You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@accumulo.apache.org by Sean Busbey <bu...@cloudera.com> on 2014/11/19 19:23:44 UTC

Re:

the error from the tablet server claims to be looking at a local filesystem
path (the part with " file:/accumulo/instance_id" should start with hdfs).

Are your hadoop client configs in the classpath?

Also, the Accumulo PMC *strongly* recommends using Hadoop 2 for Accumulo
1.5+. Is there a particular reason you're using 1.x?

On Wed, Nov 19, 2014 at 5:03 AM, Salih Kardan <ka...@gmail.com> wrote:

> Hello everyone,
>
> Currently I am testing Accumulo 1.6.1 with Hadoop 1.2.1 and Zookeeper
> 3.4.5 with 4 hadoop nodes. I gave a role to each node in hadoop cluster;
> specifically
> assume my hadoop nodes like this: hadoop1, hadoop2, hadoop3 and hadoop4.
>
> The accumulo roles are distributed to hadoop nodes;
> hadoop1 = accumulo master ( single zookeeper instance is also running on
> this node)
> hadoop2 = gc
> hadoop3 = monitor
> hadoop4 = tablet server + tracer
>
> After I initialize accumulo with *"accumulo init"* command, I invoked
> start_all.sh script from master machine. All services except tracer and
> tablet server started working.
> However on tablet server machine I see these logs :
>
> Thread "tracer" died Accumulo not initialized, there is no instance id at
> file:/accumulo/instance_id
> java.lang.RuntimeException: Accumulo not initialized, there is no instance
> id at file:/accumulo/instance_id
>     at
> org.apache.accumulo.core.zookeeper.ZooUtil.getInstanceIDFromHdfs(ZooUtil.java:62)
>     at
> org.apache.accumulo.server.client.HdfsZooInstance._getInstanceID(HdfsZooInstance.java:132)
>     at
> org.apache.accumulo.server.client.HdfsZooInstance.getInstanceID(HdfsZooInstance.java:116)
>     at
> org.apache.accumulo.server.conf.ServerConfigurationFactory.<init>(ServerConfigurationFactory.java:113)
>     at
> org.apache.accumulo.server.conf.ServerConfiguration.<init>(ServerConfiguration.java:79)
>     at org.apache.accumulo.tracer.TraceServer.main(TraceServer.java:290)
>     at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
>     at
> sun.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:57)
>     at
> sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
>     at java.lang.reflect.Method.invoke(Method.java:606)
>     at org.apache.accumulo.start.Main$1.run(Main.java:141)
>     at java.lang.Thread.run(Thread.java:745)
>
> I checked hdfs and it seems accumulo is initialized. Here is the output of *"hadoop
> dfs -ls /accumulo/instance_id" *command*, *
>
> Found 1 items
> -rw-r--r--   1 root supergroup          0 2014-11-19 11:19
> /accumulo/instance_id/268acc40-e20b-4a35-8d8a-0e46e7859a0d
>
> I googled the problem, some comments stating that the problem may occur
> due to missing hadoop libs in classpath, but I checked classpath of
> with *"accumulo classpath"*  command, it also seems correct, both hadoop
> and zookeeper libs listed in classspath.
>
>
> Then I tried single node accumulo installation, that way all services
> including tablet server seems working.
> What can be the problem when I use multiple nodes? Any help is appreciated.
>
> Thanks.
>



-- 
Sean

Re:

Posted by Christopher <ct...@apache.org>.
You may want to look at the bin/bootstrap_config.sh script provided in
Accumulo 1.6.1. The default settings in a lot of Accumulo's example configs
for 1.6 is for Hadoop 2. The bootstrap_config.sh script will generate
config examples from a template, after asking a few questions about your
environment. You may want compare the configs generated from that with
those you are currently using.

--
Christopher L Tubbs II
http://gravatar.com/ctubbsii

On Wed, Nov 19, 2014 at 1:43 PM, Salih Kardan <ka...@gmail.com> wrote:

> Hi Sean,
>
> I have hadoop an environment and along with Hadoop I am using some other
> Apache products such Hive, Pig and HBase. Actually I don't have a specific
> reason to use Hadoop 1.x but changing hadoop version require me to change
> also other products' version. That's why just for now I avoid changing it
> and just wanted to test If I can upgrade Accumulo version that I am using
> right now due to some bugs in accumulo-1.4.2.
>
> Before I see your mail, I have just send accumulo-env.sh script and output
> of accumulo classpath command. Could u please take a look at it and how can
> I force tablet server look hdfs instead of local file system?
>
> Thanks.
>
> Salih Kardan
>
> On Wed, Nov 19, 2014 at 8:33 PM, Christopher <ct...@apache.org> wrote:
>
>> On Wed, Nov 19, 2014 at 1:23 PM, Sean Busbey <bu...@cloudera.com> wrote:
>>
>>> the error from the tablet server claims to be looking at a local
>>> filesystem path (the part with " file:/accumulo/instance_id" should
>>> start with hdfs).
>>>
>>> Are your hadoop client configs in the classpath?
>>>
>>> Also, the Accumulo PMC *strongly* recommends using Hadoop 2 for Accumulo
>>> 1.5+. Is there a particular reason you're using 1.x?
>>>
>>>
>> Just to clarify...
>>
>> The default for 1.5 is still Hadoop 1.x. It supports Hadoop 2.x, and I'd
>> personally agree with the recommendation to use Hadoop 2.x when possible,
>> even when running 1.5 (report any bugs), but the default build and testing
>> and release packaging environment for 1.5 is still Hadoop 1.x, so I don't
>> know that the it's "strongly" recommended. Still, probably a good idea.
>>
>> --
>> Christopher L Tubbs II
>> http://gravatar.com/ctubbsii
>>
>>
>>
>>
>>> On Wed, Nov 19, 2014 at 5:03 AM, Salih Kardan <ka...@gmail.com>
>>> wrote:
>>>
>>>> Hello everyone,
>>>>
>>>> Currently I am testing Accumulo 1.6.1 with Hadoop 1.2.1 and Zookeeper
>>>> 3.4.5 with 4 hadoop nodes. I gave a role to each node in hadoop cluster;
>>>> specifically
>>>> assume my hadoop nodes like this: hadoop1, hadoop2, hadoop3 and
>>>> hadoop4.
>>>>
>>>> The accumulo roles are distