You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@hadoop.apache.org by Sai Sai <sa...@yahoo.in> on 2013/09/26 12:58:46 UTC
Re: 2 Map tasks running for a small input file
Hi
Here is the input file for the wordcount job:
******************
Hi This is a simple test.
Hi Hadoop how r u.
Hello Hello.
Hi Hi.
Hadoop Hadoop Welcome.
******************
After running the wordcount successfully
here r the counters info:
***************
Job Counters SLOTS_MILLIS_MAPS 0 0 8,386
Launched reduce tasks 0 0 1
Total time spent by all reduces waiting after reserving slots (ms) 0 0 0
Total time spent by all maps waiting after reserving slots (ms) 0 0 0
Launched map tasks 0 0 2
Data-local map tasks 0 0 2
SLOTS_MILLIS_REDUCES 0 0 9,199
***************
My question why r there 2 launched map tasks when i have only a small file.
Per my understanding it is only 1 block.
and should be only 1 split.
Then for each line a map computation should occur
but it shows 2 map tasks.
Please let me know.
Thanks
Sai
Re: Input Split vs Task vs attempt vs computation
Posted by Sonal Goyal <so...@gmail.com>.
Inline
Best Regards,
Sonal
Nube Technologies <http://www.nubetech.co>
<http://in.linkedin.com/in/sonalgoyal>
On Fri, Sep 27, 2013 at 10:42 AM, Sai Sai <sa...@yahoo.in> wrote:
> Hi
> I have a few questions i am trying to understand:
>
> 1. Is each input split same as a record, (a rec can be a single line or
> multiple lines).
>
An InputSplit is a chunk of input that is handled by a map task. It will
generally contain multiple records. The RecordReader provides the key
values to the map task. Check
http://hadoop.apache.org/docs/stable/api/org/apache/hadoop/mapreduce/InputSplit.html
>
> 2. Is each Task a collection of few computations or attempts.
>
> For ex: if i have a small file with 5 lines.
> By default there will be 1 line on which each map computation is performed.
> So totally 5 computations r done on 1 node.
>
> This means JT will spawn 1 JVM for 1 Tasktracker on a node
> and another JVM for map task which will instantiate 5 map objects 1 for
> each line.
>
> i am not sure what you mean by 5 map objects. But yes, the mapper will be
invoked 5 times, once for each line.
> The MT JVM is called the task which will have 5 attempts for each line.
> This means attempt is same as computation.
>
> Please let me know if anything is incorrect.
> Thanks
> Sai
>
>
Re: Input Split vs Task vs attempt vs computation
Posted by Sonal Goyal <so...@gmail.com>.
Inline
Best Regards,
Sonal
Nube Technologies <http://www.nubetech.co>
<http://in.linkedin.com/in/sonalgoyal>
On Fri, Sep 27, 2013 at 10:42 AM, Sai Sai <sa...@yahoo.in> wrote:
> Hi
> I have a few questions i am trying to understand:
>
> 1. Is each input split same as a record, (a rec can be a single line or
> multiple lines).
>
An InputSplit is a chunk of input that is handled by a map task. It will
generally contain multiple records. The RecordReader provides the key
values to the map task. Check
http://hadoop.apache.org/docs/stable/api/org/apache/hadoop/mapreduce/InputSplit.html
>
> 2. Is each Task a collection of few computations or attempts.
>
> For ex: if i have a small file with 5 lines.
> By default there will be 1 line on which each map computation is performed.
> So totally 5 computations r done on 1 node.
>
> This means JT will spawn 1 JVM for 1 Tasktracker on a node
> and another JVM for map task which will instantiate 5 map objects 1 for
> each line.
>
> i am not sure what you mean by 5 map objects. But yes, the mapper will be
invoked 5 times, once for each line.
> The MT JVM is called the task which will have 5 attempts for each line.
> This means attempt is same as computation.
>
> Please let me know if anything is incorrect.
> Thanks
> Sai
>
>
Re: Input Split vs Task vs attempt vs computation
Posted by Sonal Goyal <so...@gmail.com>.
Inline
Best Regards,
Sonal
Nube Technologies <http://www.nubetech.co>
<http://in.linkedin.com/in/sonalgoyal>
On Fri, Sep 27, 2013 at 10:42 AM, Sai Sai <sa...@yahoo.in> wrote:
> Hi
> I have a few questions i am trying to understand:
>
> 1. Is each input split same as a record, (a rec can be a single line or
> multiple lines).
>
An InputSplit is a chunk of input that is handled by a map task. It will
generally contain multiple records. The RecordReader provides the key
values to the map task. Check
http://hadoop.apache.org/docs/stable/api/org/apache/hadoop/mapreduce/InputSplit.html
>
> 2. Is each Task a collection of few computations or attempts.
>
> For ex: if i have a small file with 5 lines.
> By default there will be 1 line on which each map computation is performed.
> So totally 5 computations r done on 1 node.
>
> This means JT will spawn 1 JVM for 1 Tasktracker on a node
> and another JVM for map task which will instantiate 5 map objects 1 for
> each line.
>
> i am not sure what you mean by 5 map objects. But yes, the mapper will be
invoked 5 times, once for each line.
> The MT JVM is called the task which will have 5 attempts for each line.
> This means attempt is same as computation.
>
> Please let me know if anything is incorrect.
> Thanks
> Sai
>
>
Re: Input Split vs Task vs attempt vs computation
Posted by Sonal Goyal <so...@gmail.com>.
Inline
Best Regards,
Sonal
Nube Technologies <http://www.nubetech.co>
<http://in.linkedin.com/in/sonalgoyal>
On Fri, Sep 27, 2013 at 10:42 AM, Sai Sai <sa...@yahoo.in> wrote:
> Hi
> I have a few questions i am trying to understand:
>
> 1. Is each input split same as a record, (a rec can be a single line or
> multiple lines).
>
An InputSplit is a chunk of input that is handled by a map task. It will
generally contain multiple records. The RecordReader provides the key
values to the map task. Check
http://hadoop.apache.org/docs/stable/api/org/apache/hadoop/mapreduce/InputSplit.html
>
> 2. Is each Task a collection of few computations or attempts.
>
> For ex: if i have a small file with 5 lines.
> By default there will be 1 line on which each map computation is performed.
> So totally 5 computations r done on 1 node.
>
> This means JT will spawn 1 JVM for 1 Tasktracker on a node
> and another JVM for map task which will instantiate 5 map objects 1 for
> each line.
>
> i am not sure what you mean by 5 map objects. But yes, the mapper will be
invoked 5 times, once for each line.
> The MT JVM is called the task which will have 5 attempts for each line.
> This means attempt is same as computation.
>
> Please let me know if anything is incorrect.
> Thanks
> Sai
>
>
Re: Input Split vs Task vs attempt vs computation
Posted by Sai Sai <sa...@yahoo.in>.
Hi
I have a few questions i am trying to understand:
1. Is each input split same as a record, (a rec can be a single line or multiple lines).
2. Is each Task a collection of few computations or attempts.
For ex: if i have a small file with 5 lines.
By default there will be 1 line on which each map computation is performed.
So totally 5 computations r done on 1 node.
This means JT will spawn 1 JVM for 1 Tasktracker on a node
and another JVM for map task which will instantiate 5 map objects 1 for each line.
The MT JVM is called the task which will have 5 attempts for each line.
This means attempt is same as computation.
Please let me know if anything is incorrect.
Thanks
Sai
Re: 2 Map tasks running for a small input file
Posted by Shekhar Sharma <sh...@gmail.com>.
Number of map tasks on a mapreduce job doesnt depend on this
property..it depends on the number of input splits...( or equal to
number blocks if input split size = block size)
1. What is the input format you are using? if yes what is the value of
N, you are using?
2. WHat is the propety mapred.min.split.size? have you changed to
something else or is it default which is 1?
Regards,
Som Shekhar Sharma
+91-8197243810
On Thu, Sep 26, 2013 at 4:39 PM, Viji R <vi...@cloudera.com> wrote:
> Hi,
>
> Default number of map tasks is 2. You can set mapred.map.tasks to 1 to
> avoid this.
>
> Regards,
> Viji
>
> On Thu, Sep 26, 2013 at 4:28 PM, Sai Sai <sa...@yahoo.in> wrote:
>> Hi
>> Here is the input file for the wordcount job:
>> ******************
>> Hi This is a simple test.
>> Hi Hadoop how r u.
>> Hello Hello.
>> Hi Hi.
>> Hadoop Hadoop Welcome.
>> ******************
>>
>> After running the wordcount successfully
>> here r the counters info:
>>
>> ***************
>> Job Counters SLOTS_MILLIS_MAPS 0 0 8,386
>> Launched reduce tasks 0 0 1
>> Total time spent by all reduces waiting after reserving slots (ms) 0 0 0
>> Total time spent by all maps waiting after reserving slots (ms) 0 0 0
>> Launched map tasks 0 0 2
>> Data-local map tasks 0 0 2
>> SLOTS_MILLIS_REDUCES 0 0 9,199
>> ***************
>> My question why r there 2 launched map tasks when i have only a small file.
>> Per my understanding it is only 1 block.
>> and should be only 1 split.
>> Then for each line a map computation should occur
>> but it shows 2 map tasks.
>> Please let me know.
>> Thanks
>> Sai
>>
Re: 2 Map tasks running for a small input file
Posted by Shekhar Sharma <sh...@gmail.com>.
Number of map tasks on a mapreduce job doesnt depend on this
property..it depends on the number of input splits...( or equal to
number blocks if input split size = block size)
1. What is the input format you are using? if yes what is the value of
N, you are using?
2. WHat is the propety mapred.min.split.size? have you changed to
something else or is it default which is 1?
Regards,
Som Shekhar Sharma
+91-8197243810
On Thu, Sep 26, 2013 at 4:39 PM, Viji R <vi...@cloudera.com> wrote:
> Hi,
>
> Default number of map tasks is 2. You can set mapred.map.tasks to 1 to
> avoid this.
>
> Regards,
> Viji
>
> On Thu, Sep 26, 2013 at 4:28 PM, Sai Sai <sa...@yahoo.in> wrote:
>> Hi
>> Here is the input file for the wordcount job:
>> ******************
>> Hi This is a simple test.
>> Hi Hadoop how r u.
>> Hello Hello.
>> Hi Hi.
>> Hadoop Hadoop Welcome.
>> ******************
>>
>> After running the wordcount successfully
>> here r the counters info:
>>
>> ***************
>> Job Counters SLOTS_MILLIS_MAPS 0 0 8,386
>> Launched reduce tasks 0 0 1
>> Total time spent by all reduces waiting after reserving slots (ms) 0 0 0
>> Total time spent by all maps waiting after reserving slots (ms) 0 0 0
>> Launched map tasks 0 0 2
>> Data-local map tasks 0 0 2
>> SLOTS_MILLIS_REDUCES 0 0 9,199
>> ***************
>> My question why r there 2 launched map tasks when i have only a small file.
>> Per my understanding it is only 1 block.
>> and should be only 1 split.
>> Then for each line a map computation should occur
>> but it shows 2 map tasks.
>> Please let me know.
>> Thanks
>> Sai
>>
Re: 2 Map tasks running for a small input file
Posted by Shekhar Sharma <sh...@gmail.com>.
Dmapred.tasktracker.map.tasks.maximum=1 ...Guys this property is set
for task tracker...when you set this property it means, that
particular task tracker will not run more than 1 mapper task
parallely..
FOr example: if a map reduce job requires 5 mapper tasks and if you
set this property to 1, then only 1 mapper task will run and other
will wait..once the task is completed other tasks will be scheduled...
Could you please send the code, you are trying to run..the driver code
and mapred-site.xml contents..?
You can controll the numbr of map task through input split size(
mapred.min.split.size, mapred.max.split.size and dfs.block.size)
max(minSPlitSize, min(maxSPlitsize, blocksize))
Regards,
Som Shekhar Sharma
+91-8197243810
On Thu, Sep 26, 2013 at 6:07 PM, shashwat shriparv
<dw...@gmail.com> wrote:
> just try giving -Dmapred.tasktracker.map.tasks.maximum=1 on the command line
> and check how many map task its running. and also set this in
> mapred-site.xml and check.
>
> Thanks & Regards
>
> ∞
>
> Shashwat Shriparv
>
>
>
> On Thu, Sep 26, 2013 at 5:24 PM, Harsh J <ha...@cloudera.com> wrote:
>>
>> Hi Sai,
>>
>> What Viji indicated is that the default Apache Hadoop setting for any
>> input is 2 maps. If the input is larger than one block, regular
>> policies of splitting such as those stated by Shekhar would apply. But
>> for smaller inputs, just for an out-of-box "parallelism experience",
>> Hadoop ships with a 2-maps forced splitting default
>> (mapred.map.tasks=2).
>>
>> This means your 5 lines is probably divided as 2:3 or other ratios and
>> is processed by 2 different Tasks. As Viji also indicated, to turn off
>> this behavior, you can set the mapred.map.tasks to 1 in your configs
>> and then you'll see only one map task process all 5 lines.
>>
>> On Thu, Sep 26, 2013 at 4:59 PM, Sai Sai <sa...@yahoo.in> wrote:
>> > Thanks Viji.
>> > I am confused a little when the data is small y would there b 2 tasks.
>> > U will use the min as 2 if u need it but in this case it is not needed
>> > due
>> > to size of the data being small
>> > so y would 2 map tasks exec.
>> > Since it results in 1 block with 5 lines of data in it
>> > i am assuming this results in 5 map computations 1 per each line
>> > and all of em in 1 process/node since i m using a pseudo vm.
>> > Where is the second task coming from.
>> > The 5 computations of map on each line is 1 task.
>> > Is this right.
>> > Please help.
>> > Thanks
>> >
>> >
>> > ________________________________
>> > From: Viji R <vi...@cloudera.com>
>> > To: user@hadoop.apache.org; Sai Sai <sa...@yahoo.in>
>> > Sent: Thursday, 26 September 2013 5:09 PM
>> > Subject: Re: 2 Map tasks running for a small input file
>> >
>> > Hi,
>> >
>> > Default number of map tasks is 2. You can set mapred.map.tasks to 1 to
>> > avoid this.
>> >
>> > Regards,
>> > Viji
>> >
>> > On Thu, Sep 26, 2013 at 4:28 PM, Sai Sai <sa...@yahoo.in> wrote:
>> >> Hi
>> >> Here is the input file for the wordcount job:
>> >> ******************
>> >> Hi This is a simple test.
>> >> Hi Hadoop how r u.
>> >> Hello Hello.
>> >> Hi Hi.
>> >> Hadoop Hadoop Welcome.
>> >> ******************
>> >>
>> >> After running the wordcount successfully
>> >> here r the counters info:
>> >>
>> >> ***************
>> >> Job Counters SLOTS_MILLIS_MAPS 0 0 8,386
>> >> Launched reduce tasks 0 0 1
>> >> Total time spent by all reduces waiting after reserving slots (ms) 0 0
>> >> 0
>> >> Total time spent by all maps waiting after reserving slots (ms) 0 0 0
>> >> Launched map tasks 0 0 2
>> >> Data-local map tasks 0 0 2
>> >> SLOTS_MILLIS_REDUCES 0 0 9,199
>> >> ***************
>> >> My question why r there 2 launched map tasks when i have only a small
>> >> file.
>> >> Per my understanding it is only 1 block.
>> >> and should be only 1 split.
>> >> Then for each line a map computation should occur
>> >> but it shows 2 map tasks.
>> >> Please let me know.
>> >> Thanks
>> >> Sai
>> >>
>> >
>> >
>>
>>
>>
>> --
>> Harsh J
>
>
Re: 2 Map tasks running for a small input file
Posted by Shekhar Sharma <sh...@gmail.com>.
Dmapred.tasktracker.map.tasks.maximum=1 ...Guys this property is set
for task tracker...when you set this property it means, that
particular task tracker will not run more than 1 mapper task
parallely..
FOr example: if a map reduce job requires 5 mapper tasks and if you
set this property to 1, then only 1 mapper task will run and other
will wait..once the task is completed other tasks will be scheduled...
Could you please send the code, you are trying to run..the driver code
and mapred-site.xml contents..?
You can controll the numbr of map task through input split size(
mapred.min.split.size, mapred.max.split.size and dfs.block.size)
max(minSPlitSize, min(maxSPlitsize, blocksize))
Regards,
Som Shekhar Sharma
+91-8197243810
On Thu, Sep 26, 2013 at 6:07 PM, shashwat shriparv
<dw...@gmail.com> wrote:
> just try giving -Dmapred.tasktracker.map.tasks.maximum=1 on the command line
> and check how many map task its running. and also set this in
> mapred-site.xml and check.
>
> Thanks & Regards
>
> ∞
>
> Shashwat Shriparv
>
>
>
> On Thu, Sep 26, 2013 at 5:24 PM, Harsh J <ha...@cloudera.com> wrote:
>>
>> Hi Sai,
>>
>> What Viji indicated is that the default Apache Hadoop setting for any
>> input is 2 maps. If the input is larger than one block, regular
>> policies of splitting such as those stated by Shekhar would apply. But
>> for smaller inputs, just for an out-of-box "parallelism experience",
>> Hadoop ships with a 2-maps forced splitting default
>> (mapred.map.tasks=2).
>>
>> This means your 5 lines is probably divided as 2:3 or other ratios and
>> is processed by 2 different Tasks. As Viji also indicated, to turn off
>> this behavior, you can set the mapred.map.tasks to 1 in your configs
>> and then you'll see only one map task process all 5 lines.
>>
>> On Thu, Sep 26, 2013 at 4:59 PM, Sai Sai <sa...@yahoo.in> wrote:
>> > Thanks Viji.
>> > I am confused a little when the data is small y would there b 2 tasks.
>> > U will use the min as 2 if u need it but in this case it is not needed
>> > due
>> > to size of the data being small
>> > so y would 2 map tasks exec.
>> > Since it results in 1 block with 5 lines of data in it
>> > i am assuming this results in 5 map computations 1 per each line
>> > and all of em in 1 process/node since i m using a pseudo vm.
>> > Where is the second task coming from.
>> > The 5 computations of map on each line is 1 task.
>> > Is this right.
>> > Please help.
>> > Thanks
>> >
>> >
>> > ________________________________
>> > From: Viji R <vi...@cloudera.com>
>> > To: user@hadoop.apache.org; Sai Sai <sa...@yahoo.in>
>> > Sent: Thursday, 26 September 2013 5:09 PM
>> > Subject: Re: 2 Map tasks running for a small input file
>> >
>> > Hi,
>> >
>> > Default number of map tasks is 2. You can set mapred.map.tasks to 1 to
>> > avoid this.
>> >
>> > Regards,
>> > Viji
>> >
>> > On Thu, Sep 26, 2013 at 4:28 PM, Sai Sai <sa...@yahoo.in> wrote:
>> >> Hi
>> >> Here is the input file for the wordcount job:
>> >> ******************
>> >> Hi This is a simple test.
>> >> Hi Hadoop how r u.
>> >> Hello Hello.
>> >> Hi Hi.
>> >> Hadoop Hadoop Welcome.
>> >> ******************
>> >>
>> >> After running the wordcount successfully
>> >> here r the counters info:
>> >>
>> >> ***************
>> >> Job Counters SLOTS_MILLIS_MAPS 0 0 8,386
>> >> Launched reduce tasks 0 0 1
>> >> Total time spent by all reduces waiting after reserving slots (ms) 0 0
>> >> 0
>> >> Total time spent by all maps waiting after reserving slots (ms) 0 0 0
>> >> Launched map tasks 0 0 2
>> >> Data-local map tasks 0 0 2
>> >> SLOTS_MILLIS_REDUCES 0 0 9,199
>> >> ***************
>> >> My question why r there 2 launched map tasks when i have only a small
>> >> file.
>> >> Per my understanding it is only 1 block.
>> >> and should be only 1 split.
>> >> Then for each line a map computation should occur
>> >> but it shows 2 map tasks.
>> >> Please let me know.
>> >> Thanks
>> >> Sai
>> >>
>> >
>> >
>>
>>
>>
>> --
>> Harsh J
>
>
Re: 2 Map tasks running for a small input file
Posted by Shekhar Sharma <sh...@gmail.com>.
Dmapred.tasktracker.map.tasks.maximum=1 ...Guys this property is set
for task tracker...when you set this property it means, that
particular task tracker will not run more than 1 mapper task
parallely..
FOr example: if a map reduce job requires 5 mapper tasks and if you
set this property to 1, then only 1 mapper task will run and other
will wait..once the task is completed other tasks will be scheduled...
Could you please send the code, you are trying to run..the driver code
and mapred-site.xml contents..?
You can controll the numbr of map task through input split size(
mapred.min.split.size, mapred.max.split.size and dfs.block.size)
max(minSPlitSize, min(maxSPlitsize, blocksize))
Regards,
Som Shekhar Sharma
+91-8197243810
On Thu, Sep 26, 2013 at 6:07 PM, shashwat shriparv
<dw...@gmail.com> wrote:
> just try giving -Dmapred.tasktracker.map.tasks.maximum=1 on the command line
> and check how many map task its running. and also set this in
> mapred-site.xml and check.
>
> Thanks & Regards
>
> ∞
>
> Shashwat Shriparv
>
>
>
> On Thu, Sep 26, 2013 at 5:24 PM, Harsh J <ha...@cloudera.com> wrote:
>>
>> Hi Sai,
>>
>> What Viji indicated is that the default Apache Hadoop setting for any
>> input is 2 maps. If the input is larger than one block, regular
>> policies of splitting such as those stated by Shekhar would apply. But
>> for smaller inputs, just for an out-of-box "parallelism experience",
>> Hadoop ships with a 2-maps forced splitting default
>> (mapred.map.tasks=2).
>>
>> This means your 5 lines is probably divided as 2:3 or other ratios and
>> is processed by 2 different Tasks. As Viji also indicated, to turn off
>> this behavior, you can set the mapred.map.tasks to 1 in your configs
>> and then you'll see only one map task process all 5 lines.
>>
>> On Thu, Sep 26, 2013 at 4:59 PM, Sai Sai <sa...@yahoo.in> wrote:
>> > Thanks Viji.
>> > I am confused a little when the data is small y would there b 2 tasks.
>> > U will use the min as 2 if u need it but in this case it is not needed
>> > due
>> > to size of the data being small
>> > so y would 2 map tasks exec.
>> > Since it results in 1 block with 5 lines of data in it
>> > i am assuming this results in 5 map computations 1 per each line
>> > and all of em in 1 process/node since i m using a pseudo vm.
>> > Where is the second task coming from.
>> > The 5 computations of map on each line is 1 task.
>> > Is this right.
>> > Please help.
>> > Thanks
>> >
>> >
>> > ________________________________
>> > From: Viji R <vi...@cloudera.com>
>> > To: user@hadoop.apache.org; Sai Sai <sa...@yahoo.in>
>> > Sent: Thursday, 26 September 2013 5:09 PM
>> > Subject: Re: 2 Map tasks running for a small input file
>> >
>> > Hi,
>> >
>> > Default number of map tasks is 2. You can set mapred.map.tasks to 1 to
>> > avoid this.
>> >
>> > Regards,
>> > Viji
>> >
>> > On Thu, Sep 26, 2013 at 4:28 PM, Sai Sai <sa...@yahoo.in> wrote:
>> >> Hi
>> >> Here is the input file for the wordcount job:
>> >> ******************
>> >> Hi This is a simple test.
>> >> Hi Hadoop how r u.
>> >> Hello Hello.
>> >> Hi Hi.
>> >> Hadoop Hadoop Welcome.
>> >> ******************
>> >>
>> >> After running the wordcount successfully
>> >> here r the counters info:
>> >>
>> >> ***************
>> >> Job Counters SLOTS_MILLIS_MAPS 0 0 8,386
>> >> Launched reduce tasks 0 0 1
>> >> Total time spent by all reduces waiting after reserving slots (ms) 0 0
>> >> 0
>> >> Total time spent by all maps waiting after reserving slots (ms) 0 0 0
>> >> Launched map tasks 0 0 2
>> >> Data-local map tasks 0 0 2
>> >> SLOTS_MILLIS_REDUCES 0 0 9,199
>> >> ***************
>> >> My question why r there 2 launched map tasks when i have only a small
>> >> file.
>> >> Per my understanding it is only 1 block.
>> >> and should be only 1 split.
>> >> Then for each line a map computation should occur
>> >> but it shows 2 map tasks.
>> >> Please let me know.
>> >> Thanks
>> >> Sai
>> >>
>> >
>> >
>>
>>
>>
>> --
>> Harsh J
>
>
Re: 2 Map tasks running for a small input file
Posted by Shekhar Sharma <sh...@gmail.com>.
Dmapred.tasktracker.map.tasks.maximum=1 ...Guys this property is set
for task tracker...when you set this property it means, that
particular task tracker will not run more than 1 mapper task
parallely..
FOr example: if a map reduce job requires 5 mapper tasks and if you
set this property to 1, then only 1 mapper task will run and other
will wait..once the task is completed other tasks will be scheduled...
Could you please send the code, you are trying to run..the driver code
and mapred-site.xml contents..?
You can controll the numbr of map task through input split size(
mapred.min.split.size, mapred.max.split.size and dfs.block.size)
max(minSPlitSize, min(maxSPlitsize, blocksize))
Regards,
Som Shekhar Sharma
+91-8197243810
On Thu, Sep 26, 2013 at 6:07 PM, shashwat shriparv
<dw...@gmail.com> wrote:
> just try giving -Dmapred.tasktracker.map.tasks.maximum=1 on the command line
> and check how many map task its running. and also set this in
> mapred-site.xml and check.
>
> Thanks & Regards
>
> ∞
>
> Shashwat Shriparv
>
>
>
> On Thu, Sep 26, 2013 at 5:24 PM, Harsh J <ha...@cloudera.com> wrote:
>>
>> Hi Sai,
>>
>> What Viji indicated is that the default Apache Hadoop setting for any
>> input is 2 maps. If the input is larger than one block, regular
>> policies of splitting such as those stated by Shekhar would apply. But
>> for smaller inputs, just for an out-of-box "parallelism experience",
>> Hadoop ships with a 2-maps forced splitting default
>> (mapred.map.tasks=2).
>>
>> This means your 5 lines is probably divided as 2:3 or other ratios and
>> is processed by 2 different Tasks. As Viji also indicated, to turn off
>> this behavior, you can set the mapred.map.tasks to 1 in your configs
>> and then you'll see only one map task process all 5 lines.
>>
>> On Thu, Sep 26, 2013 at 4:59 PM, Sai Sai <sa...@yahoo.in> wrote:
>> > Thanks Viji.
>> > I am confused a little when the data is small y would there b 2 tasks.
>> > U will use the min as 2 if u need it but in this case it is not needed
>> > due
>> > to size of the data being small
>> > so y would 2 map tasks exec.
>> > Since it results in 1 block with 5 lines of data in it
>> > i am assuming this results in 5 map computations 1 per each line
>> > and all of em in 1 process/node since i m using a pseudo vm.
>> > Where is the second task coming from.
>> > The 5 computations of map on each line is 1 task.
>> > Is this right.
>> > Please help.
>> > Thanks
>> >
>> >
>> > ________________________________
>> > From: Viji R <vi...@cloudera.com>
>> > To: user@hadoop.apache.org; Sai Sai <sa...@yahoo.in>
>> > Sent: Thursday, 26 September 2013 5:09 PM
>> > Subject: Re: 2 Map tasks running for a small input file
>> >
>> > Hi,
>> >
>> > Default number of map tasks is 2. You can set mapred.map.tasks to 1 to
>> > avoid this.
>> >
>> > Regards,
>> > Viji
>> >
>> > On Thu, Sep 26, 2013 at 4:28 PM, Sai Sai <sa...@yahoo.in> wrote:
>> >> Hi
>> >> Here is the input file for the wordcount job:
>> >> ******************
>> >> Hi This is a simple test.
>> >> Hi Hadoop how r u.
>> >> Hello Hello.
>> >> Hi Hi.
>> >> Hadoop Hadoop Welcome.
>> >> ******************
>> >>
>> >> After running the wordcount successfully
>> >> here r the counters info:
>> >>
>> >> ***************
>> >> Job Counters SLOTS_MILLIS_MAPS 0 0 8,386
>> >> Launched reduce tasks 0 0 1
>> >> Total time spent by all reduces waiting after reserving slots (ms) 0 0
>> >> 0
>> >> Total time spent by all maps waiting after reserving slots (ms) 0 0 0
>> >> Launched map tasks 0 0 2
>> >> Data-local map tasks 0 0 2
>> >> SLOTS_MILLIS_REDUCES 0 0 9,199
>> >> ***************
>> >> My question why r there 2 launched map tasks when i have only a small
>> >> file.
>> >> Per my understanding it is only 1 block.
>> >> and should be only 1 split.
>> >> Then for each line a map computation should occur
>> >> but it shows 2 map tasks.
>> >> Please let me know.
>> >> Thanks
>> >> Sai
>> >>
>> >
>> >
>>
>>
>>
>> --
>> Harsh J
>
>
Re: 2 Map tasks running for a small input file
Posted by shashwat shriparv <dw...@gmail.com>.
just try giving -Dmapred.tasktracker.map.tasks.maximum=1 on the command
line and check how many map task its running. and also set this in
mapred-site.xml and check.
*Thanks & Regards *
∞
Shashwat Shriparv
On Thu, Sep 26, 2013 at 5:24 PM, Harsh J <ha...@cloudera.com> wrote:
> Hi Sai,
>
> What Viji indicated is that the default Apache Hadoop setting for any
> input is 2 maps. If the input is larger than one block, regular
> policies of splitting such as those stated by Shekhar would apply. But
> for smaller inputs, just for an out-of-box "parallelism experience",
> Hadoop ships with a 2-maps forced splitting default
> (mapred.map.tasks=2).
>
> This means your 5 lines is probably divided as 2:3 or other ratios and
> is processed by 2 different Tasks. As Viji also indicated, to turn off
> this behavior, you can set the mapred.map.tasks to 1 in your configs
> and then you'll see only one map task process all 5 lines.
>
> On Thu, Sep 26, 2013 at 4:59 PM, Sai Sai <sa...@yahoo.in> wrote:
> > Thanks Viji.
> > I am confused a little when the data is small y would there b 2 tasks.
> > U will use the min as 2 if u need it but in this case it is not needed
> due
> > to size of the data being small
> > so y would 2 map tasks exec.
> > Since it results in 1 block with 5 lines of data in it
> > i am assuming this results in 5 map computations 1 per each line
> > and all of em in 1 process/node since i m using a pseudo vm.
> > Where is the second task coming from.
> > The 5 computations of map on each line is 1 task.
> > Is this right.
> > Please help.
> > Thanks
> >
> >
> > ________________________________
> > From: Viji R <vi...@cloudera.com>
> > To: user@hadoop.apache.org; Sai Sai <sa...@yahoo.in>
> > Sent: Thursday, 26 September 2013 5:09 PM
> > Subject: Re: 2 Map tasks running for a small input file
> >
> > Hi,
> >
> > Default number of map tasks is 2. You can set mapred.map.tasks to 1 to
> > avoid this.
> >
> > Regards,
> > Viji
> >
> > On Thu, Sep 26, 2013 at 4:28 PM, Sai Sai <sa...@yahoo.in> wrote:
> >> Hi
> >> Here is the input file for the wordcount job:
> >> ******************
> >> Hi This is a simple test.
> >> Hi Hadoop how r u.
> >> Hello Hello.
> >> Hi Hi.
> >> Hadoop Hadoop Welcome.
> >> ******************
> >>
> >> After running the wordcount successfully
> >> here r the counters info:
> >>
> >> ***************
> >> Job Counters SLOTS_MILLIS_MAPS 0 0 8,386
> >> Launched reduce tasks 0 0 1
> >> Total time spent by all reduces waiting after reserving slots (ms) 0 0 0
> >> Total time spent by all maps waiting after reserving slots (ms) 0 0 0
> >> Launched map tasks 0 0 2
> >> Data-local map tasks 0 0 2
> >> SLOTS_MILLIS_REDUCES 0 0 9,199
> >> ***************
> >> My question why r there 2 launched map tasks when i have only a small
> >> file.
> >> Per my understanding it is only 1 block.
> >> and should be only 1 split.
> >> Then for each line a map computation should occur
> >> but it shows 2 map tasks.
> >> Please let me know.
> >> Thanks
> >> Sai
> >>
> >
> >
>
>
>
> --
> Harsh J
>
Re: 2 Map tasks running for a small input file
Posted by shashwat shriparv <dw...@gmail.com>.
just try giving -Dmapred.tasktracker.map.tasks.maximum=1 on the command
line and check how many map task its running. and also set this in
mapred-site.xml and check.
*Thanks & Regards *
∞
Shashwat Shriparv
On Thu, Sep 26, 2013 at 5:24 PM, Harsh J <ha...@cloudera.com> wrote:
> Hi Sai,
>
> What Viji indicated is that the default Apache Hadoop setting for any
> input is 2 maps. If the input is larger than one block, regular
> policies of splitting such as those stated by Shekhar would apply. But
> for smaller inputs, just for an out-of-box "parallelism experience",
> Hadoop ships with a 2-maps forced splitting default
> (mapred.map.tasks=2).
>
> This means your 5 lines is probably divided as 2:3 or other ratios and
> is processed by 2 different Tasks. As Viji also indicated, to turn off
> this behavior, you can set the mapred.map.tasks to 1 in your configs
> and then you'll see only one map task process all 5 lines.
>
> On Thu, Sep 26, 2013 at 4:59 PM, Sai Sai <sa...@yahoo.in> wrote:
> > Thanks Viji.
> > I am confused a little when the data is small y would there b 2 tasks.
> > U will use the min as 2 if u need it but in this case it is not needed
> due
> > to size of the data being small
> > so y would 2 map tasks exec.
> > Since it results in 1 block with 5 lines of data in it
> > i am assuming this results in 5 map computations 1 per each line
> > and all of em in 1 process/node since i m using a pseudo vm.
> > Where is the second task coming from.
> > The 5 computations of map on each line is 1 task.
> > Is this right.
> > Please help.
> > Thanks
> >
> >
> > ________________________________
> > From: Viji R <vi...@cloudera.com>
> > To: user@hadoop.apache.org; Sai Sai <sa...@yahoo.in>
> > Sent: Thursday, 26 September 2013 5:09 PM
> > Subject: Re: 2 Map tasks running for a small input file
> >
> > Hi,
> >
> > Default number of map tasks is 2. You can set mapred.map.tasks to 1 to
> > avoid this.
> >
> > Regards,
> > Viji
> >
> > On Thu, Sep 26, 2013 at 4:28 PM, Sai Sai <sa...@yahoo.in> wrote:
> >> Hi
> >> Here is the input file for the wordcount job:
> >> ******************
> >> Hi This is a simple test.
> >> Hi Hadoop how r u.
> >> Hello Hello.
> >> Hi Hi.
> >> Hadoop Hadoop Welcome.
> >> ******************
> >>
> >> After running the wordcount successfully
> >> here r the counters info:
> >>
> >> ***************
> >> Job Counters SLOTS_MILLIS_MAPS 0 0 8,386
> >> Launched reduce tasks 0 0 1
> >> Total time spent by all reduces waiting after reserving slots (ms) 0 0 0
> >> Total time spent by all maps waiting after reserving slots (ms) 0 0 0
> >> Launched map tasks 0 0 2
> >> Data-local map tasks 0 0 2
> >> SLOTS_MILLIS_REDUCES 0 0 9,199
> >> ***************
> >> My question why r there 2 launched map tasks when i have only a small
> >> file.
> >> Per my understanding it is only 1 block.
> >> and should be only 1 split.
> >> Then for each line a map computation should occur
> >> but it shows 2 map tasks.
> >> Please let me know.
> >> Thanks
> >> Sai
> >>
> >
> >
>
>
>
> --
> Harsh J
>
Re: 2 Map tasks running for a small input file
Posted by shashwat shriparv <dw...@gmail.com>.
just try giving -Dmapred.tasktracker.map.tasks.maximum=1 on the command
line and check how many map task its running. and also set this in
mapred-site.xml and check.
*Thanks & Regards *
∞
Shashwat Shriparv
On Thu, Sep 26, 2013 at 5:24 PM, Harsh J <ha...@cloudera.com> wrote:
> Hi Sai,
>
> What Viji indicated is that the default Apache Hadoop setting for any
> input is 2 maps. If the input is larger than one block, regular
> policies of splitting such as those stated by Shekhar would apply. But
> for smaller inputs, just for an out-of-box "parallelism experience",
> Hadoop ships with a 2-maps forced splitting default
> (mapred.map.tasks=2).
>
> This means your 5 lines is probably divided as 2:3 or other ratios and
> is processed by 2 different Tasks. As Viji also indicated, to turn off
> this behavior, you can set the mapred.map.tasks to 1 in your configs
> and then you'll see only one map task process all 5 lines.
>
> On Thu, Sep 26, 2013 at 4:59 PM, Sai Sai <sa...@yahoo.in> wrote:
> > Thanks Viji.
> > I am confused a little when the data is small y would there b 2 tasks.
> > U will use the min as 2 if u need it but in this case it is not needed
> due
> > to size of the data being small
> > so y would 2 map tasks exec.
> > Since it results in 1 block with 5 lines of data in it
> > i am assuming this results in 5 map computations 1 per each line
> > and all of em in 1 process/node since i m using a pseudo vm.
> > Where is the second task coming from.
> > The 5 computations of map on each line is 1 task.
> > Is this right.
> > Please help.
> > Thanks
> >
> >
> > ________________________________
> > From: Viji R <vi...@cloudera.com>
> > To: user@hadoop.apache.org; Sai Sai <sa...@yahoo.in>
> > Sent: Thursday, 26 September 2013 5:09 PM
> > Subject: Re: 2 Map tasks running for a small input file
> >
> > Hi,
> >
> > Default number of map tasks is 2. You can set mapred.map.tasks to 1 to
> > avoid this.
> >
> > Regards,
> > Viji
> >
> > On Thu, Sep 26, 2013 at 4:28 PM, Sai Sai <sa...@yahoo.in> wrote:
> >> Hi
> >> Here is the input file for the wordcount job:
> >> ******************
> >> Hi This is a simple test.
> >> Hi Hadoop how r u.
> >> Hello Hello.
> >> Hi Hi.
> >> Hadoop Hadoop Welcome.
> >> ******************
> >>
> >> After running the wordcount successfully
> >> here r the counters info:
> >>
> >> ***************
> >> Job Counters SLOTS_MILLIS_MAPS 0 0 8,386
> >> Launched reduce tasks 0 0 1
> >> Total time spent by all reduces waiting after reserving slots (ms) 0 0 0
> >> Total time spent by all maps waiting after reserving slots (ms) 0 0 0
> >> Launched map tasks 0 0 2
> >> Data-local map tasks 0 0 2
> >> SLOTS_MILLIS_REDUCES 0 0 9,199
> >> ***************
> >> My question why r there 2 launched map tasks when i have only a small
> >> file.
> >> Per my understanding it is only 1 block.
> >> and should be only 1 split.
> >> Then for each line a map computation should occur
> >> but it shows 2 map tasks.
> >> Please let me know.
> >> Thanks
> >> Sai
> >>
> >
> >
>
>
>
> --
> Harsh J
>
Re: 2 Map tasks running for a small input file
Posted by shashwat shriparv <dw...@gmail.com>.
just try giving -Dmapred.tasktracker.map.tasks.maximum=1 on the command
line and check how many map task its running. and also set this in
mapred-site.xml and check.
*Thanks & Regards *
∞
Shashwat Shriparv
On Thu, Sep 26, 2013 at 5:24 PM, Harsh J <ha...@cloudera.com> wrote:
> Hi Sai,
>
> What Viji indicated is that the default Apache Hadoop setting for any
> input is 2 maps. If the input is larger than one block, regular
> policies of splitting such as those stated by Shekhar would apply. But
> for smaller inputs, just for an out-of-box "parallelism experience",
> Hadoop ships with a 2-maps forced splitting default
> (mapred.map.tasks=2).
>
> This means your 5 lines is probably divided as 2:3 or other ratios and
> is processed by 2 different Tasks. As Viji also indicated, to turn off
> this behavior, you can set the mapred.map.tasks to 1 in your configs
> and then you'll see only one map task process all 5 lines.
>
> On Thu, Sep 26, 2013 at 4:59 PM, Sai Sai <sa...@yahoo.in> wrote:
> > Thanks Viji.
> > I am confused a little when the data is small y would there b 2 tasks.
> > U will use the min as 2 if u need it but in this case it is not needed
> due
> > to size of the data being small
> > so y would 2 map tasks exec.
> > Since it results in 1 block with 5 lines of data in it
> > i am assuming this results in 5 map computations 1 per each line
> > and all of em in 1 process/node since i m using a pseudo vm.
> > Where is the second task coming from.
> > The 5 computations of map on each line is 1 task.
> > Is this right.
> > Please help.
> > Thanks
> >
> >
> > ________________________________
> > From: Viji R <vi...@cloudera.com>
> > To: user@hadoop.apache.org; Sai Sai <sa...@yahoo.in>
> > Sent: Thursday, 26 September 2013 5:09 PM
> > Subject: Re: 2 Map tasks running for a small input file
> >
> > Hi,
> >
> > Default number of map tasks is 2. You can set mapred.map.tasks to 1 to
> > avoid this.
> >
> > Regards,
> > Viji
> >
> > On Thu, Sep 26, 2013 at 4:28 PM, Sai Sai <sa...@yahoo.in> wrote:
> >> Hi
> >> Here is the input file for the wordcount job:
> >> ******************
> >> Hi This is a simple test.
> >> Hi Hadoop how r u.
> >> Hello Hello.
> >> Hi Hi.
> >> Hadoop Hadoop Welcome.
> >> ******************
> >>
> >> After running the wordcount successfully
> >> here r the counters info:
> >>
> >> ***************
> >> Job Counters SLOTS_MILLIS_MAPS 0 0 8,386
> >> Launched reduce tasks 0 0 1
> >> Total time spent by all reduces waiting after reserving slots (ms) 0 0 0
> >> Total time spent by all maps waiting after reserving slots (ms) 0 0 0
> >> Launched map tasks 0 0 2
> >> Data-local map tasks 0 0 2
> >> SLOTS_MILLIS_REDUCES 0 0 9,199
> >> ***************
> >> My question why r there 2 launched map tasks when i have only a small
> >> file.
> >> Per my understanding it is only 1 block.
> >> and should be only 1 split.
> >> Then for each line a map computation should occur
> >> but it shows 2 map tasks.
> >> Please let me know.
> >> Thanks
> >> Sai
> >>
> >
> >
>
>
>
> --
> Harsh J
>
Re: 2 Map tasks running for a small input file
Posted by Harsh J <ha...@cloudera.com>.
Hi Sai,
What Viji indicated is that the default Apache Hadoop setting for any
input is 2 maps. If the input is larger than one block, regular
policies of splitting such as those stated by Shekhar would apply. But
for smaller inputs, just for an out-of-box "parallelism experience",
Hadoop ships with a 2-maps forced splitting default
(mapred.map.tasks=2).
This means your 5 lines is probably divided as 2:3 or other ratios and
is processed by 2 different Tasks. As Viji also indicated, to turn off
this behavior, you can set the mapred.map.tasks to 1 in your configs
and then you'll see only one map task process all 5 lines.
On Thu, Sep 26, 2013 at 4:59 PM, Sai Sai <sa...@yahoo.in> wrote:
> Thanks Viji.
> I am confused a little when the data is small y would there b 2 tasks.
> U will use the min as 2 if u need it but in this case it is not needed due
> to size of the data being small
> so y would 2 map tasks exec.
> Since it results in 1 block with 5 lines of data in it
> i am assuming this results in 5 map computations 1 per each line
> and all of em in 1 process/node since i m using a pseudo vm.
> Where is the second task coming from.
> The 5 computations of map on each line is 1 task.
> Is this right.
> Please help.
> Thanks
>
>
> ________________________________
> From: Viji R <vi...@cloudera.com>
> To: user@hadoop.apache.org; Sai Sai <sa...@yahoo.in>
> Sent: Thursday, 26 September 2013 5:09 PM
> Subject: Re: 2 Map tasks running for a small input file
>
> Hi,
>
> Default number of map tasks is 2. You can set mapred.map.tasks to 1 to
> avoid this.
>
> Regards,
> Viji
>
> On Thu, Sep 26, 2013 at 4:28 PM, Sai Sai <sa...@yahoo.in> wrote:
>> Hi
>> Here is the input file for the wordcount job:
>> ******************
>> Hi This is a simple test.
>> Hi Hadoop how r u.
>> Hello Hello.
>> Hi Hi.
>> Hadoop Hadoop Welcome.
>> ******************
>>
>> After running the wordcount successfully
>> here r the counters info:
>>
>> ***************
>> Job Counters SLOTS_MILLIS_MAPS 0 0 8,386
>> Launched reduce tasks 0 0 1
>> Total time spent by all reduces waiting after reserving slots (ms) 0 0 0
>> Total time spent by all maps waiting after reserving slots (ms) 0 0 0
>> Launched map tasks 0 0 2
>> Data-local map tasks 0 0 2
>> SLOTS_MILLIS_REDUCES 0 0 9,199
>> ***************
>> My question why r there 2 launched map tasks when i have only a small
>> file.
>> Per my understanding it is only 1 block.
>> and should be only 1 split.
>> Then for each line a map computation should occur
>> but it shows 2 map tasks.
>> Please let me know.
>> Thanks
>> Sai
>>
>
>
--
Harsh J
Re: 2 Map tasks running for a small input file
Posted by Harsh J <ha...@cloudera.com>.
Hi Sai,
What Viji indicated is that the default Apache Hadoop setting for any
input is 2 maps. If the input is larger than one block, regular
policies of splitting such as those stated by Shekhar would apply. But
for smaller inputs, just for an out-of-box "parallelism experience",
Hadoop ships with a 2-maps forced splitting default
(mapred.map.tasks=2).
This means your 5 lines is probably divided as 2:3 or other ratios and
is processed by 2 different Tasks. As Viji also indicated, to turn off
this behavior, you can set the mapred.map.tasks to 1 in your configs
and then you'll see only one map task process all 5 lines.
On Thu, Sep 26, 2013 at 4:59 PM, Sai Sai <sa...@yahoo.in> wrote:
> Thanks Viji.
> I am confused a little when the data is small y would there b 2 tasks.
> U will use the min as 2 if u need it but in this case it is not needed due
> to size of the data being small
> so y would 2 map tasks exec.
> Since it results in 1 block with 5 lines of data in it
> i am assuming this results in 5 map computations 1 per each line
> and all of em in 1 process/node since i m using a pseudo vm.
> Where is the second task coming from.
> The 5 computations of map on each line is 1 task.
> Is this right.
> Please help.
> Thanks
>
>
> ________________________________
> From: Viji R <vi...@cloudera.com>
> To: user@hadoop.apache.org; Sai Sai <sa...@yahoo.in>
> Sent: Thursday, 26 September 2013 5:09 PM
> Subject: Re: 2 Map tasks running for a small input file
>
> Hi,
>
> Default number of map tasks is 2. You can set mapred.map.tasks to 1 to
> avoid this.
>
> Regards,
> Viji
>
> On Thu, Sep 26, 2013 at 4:28 PM, Sai Sai <sa...@yahoo.in> wrote:
>> Hi
>> Here is the input file for the wordcount job:
>> ******************
>> Hi This is a simple test.
>> Hi Hadoop how r u.
>> Hello Hello.
>> Hi Hi.
>> Hadoop Hadoop Welcome.
>> ******************
>>
>> After running the wordcount successfully
>> here r the counters info:
>>
>> ***************
>> Job Counters SLOTS_MILLIS_MAPS 0 0 8,386
>> Launched reduce tasks 0 0 1
>> Total time spent by all reduces waiting after reserving slots (ms) 0 0 0
>> Total time spent by all maps waiting after reserving slots (ms) 0 0 0
>> Launched map tasks 0 0 2
>> Data-local map tasks 0 0 2
>> SLOTS_MILLIS_REDUCES 0 0 9,199
>> ***************
>> My question why r there 2 launched map tasks when i have only a small
>> file.
>> Per my understanding it is only 1 block.
>> and should be only 1 split.
>> Then for each line a map computation should occur
>> but it shows 2 map tasks.
>> Please let me know.
>> Thanks
>> Sai
>>
>
>
--
Harsh J
Re: 2 Map tasks running for a small input file
Posted by Harsh J <ha...@cloudera.com>.
Hi Sai,
What Viji indicated is that the default Apache Hadoop setting for any
input is 2 maps. If the input is larger than one block, regular
policies of splitting such as those stated by Shekhar would apply. But
for smaller inputs, just for an out-of-box "parallelism experience",
Hadoop ships with a 2-maps forced splitting default
(mapred.map.tasks=2).
This means your 5 lines is probably divided as 2:3 or other ratios and
is processed by 2 different Tasks. As Viji also indicated, to turn off
this behavior, you can set the mapred.map.tasks to 1 in your configs
and then you'll see only one map task process all 5 lines.
On Thu, Sep 26, 2013 at 4:59 PM, Sai Sai <sa...@yahoo.in> wrote:
> Thanks Viji.
> I am confused a little when the data is small y would there b 2 tasks.
> U will use the min as 2 if u need it but in this case it is not needed due
> to size of the data being small
> so y would 2 map tasks exec.
> Since it results in 1 block with 5 lines of data in it
> i am assuming this results in 5 map computations 1 per each line
> and all of em in 1 process/node since i m using a pseudo vm.
> Where is the second task coming from.
> The 5 computations of map on each line is 1 task.
> Is this right.
> Please help.
> Thanks
>
>
> ________________________________
> From: Viji R <vi...@cloudera.com>
> To: user@hadoop.apache.org; Sai Sai <sa...@yahoo.in>
> Sent: Thursday, 26 September 2013 5:09 PM
> Subject: Re: 2 Map tasks running for a small input file
>
> Hi,
>
> Default number of map tasks is 2. You can set mapred.map.tasks to 1 to
> avoid this.
>
> Regards,
> Viji
>
> On Thu, Sep 26, 2013 at 4:28 PM, Sai Sai <sa...@yahoo.in> wrote:
>> Hi
>> Here is the input file for the wordcount job:
>> ******************
>> Hi This is a simple test.
>> Hi Hadoop how r u.
>> Hello Hello.
>> Hi Hi.
>> Hadoop Hadoop Welcome.
>> ******************
>>
>> After running the wordcount successfully
>> here r the counters info:
>>
>> ***************
>> Job Counters SLOTS_MILLIS_MAPS 0 0 8,386
>> Launched reduce tasks 0 0 1
>> Total time spent by all reduces waiting after reserving slots (ms) 0 0 0
>> Total time spent by all maps waiting after reserving slots (ms) 0 0 0
>> Launched map tasks 0 0 2
>> Data-local map tasks 0 0 2
>> SLOTS_MILLIS_REDUCES 0 0 9,199
>> ***************
>> My question why r there 2 launched map tasks when i have only a small
>> file.
>> Per my understanding it is only 1 block.
>> and should be only 1 split.
>> Then for each line a map computation should occur
>> but it shows 2 map tasks.
>> Please let me know.
>> Thanks
>> Sai
>>
>
>
--
Harsh J
Re: 2 Map tasks running for a small input file
Posted by Harsh J <ha...@cloudera.com>.
Hi Sai,
What Viji indicated is that the default Apache Hadoop setting for any
input is 2 maps. If the input is larger than one block, regular
policies of splitting such as those stated by Shekhar would apply. But
for smaller inputs, just for an out-of-box "parallelism experience",
Hadoop ships with a 2-maps forced splitting default
(mapred.map.tasks=2).
This means your 5 lines is probably divided as 2:3 or other ratios and
is processed by 2 different Tasks. As Viji also indicated, to turn off
this behavior, you can set the mapred.map.tasks to 1 in your configs
and then you'll see only one map task process all 5 lines.
On Thu, Sep 26, 2013 at 4:59 PM, Sai Sai <sa...@yahoo.in> wrote:
> Thanks Viji.
> I am confused a little when the data is small y would there b 2 tasks.
> U will use the min as 2 if u need it but in this case it is not needed due
> to size of the data being small
> so y would 2 map tasks exec.
> Since it results in 1 block with 5 lines of data in it
> i am assuming this results in 5 map computations 1 per each line
> and all of em in 1 process/node since i m using a pseudo vm.
> Where is the second task coming from.
> The 5 computations of map on each line is 1 task.
> Is this right.
> Please help.
> Thanks
>
>
> ________________________________
> From: Viji R <vi...@cloudera.com>
> To: user@hadoop.apache.org; Sai Sai <sa...@yahoo.in>
> Sent: Thursday, 26 September 2013 5:09 PM
> Subject: Re: 2 Map tasks running for a small input file
>
> Hi,
>
> Default number of map tasks is 2. You can set mapred.map.tasks to 1 to
> avoid this.
>
> Regards,
> Viji
>
> On Thu, Sep 26, 2013 at 4:28 PM, Sai Sai <sa...@yahoo.in> wrote:
>> Hi
>> Here is the input file for the wordcount job:
>> ******************
>> Hi This is a simple test.
>> Hi Hadoop how r u.
>> Hello Hello.
>> Hi Hi.
>> Hadoop Hadoop Welcome.
>> ******************
>>
>> After running the wordcount successfully
>> here r the counters info:
>>
>> ***************
>> Job Counters SLOTS_MILLIS_MAPS 0 0 8,386
>> Launched reduce tasks 0 0 1
>> Total time spent by all reduces waiting after reserving slots (ms) 0 0 0
>> Total time spent by all maps waiting after reserving slots (ms) 0 0 0
>> Launched map tasks 0 0 2
>> Data-local map tasks 0 0 2
>> SLOTS_MILLIS_REDUCES 0 0 9,199
>> ***************
>> My question why r there 2 launched map tasks when i have only a small
>> file.
>> Per my understanding it is only 1 block.
>> and should be only 1 split.
>> Then for each line a map computation should occur
>> but it shows 2 map tasks.
>> Please let me know.
>> Thanks
>> Sai
>>
>
>
--
Harsh J
Re: 2 Map tasks running for a small input file
Posted by Sai Sai <sa...@yahoo.in>.
Thanks Viji.
I am confused a little when the data is small y would there b 2 tasks.
U will use the min as 2 if u need it but in this case it is not needed due to size of the data being small
so y would 2 map tasks exec.
Since it results in 1 block with 5 lines of data in it
i am assuming this results in 5 map computations 1 per each line
and all of em in 1 process/node since i m using a pseudo vm.
Where is the second task coming from.
The 5 computations of map on each line is 1 task.
Is this right.
Please help.
Thanks
________________________________
From: Viji R <vi...@cloudera.com>
To: user@hadoop.apache.org; Sai Sai <sa...@yahoo.in>
Sent: Thursday, 26 September 2013 5:09 PM
Subject: Re: 2 Map tasks running for a small input file
Hi,
Default number of map tasks is 2. You can set mapred.map.tasks to 1 to
avoid this.
Regards,
Viji
On Thu, Sep 26, 2013 at 4:28 PM, Sai Sai <sa...@yahoo.in> wrote:
> Hi
> Here is the input file for the wordcount job:
> ******************
> Hi This is a simple test.
> Hi Hadoop how r u.
> Hello Hello.
> Hi Hi.
> Hadoop Hadoop Welcome.
> ******************
>
> After running the wordcount successfully
> here r the counters info:
>
> ***************
> Job Counters SLOTS_MILLIS_MAPS 0 0 8,386
> Launched reduce tasks 0 0 1
> Total time spent by all reduces waiting after reserving slots (ms) 0 0 0
> Total time spent by all maps waiting after reserving slots (ms) 0 0 0
> Launched map tasks 0 0 2
> Data-local map tasks 0 0 2
> SLOTS_MILLIS_REDUCES 0 0 9,199
> ***************
> My question why r there 2 launched map tasks when i have only a small file.
> Per my understanding it is only 1 block.
> and should be only 1 split.
> Then for each line a map computation should occur
> but it shows 2 map tasks.
> Please let me know.
> Thanks
> Sai
>
Re: 2 Map tasks running for a small input file
Posted by Sai Sai <sa...@yahoo.in>.
Thanks Viji.
I am confused a little when the data is small y would there b 2 tasks.
U will use the min as 2 if u need it but in this case it is not needed due to size of the data being small
so y would 2 map tasks exec.
Since it results in 1 block with 5 lines of data in it
i am assuming this results in 5 map computations 1 per each line
and all of em in 1 process/node since i m using a pseudo vm.
Where is the second task coming from.
The 5 computations of map on each line is 1 task.
Is this right.
Please help.
Thanks
________________________________
From: Viji R <vi...@cloudera.com>
To: user@hadoop.apache.org; Sai Sai <sa...@yahoo.in>
Sent: Thursday, 26 September 2013 5:09 PM
Subject: Re: 2 Map tasks running for a small input file
Hi,
Default number of map tasks is 2. You can set mapred.map.tasks to 1 to
avoid this.
Regards,
Viji
On Thu, Sep 26, 2013 at 4:28 PM, Sai Sai <sa...@yahoo.in> wrote:
> Hi
> Here is the input file for the wordcount job:
> ******************
> Hi This is a simple test.
> Hi Hadoop how r u.
> Hello Hello.
> Hi Hi.
> Hadoop Hadoop Welcome.
> ******************
>
> After running the wordcount successfully
> here r the counters info:
>
> ***************
> Job Counters SLOTS_MILLIS_MAPS 0 0 8,386
> Launched reduce tasks 0 0 1
> Total time spent by all reduces waiting after reserving slots (ms) 0 0 0
> Total time spent by all maps waiting after reserving slots (ms) 0 0 0
> Launched map tasks 0 0 2
> Data-local map tasks 0 0 2
> SLOTS_MILLIS_REDUCES 0 0 9,199
> ***************
> My question why r there 2 launched map tasks when i have only a small file.
> Per my understanding it is only 1 block.
> and should be only 1 split.
> Then for each line a map computation should occur
> but it shows 2 map tasks.
> Please let me know.
> Thanks
> Sai
>
Re: 2 Map tasks running for a small input file
Posted by Sai Sai <sa...@yahoo.in>.
Thanks Viji.
I am confused a little when the data is small y would there b 2 tasks.
U will use the min as 2 if u need it but in this case it is not needed due to size of the data being small
so y would 2 map tasks exec.
Since it results in 1 block with 5 lines of data in it
i am assuming this results in 5 map computations 1 per each line
and all of em in 1 process/node since i m using a pseudo vm.
Where is the second task coming from.
The 5 computations of map on each line is 1 task.
Is this right.
Please help.
Thanks
________________________________
From: Viji R <vi...@cloudera.com>
To: user@hadoop.apache.org; Sai Sai <sa...@yahoo.in>
Sent: Thursday, 26 September 2013 5:09 PM
Subject: Re: 2 Map tasks running for a small input file
Hi,
Default number of map tasks is 2. You can set mapred.map.tasks to 1 to
avoid this.
Regards,
Viji
On Thu, Sep 26, 2013 at 4:28 PM, Sai Sai <sa...@yahoo.in> wrote:
> Hi
> Here is the input file for the wordcount job:
> ******************
> Hi This is a simple test.
> Hi Hadoop how r u.
> Hello Hello.
> Hi Hi.
> Hadoop Hadoop Welcome.
> ******************
>
> After running the wordcount successfully
> here r the counters info:
>
> ***************
> Job Counters SLOTS_MILLIS_MAPS 0 0 8,386
> Launched reduce tasks 0 0 1
> Total time spent by all reduces waiting after reserving slots (ms) 0 0 0
> Total time spent by all maps waiting after reserving slots (ms) 0 0 0
> Launched map tasks 0 0 2
> Data-local map tasks 0 0 2
> SLOTS_MILLIS_REDUCES 0 0 9,199
> ***************
> My question why r there 2 launched map tasks when i have only a small file.
> Per my understanding it is only 1 block.
> and should be only 1 split.
> Then for each line a map computation should occur
> but it shows 2 map tasks.
> Please let me know.
> Thanks
> Sai
>
Re: 2 Map tasks running for a small input file
Posted by Shekhar Sharma <sh...@gmail.com>.
Number of map tasks on a mapreduce job doesnt depend on this
property..it depends on the number of input splits...( or equal to
number blocks if input split size = block size)
1. What is the input format you are using? if yes what is the value of
N, you are using?
2. WHat is the propety mapred.min.split.size? have you changed to
something else or is it default which is 1?
Regards,
Som Shekhar Sharma
+91-8197243810
On Thu, Sep 26, 2013 at 4:39 PM, Viji R <vi...@cloudera.com> wrote:
> Hi,
>
> Default number of map tasks is 2. You can set mapred.map.tasks to 1 to
> avoid this.
>
> Regards,
> Viji
>
> On Thu, Sep 26, 2013 at 4:28 PM, Sai Sai <sa...@yahoo.in> wrote:
>> Hi
>> Here is the input file for the wordcount job:
>> ******************
>> Hi This is a simple test.
>> Hi Hadoop how r u.
>> Hello Hello.
>> Hi Hi.
>> Hadoop Hadoop Welcome.
>> ******************
>>
>> After running the wordcount successfully
>> here r the counters info:
>>
>> ***************
>> Job Counters SLOTS_MILLIS_MAPS 0 0 8,386
>> Launched reduce tasks 0 0 1
>> Total time spent by all reduces waiting after reserving slots (ms) 0 0 0
>> Total time spent by all maps waiting after reserving slots (ms) 0 0 0
>> Launched map tasks 0 0 2
>> Data-local map tasks 0 0 2
>> SLOTS_MILLIS_REDUCES 0 0 9,199
>> ***************
>> My question why r there 2 launched map tasks when i have only a small file.
>> Per my understanding it is only 1 block.
>> and should be only 1 split.
>> Then for each line a map computation should occur
>> but it shows 2 map tasks.
>> Please let me know.
>> Thanks
>> Sai
>>
Re: 2 Map tasks running for a small input file
Posted by Shekhar Sharma <sh...@gmail.com>.
Number of map tasks on a mapreduce job doesnt depend on this
property..it depends on the number of input splits...( or equal to
number blocks if input split size = block size)
1. What is the input format you are using? if yes what is the value of
N, you are using?
2. WHat is the propety mapred.min.split.size? have you changed to
something else or is it default which is 1?
Regards,
Som Shekhar Sharma
+91-8197243810
On Thu, Sep 26, 2013 at 4:39 PM, Viji R <vi...@cloudera.com> wrote:
> Hi,
>
> Default number of map tasks is 2. You can set mapred.map.tasks to 1 to
> avoid this.
>
> Regards,
> Viji
>
> On Thu, Sep 26, 2013 at 4:28 PM, Sai Sai <sa...@yahoo.in> wrote:
>> Hi
>> Here is the input file for the wordcount job:
>> ******************
>> Hi This is a simple test.
>> Hi Hadoop how r u.
>> Hello Hello.
>> Hi Hi.
>> Hadoop Hadoop Welcome.
>> ******************
>>
>> After running the wordcount successfully
>> here r the counters info:
>>
>> ***************
>> Job Counters SLOTS_MILLIS_MAPS 0 0 8,386
>> Launched reduce tasks 0 0 1
>> Total time spent by all reduces waiting after reserving slots (ms) 0 0 0
>> Total time spent by all maps waiting after reserving slots (ms) 0 0 0
>> Launched map tasks 0 0 2
>> Data-local map tasks 0 0 2
>> SLOTS_MILLIS_REDUCES 0 0 9,199
>> ***************
>> My question why r there 2 launched map tasks when i have only a small file.
>> Per my understanding it is only 1 block.
>> and should be only 1 split.
>> Then for each line a map computation should occur
>> but it shows 2 map tasks.
>> Please let me know.
>> Thanks
>> Sai
>>
Re: 2 Map tasks running for a small input file
Posted by Sai Sai <sa...@yahoo.in>.
Thanks Viji.
I am confused a little when the data is small y would there b 2 tasks.
U will use the min as 2 if u need it but in this case it is not needed due to size of the data being small
so y would 2 map tasks exec.
Since it results in 1 block with 5 lines of data in it
i am assuming this results in 5 map computations 1 per each line
and all of em in 1 process/node since i m using a pseudo vm.
Where is the second task coming from.
The 5 computations of map on each line is 1 task.
Is this right.
Please help.
Thanks
________________________________
From: Viji R <vi...@cloudera.com>
To: user@hadoop.apache.org; Sai Sai <sa...@yahoo.in>
Sent: Thursday, 26 September 2013 5:09 PM
Subject: Re: 2 Map tasks running for a small input file
Hi,
Default number of map tasks is 2. You can set mapred.map.tasks to 1 to
avoid this.
Regards,
Viji
On Thu, Sep 26, 2013 at 4:28 PM, Sai Sai <sa...@yahoo.in> wrote:
> Hi
> Here is the input file for the wordcount job:
> ******************
> Hi This is a simple test.
> Hi Hadoop how r u.
> Hello Hello.
> Hi Hi.
> Hadoop Hadoop Welcome.
> ******************
>
> After running the wordcount successfully
> here r the counters info:
>
> ***************
> Job Counters SLOTS_MILLIS_MAPS 0 0 8,386
> Launched reduce tasks 0 0 1
> Total time spent by all reduces waiting after reserving slots (ms) 0 0 0
> Total time spent by all maps waiting after reserving slots (ms) 0 0 0
> Launched map tasks 0 0 2
> Data-local map tasks 0 0 2
> SLOTS_MILLIS_REDUCES 0 0 9,199
> ***************
> My question why r there 2 launched map tasks when i have only a small file.
> Per my understanding it is only 1 block.
> and should be only 1 split.
> Then for each line a map computation should occur
> but it shows 2 map tasks.
> Please let me know.
> Thanks
> Sai
>
Re: 2 Map tasks running for a small input file
Posted by Viji R <vi...@cloudera.com>.
Hi,
Default number of map tasks is 2. You can set mapred.map.tasks to 1 to
avoid this.
Regards,
Viji
On Thu, Sep 26, 2013 at 4:28 PM, Sai Sai <sa...@yahoo.in> wrote:
> Hi
> Here is the input file for the wordcount job:
> ******************
> Hi This is a simple test.
> Hi Hadoop how r u.
> Hello Hello.
> Hi Hi.
> Hadoop Hadoop Welcome.
> ******************
>
> After running the wordcount successfully
> here r the counters info:
>
> ***************
> Job Counters SLOTS_MILLIS_MAPS 0 0 8,386
> Launched reduce tasks 0 0 1
> Total time spent by all reduces waiting after reserving slots (ms) 0 0 0
> Total time spent by all maps waiting after reserving slots (ms) 0 0 0
> Launched map tasks 0 0 2
> Data-local map tasks 0 0 2
> SLOTS_MILLIS_REDUCES 0 0 9,199
> ***************
> My question why r there 2 launched map tasks when i have only a small file.
> Per my understanding it is only 1 block.
> and should be only 1 split.
> Then for each line a map computation should occur
> but it shows 2 map tasks.
> Please let me know.
> Thanks
> Sai
>
Re: 2 Map tasks running for a small input file
Posted by Viji R <vi...@cloudera.com>.
Hi,
Default number of map tasks is 2. You can set mapred.map.tasks to 1 to
avoid this.
Regards,
Viji
On Thu, Sep 26, 2013 at 4:28 PM, Sai Sai <sa...@yahoo.in> wrote:
> Hi
> Here is the input file for the wordcount job:
> ******************
> Hi This is a simple test.
> Hi Hadoop how r u.
> Hello Hello.
> Hi Hi.
> Hadoop Hadoop Welcome.
> ******************
>
> After running the wordcount successfully
> here r the counters info:
>
> ***************
> Job Counters SLOTS_MILLIS_MAPS 0 0 8,386
> Launched reduce tasks 0 0 1
> Total time spent by all reduces waiting after reserving slots (ms) 0 0 0
> Total time spent by all maps waiting after reserving slots (ms) 0 0 0
> Launched map tasks 0 0 2
> Data-local map tasks 0 0 2
> SLOTS_MILLIS_REDUCES 0 0 9,199
> ***************
> My question why r there 2 launched map tasks when i have only a small file.
> Per my understanding it is only 1 block.
> and should be only 1 split.
> Then for each line a map computation should occur
> but it shows 2 map tasks.
> Please let me know.
> Thanks
> Sai
>
Re: Input Split vs Task vs attempt vs computation
Posted by Sai Sai <sa...@yahoo.in>.
Hi
I have a few questions i am trying to understand:
1. Is each input split same as a record, (a rec can be a single line or multiple lines).
2. Is each Task a collection of few computations or attempts.
For ex: if i have a small file with 5 lines.
By default there will be 1 line on which each map computation is performed.
So totally 5 computations r done on 1 node.
This means JT will spawn 1 JVM for 1 Tasktracker on a node
and another JVM for map task which will instantiate 5 map objects 1 for each line.
The MT JVM is called the task which will have 5 attempts for each line.
This means attempt is same as computation.
Please let me know if anything is incorrect.
Thanks
Sai
Re: 2 Map tasks running for a small input file
Posted by Viji R <vi...@cloudera.com>.
Hi,
Default number of map tasks is 2. You can set mapred.map.tasks to 1 to
avoid this.
Regards,
Viji
On Thu, Sep 26, 2013 at 4:28 PM, Sai Sai <sa...@yahoo.in> wrote:
> Hi
> Here is the input file for the wordcount job:
> ******************
> Hi This is a simple test.
> Hi Hadoop how r u.
> Hello Hello.
> Hi Hi.
> Hadoop Hadoop Welcome.
> ******************
>
> After running the wordcount successfully
> here r the counters info:
>
> ***************
> Job Counters SLOTS_MILLIS_MAPS 0 0 8,386
> Launched reduce tasks 0 0 1
> Total time spent by all reduces waiting after reserving slots (ms) 0 0 0
> Total time spent by all maps waiting after reserving slots (ms) 0 0 0
> Launched map tasks 0 0 2
> Data-local map tasks 0 0 2
> SLOTS_MILLIS_REDUCES 0 0 9,199
> ***************
> My question why r there 2 launched map tasks when i have only a small file.
> Per my understanding it is only 1 block.
> and should be only 1 split.
> Then for each line a map computation should occur
> but it shows 2 map tasks.
> Please let me know.
> Thanks
> Sai
>
Re: Input Split vs Task vs attempt vs computation
Posted by Sai Sai <sa...@yahoo.in>.
Hi
I have a few questions i am trying to understand:
1. Is each input split same as a record, (a rec can be a single line or multiple lines).
2. Is each Task a collection of few computations or attempts.
For ex: if i have a small file with 5 lines.
By default there will be 1 line on which each map computation is performed.
So totally 5 computations r done on 1 node.
This means JT will spawn 1 JVM for 1 Tasktracker on a node
and another JVM for map task which will instantiate 5 map objects 1 for each line.
The MT JVM is called the task which will have 5 attempts for each line.
This means attempt is same as computation.
Please let me know if anything is incorrect.
Thanks
Sai
Re: Input Split vs Task vs attempt vs computation
Posted by Sai Sai <sa...@yahoo.in>.
Hi
I have a few questions i am trying to understand:
1. Is each input split same as a record, (a rec can be a single line or multiple lines).
2. Is each Task a collection of few computations or attempts.
For ex: if i have a small file with 5 lines.
By default there will be 1 line on which each map computation is performed.
So totally 5 computations r done on 1 node.
This means JT will spawn 1 JVM for 1 Tasktracker on a node
and another JVM for map task which will instantiate 5 map objects 1 for each line.
The MT JVM is called the task which will have 5 attempts for each line.
This means attempt is same as computation.
Please let me know if anything is incorrect.
Thanks
Sai
Re: Retrieve and compute input splits
Posted by Sai Sai <sa...@yahoo.in>.
Thanks for your suggestions and replies.
I am still confused about this:
To create the list of tasks to run, the job scheduler first retrieves the input splits computed by the JobClient from the shared filesystem (step 6).
My question:
Does the input split in the above statement refer to the physical block or the logical input split.
I undersstand that the client will split the file and save the blocks at the time of writing the file to the cluster and the meta data
about the blocks is in Namenode.
The only place where the meta data about the blocks is in NN so can v assume in step 6 is the scheduler goes to
NN for retrieving this meta data from NN and thats what is indicated in the diagram as Shared File System HDFS.
And if this is right the input split is the physical blocks info and not the logical input split info which could be just a single line
if v r using TextInuptFormat the default one.
Any suggestions.
Thanks
Sai
________________________________
From: Jay Vyas <ja...@gmail.com>
To: "common-user@hadoop.apache.org" <us...@hadoop.apache.org>
Cc: Sai Sai <sa...@yahoo.in>
Sent: Saturday, 28 September 2013 5:35 AM
Subject: Re: Retrieve and compute input splits
Technically, the block locations are provided by the InputSplit which in the FileInputFormat case, is provided by the FileSystem Interface.
http://hadoop.apache.org/docs/current/api/org/apache/hadoop/mapred/InputSplit.html
The thing to realize here is that the FileSystem implementation is provided at runtime - so the InputSplit class is responsible to create a FileSystem implementation using reflection, and then call the getBlockLocations of on a given file or set of files which the input split is corresponding to.
I think your confusion here lies in the fact that the input splits create a filesystem, however, they dont know what the filesystem implementation actually is - they only rely on the abstract contract, which provides a set of block locations.
See the FileSystem abstract class for details on that.
On Fri, Sep 27, 2013 at 7:02 PM, Peyman Mohajerian <mo...@gmail.com> wrote:
For the JobClient to compute the input splits doesn't it need to contact Name Node. Only Name Node knows where the splits are, how can it compute it without that additional call?
>
>
>
>
>On Fri, Sep 27, 2013 at 1:41 AM, Sonal Goyal <so...@gmail.com> wrote:
>
>The input splits are not copied, only the information on the location of the splits is copied to the jobtracker so that it can assign tasktrackers which are local to the split.
>>
>>
>>Check the Job Initialization section at
>>http://answers.oreilly.com/topic/459-anatomy-of-a-mapreduce-job-run-with-hadoop/
>>
>>
>>
>>To create the list of tasks to run, the job scheduler first retrieves the input splits computed by the JobClient from the shared filesystem (step 6). It then creates one map task for each split. The number of reduce tasks to create is determined by the mapred.reduce.tasks property in the JobConf, which is set by the setNumReduceTasks() method, and the scheduler simply creates this number of reduce tasks to be run. Tasks are given IDs at this point.
>>
>>
>>
>>Best Regards,
>>Sonal
>>Nube Technologies
>>
>>
>>
>>
>>
>>
>>
>>
>>On Fri, Sep 27, 2013 at 10:55 AM, Sai Sai <sa...@yahoo.in> wrote:
>>
>>Hi
>>>I have attached the anatomy of MR from definitive guide.
>>>
>>>
>>>In step 6 it says JT/Scheduler retrieve input splits computed by the client from hdfs.
>>>
>>>
>>>In the above line it refers to as the client computes input splits.
>>>
>>>
>>>
>>>1. Why does the JT/Scheduler retrieve the input splits and what does it do.
>>>If it is retrieving the input split does this mean it goes to the block and reads each record
>>>and gets the record back to JT. If so this is a lot of data movement for large files.
>>>which is not data locality. so i m getting confused.
>>>
>>>
>>>2. How does the client know how to calculate the input splits.
>>>
>>>
>>>Any help please.
>>>ThanksSai
>>
>
--
Jay Vyas
http://jayunit100.blogspot.com
Re: Retrieve and compute input splits
Posted by Sai Sai <sa...@yahoo.in>.
Thanks for your suggestions and replies.
I am still confused about this:
To create the list of tasks to run, the job scheduler first retrieves the input splits computed by the JobClient from the shared filesystem (step 6).
My question:
Does the input split in the above statement refer to the physical block or the logical input split.
I undersstand that the client will split the file and save the blocks at the time of writing the file to the cluster and the meta data
about the blocks is in Namenode.
The only place where the meta data about the blocks is in NN so can v assume in step 6 is the scheduler goes to
NN for retrieving this meta data from NN and thats what is indicated in the diagram as Shared File System HDFS.
And if this is right the input split is the physical blocks info and not the logical input split info which could be just a single line
if v r using TextInuptFormat the default one.
Any suggestions.
Thanks
Sai
________________________________
From: Jay Vyas <ja...@gmail.com>
To: "common-user@hadoop.apache.org" <us...@hadoop.apache.org>
Cc: Sai Sai <sa...@yahoo.in>
Sent: Saturday, 28 September 2013 5:35 AM
Subject: Re: Retrieve and compute input splits
Technically, the block locations are provided by the InputSplit which in the FileInputFormat case, is provided by the FileSystem Interface.
http://hadoop.apache.org/docs/current/api/org/apache/hadoop/mapred/InputSplit.html
The thing to realize here is that the FileSystem implementation is provided at runtime - so the InputSplit class is responsible to create a FileSystem implementation using reflection, and then call the getBlockLocations of on a given file or set of files which the input split is corresponding to.
I think your confusion here lies in the fact that the input splits create a filesystem, however, they dont know what the filesystem implementation actually is - they only rely on the abstract contract, which provides a set of block locations.
See the FileSystem abstract class for details on that.
On Fri, Sep 27, 2013 at 7:02 PM, Peyman Mohajerian <mo...@gmail.com> wrote:
For the JobClient to compute the input splits doesn't it need to contact Name Node. Only Name Node knows where the splits are, how can it compute it without that additional call?
>
>
>
>
>On Fri, Sep 27, 2013 at 1:41 AM, Sonal Goyal <so...@gmail.com> wrote:
>
>The input splits are not copied, only the information on the location of the splits is copied to the jobtracker so that it can assign tasktrackers which are local to the split.
>>
>>
>>Check the Job Initialization section at
>>http://answers.oreilly.com/topic/459-anatomy-of-a-mapreduce-job-run-with-hadoop/
>>
>>
>>
>>To create the list of tasks to run, the job scheduler first retrieves the input splits computed by the JobClient from the shared filesystem (step 6). It then creates one map task for each split. The number of reduce tasks to create is determined by the mapred.reduce.tasks property in the JobConf, which is set by the setNumReduceTasks() method, and the scheduler simply creates this number of reduce tasks to be run. Tasks are given IDs at this point.
>>
>>
>>
>>Best Regards,
>>Sonal
>>Nube Technologies
>>
>>
>>
>>
>>
>>
>>
>>
>>On Fri, Sep 27, 2013 at 10:55 AM, Sai Sai <sa...@yahoo.in> wrote:
>>
>>Hi
>>>I have attached the anatomy of MR from definitive guide.
>>>
>>>
>>>In step 6 it says JT/Scheduler retrieve input splits computed by the client from hdfs.
>>>
>>>
>>>In the above line it refers to as the client computes input splits.
>>>
>>>
>>>
>>>1. Why does the JT/Scheduler retrieve the input splits and what does it do.
>>>If it is retrieving the input split does this mean it goes to the block and reads each record
>>>and gets the record back to JT. If so this is a lot of data movement for large files.
>>>which is not data locality. so i m getting confused.
>>>
>>>
>>>2. How does the client know how to calculate the input splits.
>>>
>>>
>>>Any help please.
>>>ThanksSai
>>
>
--
Jay Vyas
http://jayunit100.blogspot.com
Re: Retrieve and compute input splits
Posted by Sai Sai <sa...@yahoo.in>.
Thanks for your suggestions and replies.
I am still confused about this:
To create the list of tasks to run, the job scheduler first retrieves the input splits computed by the JobClient from the shared filesystem (step 6).
My question:
Does the input split in the above statement refer to the physical block or the logical input split.
I undersstand that the client will split the file and save the blocks at the time of writing the file to the cluster and the meta data
about the blocks is in Namenode.
The only place where the meta data about the blocks is in NN so can v assume in step 6 is the scheduler goes to
NN for retrieving this meta data from NN and thats what is indicated in the diagram as Shared File System HDFS.
And if this is right the input split is the physical blocks info and not the logical input split info which could be just a single line
if v r using TextInuptFormat the default one.
Any suggestions.
Thanks
Sai
________________________________
From: Jay Vyas <ja...@gmail.com>
To: "common-user@hadoop.apache.org" <us...@hadoop.apache.org>
Cc: Sai Sai <sa...@yahoo.in>
Sent: Saturday, 28 September 2013 5:35 AM
Subject: Re: Retrieve and compute input splits
Technically, the block locations are provided by the InputSplit which in the FileInputFormat case, is provided by the FileSystem Interface.
http://hadoop.apache.org/docs/current/api/org/apache/hadoop/mapred/InputSplit.html
The thing to realize here is that the FileSystem implementation is provided at runtime - so the InputSplit class is responsible to create a FileSystem implementation using reflection, and then call the getBlockLocations of on a given file or set of files which the input split is corresponding to.
I think your confusion here lies in the fact that the input splits create a filesystem, however, they dont know what the filesystem implementation actually is - they only rely on the abstract contract, which provides a set of block locations.
See the FileSystem abstract class for details on that.
On Fri, Sep 27, 2013 at 7:02 PM, Peyman Mohajerian <mo...@gmail.com> wrote:
For the JobClient to compute the input splits doesn't it need to contact Name Node. Only Name Node knows where the splits are, how can it compute it without that additional call?
>
>
>
>
>On Fri, Sep 27, 2013 at 1:41 AM, Sonal Goyal <so...@gmail.com> wrote:
>
>The input splits are not copied, only the information on the location of the splits is copied to the jobtracker so that it can assign tasktrackers which are local to the split.
>>
>>
>>Check the Job Initialization section at
>>http://answers.oreilly.com/topic/459-anatomy-of-a-mapreduce-job-run-with-hadoop/
>>
>>
>>
>>To create the list of tasks to run, the job scheduler first retrieves the input splits computed by the JobClient from the shared filesystem (step 6). It then creates one map task for each split. The number of reduce tasks to create is determined by the mapred.reduce.tasks property in the JobConf, which is set by the setNumReduceTasks() method, and the scheduler simply creates this number of reduce tasks to be run. Tasks are given IDs at this point.
>>
>>
>>
>>Best Regards,
>>Sonal
>>Nube Technologies
>>
>>
>>
>>
>>
>>
>>
>>
>>On Fri, Sep 27, 2013 at 10:55 AM, Sai Sai <sa...@yahoo.in> wrote:
>>
>>Hi
>>>I have attached the anatomy of MR from definitive guide.
>>>
>>>
>>>In step 6 it says JT/Scheduler retrieve input splits computed by the client from hdfs.
>>>
>>>
>>>In the above line it refers to as the client computes input splits.
>>>
>>>
>>>
>>>1. Why does the JT/Scheduler retrieve the input splits and what does it do.
>>>If it is retrieving the input split does this mean it goes to the block and reads each record
>>>and gets the record back to JT. If so this is a lot of data movement for large files.
>>>which is not data locality. so i m getting confused.
>>>
>>>
>>>2. How does the client know how to calculate the input splits.
>>>
>>>
>>>Any help please.
>>>ThanksSai
>>
>
--
Jay Vyas
http://jayunit100.blogspot.com
Re: Retrieve and compute input splits
Posted by Sai Sai <sa...@yahoo.in>.
Thanks for your suggestions and replies.
I am still confused about this:
To create the list of tasks to run, the job scheduler first retrieves the input splits computed by the JobClient from the shared filesystem (step 6).
My question:
Does the input split in the above statement refer to the physical block or the logical input split.
I undersstand that the client will split the file and save the blocks at the time of writing the file to the cluster and the meta data
about the blocks is in Namenode.
The only place where the meta data about the blocks is in NN so can v assume in step 6 is the scheduler goes to
NN for retrieving this meta data from NN and thats what is indicated in the diagram as Shared File System HDFS.
And if this is right the input split is the physical blocks info and not the logical input split info which could be just a single line
if v r using TextInuptFormat the default one.
Any suggestions.
Thanks
Sai
________________________________
From: Jay Vyas <ja...@gmail.com>
To: "common-user@hadoop.apache.org" <us...@hadoop.apache.org>
Cc: Sai Sai <sa...@yahoo.in>
Sent: Saturday, 28 September 2013 5:35 AM
Subject: Re: Retrieve and compute input splits
Technically, the block locations are provided by the InputSplit which in the FileInputFormat case, is provided by the FileSystem Interface.
http://hadoop.apache.org/docs/current/api/org/apache/hadoop/mapred/InputSplit.html
The thing to realize here is that the FileSystem implementation is provided at runtime - so the InputSplit class is responsible to create a FileSystem implementation using reflection, and then call the getBlockLocations of on a given file or set of files which the input split is corresponding to.
I think your confusion here lies in the fact that the input splits create a filesystem, however, they dont know what the filesystem implementation actually is - they only rely on the abstract contract, which provides a set of block locations.
See the FileSystem abstract class for details on that.
On Fri, Sep 27, 2013 at 7:02 PM, Peyman Mohajerian <mo...@gmail.com> wrote:
For the JobClient to compute the input splits doesn't it need to contact Name Node. Only Name Node knows where the splits are, how can it compute it without that additional call?
>
>
>
>
>On Fri, Sep 27, 2013 at 1:41 AM, Sonal Goyal <so...@gmail.com> wrote:
>
>The input splits are not copied, only the information on the location of the splits is copied to the jobtracker so that it can assign tasktrackers which are local to the split.
>>
>>
>>Check the Job Initialization section at
>>http://answers.oreilly.com/topic/459-anatomy-of-a-mapreduce-job-run-with-hadoop/
>>
>>
>>
>>To create the list of tasks to run, the job scheduler first retrieves the input splits computed by the JobClient from the shared filesystem (step 6). It then creates one map task for each split. The number of reduce tasks to create is determined by the mapred.reduce.tasks property in the JobConf, which is set by the setNumReduceTasks() method, and the scheduler simply creates this number of reduce tasks to be run. Tasks are given IDs at this point.
>>
>>
>>
>>Best Regards,
>>Sonal
>>Nube Technologies
>>
>>
>>
>>
>>
>>
>>
>>
>>On Fri, Sep 27, 2013 at 10:55 AM, Sai Sai <sa...@yahoo.in> wrote:
>>
>>Hi
>>>I have attached the anatomy of MR from definitive guide.
>>>
>>>
>>>In step 6 it says JT/Scheduler retrieve input splits computed by the client from hdfs.
>>>
>>>
>>>In the above line it refers to as the client computes input splits.
>>>
>>>
>>>
>>>1. Why does the JT/Scheduler retrieve the input splits and what does it do.
>>>If it is retrieving the input split does this mean it goes to the block and reads each record
>>>and gets the record back to JT. If so this is a lot of data movement for large files.
>>>which is not data locality. so i m getting confused.
>>>
>>>
>>>2. How does the client know how to calculate the input splits.
>>>
>>>
>>>Any help please.
>>>ThanksSai
>>
>
--
Jay Vyas
http://jayunit100.blogspot.com
Re: Retrieve and compute input splits
Posted by Jay Vyas <ja...@gmail.com>.
Technically, the block locations are provided by the InputSplit which in
the FileInputFormat case, is provided by the FileSystem Interface.
http://hadoop.apache.org/docs/current/api/org/apache/hadoop/mapred/InputSplit.html
The thing to realize here is that the FileSystem implementation is provided
at runtime - so the InputSplit class is responsible to create a FileSystem
implementation using reflection, and then call the getBlockLocations of on
a given file or set of files which the input split is corresponding to.
I think your confusion here lies in the fact that the input splits create a
filesystem, however, they dont know what the filesystem implementation
actually is - they only rely on the abstract contract, which provides a set
of block locations.
See the FileSystem abstract class for details on that.
On Fri, Sep 27, 2013 at 7:02 PM, Peyman Mohajerian <mo...@gmail.com>wrote:
> For the JobClient to compute the input splits doesn't it need to contact
> Name Node. Only Name Node knows where the splits are, how can it compute it
> without that additional call?
>
>
> On Fri, Sep 27, 2013 at 1:41 AM, Sonal Goyal <so...@gmail.com>wrote:
>
>> The input splits are not copied, only the information on the location of
>> the splits is copied to the jobtracker so that it can assign tasktrackers
>> which are local to the split.
>>
>> Check the Job Initialization section at
>>
>> http://answers.oreilly.com/topic/459-anatomy-of-a-mapreduce-job-run-with-hadoop/
>>
>> To create the list of tasks to run, the job scheduler first retrieves
>> the input splits computed by the JobClient from the shared filesystem
>> (step 6). It then creates one map task for each split. The number of reduce
>> tasks to create is determined by the mapred.reduce.tasks property in the
>> JobConf, which is set by the setNumReduceTasks() method, and the
>> scheduler simply creates this number of reduce tasks to be run. Tasks are
>> given IDs at this point.
>>
>> Best Regards,
>> Sonal
>> Nube Technologies <http://www.nubetech.co>
>>
>> <http://in.linkedin.com/in/sonalgoyal>
>>
>>
>>
>>
>> On Fri, Sep 27, 2013 at 10:55 AM, Sai Sai <sa...@yahoo.in> wrote:
>>
>>> Hi
>>> I have attached the anatomy of MR from definitive guide.
>>>
>>> In step 6 it says JT/Scheduler retrieve input splits computed by the
>>> client from hdfs.
>>>
>>> In the above line it refers to as the client computes input splits.
>>>
>>> 1. Why does the JT/Scheduler retrieve the input splits and what does it
>>> do.
>>> If it is retrieving the input split does this mean it goes to the block
>>> and reads each record
>>> and gets the record back to JT. If so this is a lot of data movement for
>>> large files.
>>> which is not data locality. so i m getting confused.
>>>
>>> 2. How does the client know how to calculate the input splits.
>>>
>>> Any help please.
>>> Thanks
>>> Sai
>>>
>>
>>
>
--
Jay Vyas
http://jayunit100.blogspot.com
Re: Retrieve and compute input splits
Posted by Jay Vyas <ja...@gmail.com>.
Technically, the block locations are provided by the InputSplit which in
the FileInputFormat case, is provided by the FileSystem Interface.
http://hadoop.apache.org/docs/current/api/org/apache/hadoop/mapred/InputSplit.html
The thing to realize here is that the FileSystem implementation is provided
at runtime - so the InputSplit class is responsible to create a FileSystem
implementation using reflection, and then call the getBlockLocations of on
a given file or set of files which the input split is corresponding to.
I think your confusion here lies in the fact that the input splits create a
filesystem, however, they dont know what the filesystem implementation
actually is - they only rely on the abstract contract, which provides a set
of block locations.
See the FileSystem abstract class for details on that.
On Fri, Sep 27, 2013 at 7:02 PM, Peyman Mohajerian <mo...@gmail.com>wrote:
> For the JobClient to compute the input splits doesn't it need to contact
> Name Node. Only Name Node knows where the splits are, how can it compute it
> without that additional call?
>
>
> On Fri, Sep 27, 2013 at 1:41 AM, Sonal Goyal <so...@gmail.com>wrote:
>
>> The input splits are not copied, only the information on the location of
>> the splits is copied to the jobtracker so that it can assign tasktrackers
>> which are local to the split.
>>
>> Check the Job Initialization section at
>>
>> http://answers.oreilly.com/topic/459-anatomy-of-a-mapreduce-job-run-with-hadoop/
>>
>> To create the list of tasks to run, the job scheduler first retrieves
>> the input splits computed by the JobClient from the shared filesystem
>> (step 6). It then creates one map task for each split. The number of reduce
>> tasks to create is determined by the mapred.reduce.tasks property in the
>> JobConf, which is set by the setNumReduceTasks() method, and the
>> scheduler simply creates this number of reduce tasks to be run. Tasks are
>> given IDs at this point.
>>
>> Best Regards,
>> Sonal
>> Nube Technologies <http://www.nubetech.co>
>>
>> <http://in.linkedin.com/in/sonalgoyal>
>>
>>
>>
>>
>> On Fri, Sep 27, 2013 at 10:55 AM, Sai Sai <sa...@yahoo.in> wrote:
>>
>>> Hi
>>> I have attached the anatomy of MR from definitive guide.
>>>
>>> In step 6 it says JT/Scheduler retrieve input splits computed by the
>>> client from hdfs.
>>>
>>> In the above line it refers to as the client computes input splits.
>>>
>>> 1. Why does the JT/Scheduler retrieve the input splits and what does it
>>> do.
>>> If it is retrieving the input split does this mean it goes to the block
>>> and reads each record
>>> and gets the record back to JT. If so this is a lot of data movement for
>>> large files.
>>> which is not data locality. so i m getting confused.
>>>
>>> 2. How does the client know how to calculate the input splits.
>>>
>>> Any help please.
>>> Thanks
>>> Sai
>>>
>>
>>
>
--
Jay Vyas
http://jayunit100.blogspot.com
Re: Retrieve and compute input splits
Posted by Jay Vyas <ja...@gmail.com>.
Technically, the block locations are provided by the InputSplit which in
the FileInputFormat case, is provided by the FileSystem Interface.
http://hadoop.apache.org/docs/current/api/org/apache/hadoop/mapred/InputSplit.html
The thing to realize here is that the FileSystem implementation is provided
at runtime - so the InputSplit class is responsible to create a FileSystem
implementation using reflection, and then call the getBlockLocations of on
a given file or set of files which the input split is corresponding to.
I think your confusion here lies in the fact that the input splits create a
filesystem, however, they dont know what the filesystem implementation
actually is - they only rely on the abstract contract, which provides a set
of block locations.
See the FileSystem abstract class for details on that.
On Fri, Sep 27, 2013 at 7:02 PM, Peyman Mohajerian <mo...@gmail.com>wrote:
> For the JobClient to compute the input splits doesn't it need to contact
> Name Node. Only Name Node knows where the splits are, how can it compute it
> without that additional call?
>
>
> On Fri, Sep 27, 2013 at 1:41 AM, Sonal Goyal <so...@gmail.com>wrote:
>
>> The input splits are not copied, only the information on the location of
>> the splits is copied to the jobtracker so that it can assign tasktrackers
>> which are local to the split.
>>
>> Check the Job Initialization section at
>>
>> http://answers.oreilly.com/topic/459-anatomy-of-a-mapreduce-job-run-with-hadoop/
>>
>> To create the list of tasks to run, the job scheduler first retrieves
>> the input splits computed by the JobClient from the shared filesystem
>> (step 6). It then creates one map task for each split. The number of reduce
>> tasks to create is determined by the mapred.reduce.tasks property in the
>> JobConf, which is set by the setNumReduceTasks() method, and the
>> scheduler simply creates this number of reduce tasks to be run. Tasks are
>> given IDs at this point.
>>
>> Best Regards,
>> Sonal
>> Nube Technologies <http://www.nubetech.co>
>>
>> <http://in.linkedin.com/in/sonalgoyal>
>>
>>
>>
>>
>> On Fri, Sep 27, 2013 at 10:55 AM, Sai Sai <sa...@yahoo.in> wrote:
>>
>>> Hi
>>> I have attached the anatomy of MR from definitive guide.
>>>
>>> In step 6 it says JT/Scheduler retrieve input splits computed by the
>>> client from hdfs.
>>>
>>> In the above line it refers to as the client computes input splits.
>>>
>>> 1. Why does the JT/Scheduler retrieve the input splits and what does it
>>> do.
>>> If it is retrieving the input split does this mean it goes to the block
>>> and reads each record
>>> and gets the record back to JT. If so this is a lot of data movement for
>>> large files.
>>> which is not data locality. so i m getting confused.
>>>
>>> 2. How does the client know how to calculate the input splits.
>>>
>>> Any help please.
>>> Thanks
>>> Sai
>>>
>>
>>
>
--
Jay Vyas
http://jayunit100.blogspot.com
Re: Retrieve and compute input splits
Posted by Jay Vyas <ja...@gmail.com>.
Technically, the block locations are provided by the InputSplit which in
the FileInputFormat case, is provided by the FileSystem Interface.
http://hadoop.apache.org/docs/current/api/org/apache/hadoop/mapred/InputSplit.html
The thing to realize here is that the FileSystem implementation is provided
at runtime - so the InputSplit class is responsible to create a FileSystem
implementation using reflection, and then call the getBlockLocations of on
a given file or set of files which the input split is corresponding to.
I think your confusion here lies in the fact that the input splits create a
filesystem, however, they dont know what the filesystem implementation
actually is - they only rely on the abstract contract, which provides a set
of block locations.
See the FileSystem abstract class for details on that.
On Fri, Sep 27, 2013 at 7:02 PM, Peyman Mohajerian <mo...@gmail.com>wrote:
> For the JobClient to compute the input splits doesn't it need to contact
> Name Node. Only Name Node knows where the splits are, how can it compute it
> without that additional call?
>
>
> On Fri, Sep 27, 2013 at 1:41 AM, Sonal Goyal <so...@gmail.com>wrote:
>
>> The input splits are not copied, only the information on the location of
>> the splits is copied to the jobtracker so that it can assign tasktrackers
>> which are local to the split.
>>
>> Check the Job Initialization section at
>>
>> http://answers.oreilly.com/topic/459-anatomy-of-a-mapreduce-job-run-with-hadoop/
>>
>> To create the list of tasks to run, the job scheduler first retrieves
>> the input splits computed by the JobClient from the shared filesystem
>> (step 6). It then creates one map task for each split. The number of reduce
>> tasks to create is determined by the mapred.reduce.tasks property in the
>> JobConf, which is set by the setNumReduceTasks() method, and the
>> scheduler simply creates this number of reduce tasks to be run. Tasks are
>> given IDs at this point.
>>
>> Best Regards,
>> Sonal
>> Nube Technologies <http://www.nubetech.co>
>>
>> <http://in.linkedin.com/in/sonalgoyal>
>>
>>
>>
>>
>> On Fri, Sep 27, 2013 at 10:55 AM, Sai Sai <sa...@yahoo.in> wrote:
>>
>>> Hi
>>> I have attached the anatomy of MR from definitive guide.
>>>
>>> In step 6 it says JT/Scheduler retrieve input splits computed by the
>>> client from hdfs.
>>>
>>> In the above line it refers to as the client computes input splits.
>>>
>>> 1. Why does the JT/Scheduler retrieve the input splits and what does it
>>> do.
>>> If it is retrieving the input split does this mean it goes to the block
>>> and reads each record
>>> and gets the record back to JT. If so this is a lot of data movement for
>>> large files.
>>> which is not data locality. so i m getting confused.
>>>
>>> 2. How does the client know how to calculate the input splits.
>>>
>>> Any help please.
>>> Thanks
>>> Sai
>>>
>>
>>
>
--
Jay Vyas
http://jayunit100.blogspot.com
Re: Retrieve and compute input splits
Posted by Peyman Mohajerian <mo...@gmail.com>.
For the JobClient to compute the input splits doesn't it need to contact
Name Node. Only Name Node knows where the splits are, how can it compute it
without that additional call?
On Fri, Sep 27, 2013 at 1:41 AM, Sonal Goyal <so...@gmail.com> wrote:
> The input splits are not copied, only the information on the location of
> the splits is copied to the jobtracker so that it can assign tasktrackers
> which are local to the split.
>
> Check the Job Initialization section at
>
> http://answers.oreilly.com/topic/459-anatomy-of-a-mapreduce-job-run-with-hadoop/
>
> To create the list of tasks to run, the job scheduler first retrieves the
> input splits computed by the JobClient from the shared filesystem (step
> 6). It then creates one map task for each split. The number of reduce tasks
> to create is determined by the mapred.reduce.tasks property in the JobConf,
> which is set by the setNumReduceTasks() method, and the scheduler simply
> creates this number of reduce tasks to be run. Tasks are given IDs at this
> point.
>
> Best Regards,
> Sonal
> Nube Technologies <http://www.nubetech.co>
>
> <http://in.linkedin.com/in/sonalgoyal>
>
>
>
>
> On Fri, Sep 27, 2013 at 10:55 AM, Sai Sai <sa...@yahoo.in> wrote:
>
>> Hi
>> I have attached the anatomy of MR from definitive guide.
>>
>> In step 6 it says JT/Scheduler retrieve input splits computed by the
>> client from hdfs.
>>
>> In the above line it refers to as the client computes input splits.
>>
>> 1. Why does the JT/Scheduler retrieve the input splits and what does it
>> do.
>> If it is retrieving the input split does this mean it goes to the block
>> and reads each record
>> and gets the record back to JT. If so this is a lot of data movement for
>> large files.
>> which is not data locality. so i m getting confused.
>>
>> 2. How does the client know how to calculate the input splits.
>>
>> Any help please.
>> Thanks
>> Sai
>>
>
>
Re: Retrieve and compute input splits
Posted by Peyman Mohajerian <mo...@gmail.com>.
For the JobClient to compute the input splits doesn't it need to contact
Name Node. Only Name Node knows where the splits are, how can it compute it
without that additional call?
On Fri, Sep 27, 2013 at 1:41 AM, Sonal Goyal <so...@gmail.com> wrote:
> The input splits are not copied, only the information on the location of
> the splits is copied to the jobtracker so that it can assign tasktrackers
> which are local to the split.
>
> Check the Job Initialization section at
>
> http://answers.oreilly.com/topic/459-anatomy-of-a-mapreduce-job-run-with-hadoop/
>
> To create the list of tasks to run, the job scheduler first retrieves the
> input splits computed by the JobClient from the shared filesystem (step
> 6). It then creates one map task for each split. The number of reduce tasks
> to create is determined by the mapred.reduce.tasks property in the JobConf,
> which is set by the setNumReduceTasks() method, and the scheduler simply
> creates this number of reduce tasks to be run. Tasks are given IDs at this
> point.
>
> Best Regards,
> Sonal
> Nube Technologies <http://www.nubetech.co>
>
> <http://in.linkedin.com/in/sonalgoyal>
>
>
>
>
> On Fri, Sep 27, 2013 at 10:55 AM, Sai Sai <sa...@yahoo.in> wrote:
>
>> Hi
>> I have attached the anatomy of MR from definitive guide.
>>
>> In step 6 it says JT/Scheduler retrieve input splits computed by the
>> client from hdfs.
>>
>> In the above line it refers to as the client computes input splits.
>>
>> 1. Why does the JT/Scheduler retrieve the input splits and what does it
>> do.
>> If it is retrieving the input split does this mean it goes to the block
>> and reads each record
>> and gets the record back to JT. If so this is a lot of data movement for
>> large files.
>> which is not data locality. so i m getting confused.
>>
>> 2. How does the client know how to calculate the input splits.
>>
>> Any help please.
>> Thanks
>> Sai
>>
>
>
Re: Retrieve and compute input splits
Posted by Peyman Mohajerian <mo...@gmail.com>.
For the JobClient to compute the input splits doesn't it need to contact
Name Node. Only Name Node knows where the splits are, how can it compute it
without that additional call?
On Fri, Sep 27, 2013 at 1:41 AM, Sonal Goyal <so...@gmail.com> wrote:
> The input splits are not copied, only the information on the location of
> the splits is copied to the jobtracker so that it can assign tasktrackers
> which are local to the split.
>
> Check the Job Initialization section at
>
> http://answers.oreilly.com/topic/459-anatomy-of-a-mapreduce-job-run-with-hadoop/
>
> To create the list of tasks to run, the job scheduler first retrieves the
> input splits computed by the JobClient from the shared filesystem (step
> 6). It then creates one map task for each split. The number of reduce tasks
> to create is determined by the mapred.reduce.tasks property in the JobConf,
> which is set by the setNumReduceTasks() method, and the scheduler simply
> creates this number of reduce tasks to be run. Tasks are given IDs at this
> point.
>
> Best Regards,
> Sonal
> Nube Technologies <http://www.nubetech.co>
>
> <http://in.linkedin.com/in/sonalgoyal>
>
>
>
>
> On Fri, Sep 27, 2013 at 10:55 AM, Sai Sai <sa...@yahoo.in> wrote:
>
>> Hi
>> I have attached the anatomy of MR from definitive guide.
>>
>> In step 6 it says JT/Scheduler retrieve input splits computed by the
>> client from hdfs.
>>
>> In the above line it refers to as the client computes input splits.
>>
>> 1. Why does the JT/Scheduler retrieve the input splits and what does it
>> do.
>> If it is retrieving the input split does this mean it goes to the block
>> and reads each record
>> and gets the record back to JT. If so this is a lot of data movement for
>> large files.
>> which is not data locality. so i m getting confused.
>>
>> 2. How does the client know how to calculate the input splits.
>>
>> Any help please.
>> Thanks
>> Sai
>>
>
>
Re: Retrieve and compute input splits
Posted by Peyman Mohajerian <mo...@gmail.com>.
For the JobClient to compute the input splits doesn't it need to contact
Name Node. Only Name Node knows where the splits are, how can it compute it
without that additional call?
On Fri, Sep 27, 2013 at 1:41 AM, Sonal Goyal <so...@gmail.com> wrote:
> The input splits are not copied, only the information on the location of
> the splits is copied to the jobtracker so that it can assign tasktrackers
> which are local to the split.
>
> Check the Job Initialization section at
>
> http://answers.oreilly.com/topic/459-anatomy-of-a-mapreduce-job-run-with-hadoop/
>
> To create the list of tasks to run, the job scheduler first retrieves the
> input splits computed by the JobClient from the shared filesystem (step
> 6). It then creates one map task for each split. The number of reduce tasks
> to create is determined by the mapred.reduce.tasks property in the JobConf,
> which is set by the setNumReduceTasks() method, and the scheduler simply
> creates this number of reduce tasks to be run. Tasks are given IDs at this
> point.
>
> Best Regards,
> Sonal
> Nube Technologies <http://www.nubetech.co>
>
> <http://in.linkedin.com/in/sonalgoyal>
>
>
>
>
> On Fri, Sep 27, 2013 at 10:55 AM, Sai Sai <sa...@yahoo.in> wrote:
>
>> Hi
>> I have attached the anatomy of MR from definitive guide.
>>
>> In step 6 it says JT/Scheduler retrieve input splits computed by the
>> client from hdfs.
>>
>> In the above line it refers to as the client computes input splits.
>>
>> 1. Why does the JT/Scheduler retrieve the input splits and what does it
>> do.
>> If it is retrieving the input split does this mean it goes to the block
>> and reads each record
>> and gets the record back to JT. If so this is a lot of data movement for
>> large files.
>> which is not data locality. so i m getting confused.
>>
>> 2. How does the client know how to calculate the input splits.
>>
>> Any help please.
>> Thanks
>> Sai
>>
>
>
Re: Retrieve and compute input splits
Posted by Sonal Goyal <so...@gmail.com>.
The input splits are not copied, only the information on the location of
the splits is copied to the jobtracker so that it can assign tasktrackers
which are local to the split.
Check the Job Initialization section at
http://answers.oreilly.com/topic/459-anatomy-of-a-mapreduce-job-run-with-hadoop/
To create the list of tasks to run, the job scheduler first retrieves the
input splits computed by the JobClient from the shared filesystem (step 6).
It then creates one map task for each split. The number of reduce tasks to
create is determined by the mapred.reduce.tasks property in the JobConf,
which is set by the setNumReduceTasks() method, and the scheduler simply
creates this number of reduce tasks to be run. Tasks are given IDs at this
point.
Best Regards,
Sonal
Nube Technologies <http://www.nubetech.co>
<http://in.linkedin.com/in/sonalgoyal>
On Fri, Sep 27, 2013 at 10:55 AM, Sai Sai <sa...@yahoo.in> wrote:
> Hi
> I have attached the anatomy of MR from definitive guide.
>
> In step 6 it says JT/Scheduler retrieve input splits computed by the
> client from hdfs.
>
> In the above line it refers to as the client computes input splits.
>
> 1. Why does the JT/Scheduler retrieve the input splits and what does it do.
> If it is retrieving the input split does this mean it goes to the block
> and reads each record
> and gets the record back to JT. If so this is a lot of data movement for
> large files.
> which is not data locality. so i m getting confused.
>
> 2. How does the client know how to calculate the input splits.
>
> Any help please.
> Thanks
> Sai
>
Re: Retrieve and compute input splits
Posted by Sonal Goyal <so...@gmail.com>.
The input splits are not copied, only the information on the location of
the splits is copied to the jobtracker so that it can assign tasktrackers
which are local to the split.
Check the Job Initialization section at
http://answers.oreilly.com/topic/459-anatomy-of-a-mapreduce-job-run-with-hadoop/
To create the list of tasks to run, the job scheduler first retrieves the
input splits computed by the JobClient from the shared filesystem (step 6).
It then creates one map task for each split. The number of reduce tasks to
create is determined by the mapred.reduce.tasks property in the JobConf,
which is set by the setNumReduceTasks() method, and the scheduler simply
creates this number of reduce tasks to be run. Tasks are given IDs at this
point.
Best Regards,
Sonal
Nube Technologies <http://www.nubetech.co>
<http://in.linkedin.com/in/sonalgoyal>
On Fri, Sep 27, 2013 at 10:55 AM, Sai Sai <sa...@yahoo.in> wrote:
> Hi
> I have attached the anatomy of MR from definitive guide.
>
> In step 6 it says JT/Scheduler retrieve input splits computed by the
> client from hdfs.
>
> In the above line it refers to as the client computes input splits.
>
> 1. Why does the JT/Scheduler retrieve the input splits and what does it do.
> If it is retrieving the input split does this mean it goes to the block
> and reads each record
> and gets the record back to JT. If so this is a lot of data movement for
> large files.
> which is not data locality. so i m getting confused.
>
> 2. How does the client know how to calculate the input splits.
>
> Any help please.
> Thanks
> Sai
>
Re: Retrieve and compute input splits
Posted by Sonal Goyal <so...@gmail.com>.
The input splits are not copied, only the information on the location of
the splits is copied to the jobtracker so that it can assign tasktrackers
which are local to the split.
Check the Job Initialization section at
http://answers.oreilly.com/topic/459-anatomy-of-a-mapreduce-job-run-with-hadoop/
To create the list of tasks to run, the job scheduler first retrieves the
input splits computed by the JobClient from the shared filesystem (step 6).
It then creates one map task for each split. The number of reduce tasks to
create is determined by the mapred.reduce.tasks property in the JobConf,
which is set by the setNumReduceTasks() method, and the scheduler simply
creates this number of reduce tasks to be run. Tasks are given IDs at this
point.
Best Regards,
Sonal
Nube Technologies <http://www.nubetech.co>
<http://in.linkedin.com/in/sonalgoyal>
On Fri, Sep 27, 2013 at 10:55 AM, Sai Sai <sa...@yahoo.in> wrote:
> Hi
> I have attached the anatomy of MR from definitive guide.
>
> In step 6 it says JT/Scheduler retrieve input splits computed by the
> client from hdfs.
>
> In the above line it refers to as the client computes input splits.
>
> 1. Why does the JT/Scheduler retrieve the input splits and what does it do.
> If it is retrieving the input split does this mean it goes to the block
> and reads each record
> and gets the record back to JT. If so this is a lot of data movement for
> large files.
> which is not data locality. so i m getting confused.
>
> 2. How does the client know how to calculate the input splits.
>
> Any help please.
> Thanks
> Sai
>
Re: Retrieve and compute input splits
Posted by Sonal Goyal <so...@gmail.com>.
The input splits are not copied, only the information on the location of
the splits is copied to the jobtracker so that it can assign tasktrackers
which are local to the split.
Check the Job Initialization section at
http://answers.oreilly.com/topic/459-anatomy-of-a-mapreduce-job-run-with-hadoop/
To create the list of tasks to run, the job scheduler first retrieves the
input splits computed by the JobClient from the shared filesystem (step 6).
It then creates one map task for each split. The number of reduce tasks to
create is determined by the mapred.reduce.tasks property in the JobConf,
which is set by the setNumReduceTasks() method, and the scheduler simply
creates this number of reduce tasks to be run. Tasks are given IDs at this
point.
Best Regards,
Sonal
Nube Technologies <http://www.nubetech.co>
<http://in.linkedin.com/in/sonalgoyal>
On Fri, Sep 27, 2013 at 10:55 AM, Sai Sai <sa...@yahoo.in> wrote:
> Hi
> I have attached the anatomy of MR from definitive guide.
>
> In step 6 it says JT/Scheduler retrieve input splits computed by the
> client from hdfs.
>
> In the above line it refers to as the client computes input splits.
>
> 1. Why does the JT/Scheduler retrieve the input splits and what does it do.
> If it is retrieving the input split does this mean it goes to the block
> and reads each record
> and gets the record back to JT. If so this is a lot of data movement for
> large files.
> which is not data locality. so i m getting confused.
>
> 2. How does the client know how to calculate the input splits.
>
> Any help please.
> Thanks
> Sai
>
Re: Retrieve and compute input splits
Posted by Sai Sai <sa...@yahoo.in>.
Hi
I have attached the anatomy of MR from definitive guide.
In step 6 it says JT/Scheduler retrieve input splits computed by the client from hdfs.
In the above line it refers to as the client computes input splits.
1. Why does the JT/Scheduler retrieve the input splits and what does it do.
If it is retrieving the input split does this mean it goes to the block and reads each record
and gets the record back to JT. If so this is a lot of data movement for large files.
which is not data locality. so i m getting confused.
2. How does the client know how to calculate the input splits.
Any help please.
Thanks
Sai
Re: 2 Map tasks running for a small input file
Posted by Viji R <vi...@cloudera.com>.
Hi,
Default number of map tasks is 2. You can set mapred.map.tasks to 1 to
avoid this.
Regards,
Viji
On Thu, Sep 26, 2013 at 4:28 PM, Sai Sai <sa...@yahoo.in> wrote:
> Hi
> Here is the input file for the wordcount job:
> ******************
> Hi This is a simple test.
> Hi Hadoop how r u.
> Hello Hello.
> Hi Hi.
> Hadoop Hadoop Welcome.
> ******************
>
> After running the wordcount successfully
> here r the counters info:
>
> ***************
> Job Counters SLOTS_MILLIS_MAPS 0 0 8,386
> Launched reduce tasks 0 0 1
> Total time spent by all reduces waiting after reserving slots (ms) 0 0 0
> Total time spent by all maps waiting after reserving slots (ms) 0 0 0
> Launched map tasks 0 0 2
> Data-local map tasks 0 0 2
> SLOTS_MILLIS_REDUCES 0 0 9,199
> ***************
> My question why r there 2 launched map tasks when i have only a small file.
> Per my understanding it is only 1 block.
> and should be only 1 split.
> Then for each line a map computation should occur
> but it shows 2 map tasks.
> Please let me know.
> Thanks
> Sai
>
Re: Retrieve and compute input splits
Posted by Sai Sai <sa...@yahoo.in>.
Hi
I have attached the anatomy of MR from definitive guide.
In step 6 it says JT/Scheduler retrieve input splits computed by the client from hdfs.
In the above line it refers to as the client computes input splits.
1. Why does the JT/Scheduler retrieve the input splits and what does it do.
If it is retrieving the input split does this mean it goes to the block and reads each record
and gets the record back to JT. If so this is a lot of data movement for large files.
which is not data locality. so i m getting confused.
2. How does the client know how to calculate the input splits.
Any help please.
Thanks
Sai
Re: Retrieve and compute input splits
Posted by Sai Sai <sa...@yahoo.in>.
Hi
I have attached the anatomy of MR from definitive guide.
In step 6 it says JT/Scheduler retrieve input splits computed by the client from hdfs.
In the above line it refers to as the client computes input splits.
1. Why does the JT/Scheduler retrieve the input splits and what does it do.
If it is retrieving the input split does this mean it goes to the block and reads each record
and gets the record back to JT. If so this is a lot of data movement for large files.
which is not data locality. so i m getting confused.
2. How does the client know how to calculate the input splits.
Any help please.
Thanks
Sai
Re: Retrieve and compute input splits
Posted by Sai Sai <sa...@yahoo.in>.
Hi
I have attached the anatomy of MR from definitive guide.
In step 6 it says JT/Scheduler retrieve input splits computed by the client from hdfs.
In the above line it refers to as the client computes input splits.
1. Why does the JT/Scheduler retrieve the input splits and what does it do.
If it is retrieving the input split does this mean it goes to the block and reads each record
and gets the record back to JT. If so this is a lot of data movement for large files.
which is not data locality. so i m getting confused.
2. How does the client know how to calculate the input splits.
Any help please.
Thanks
Sai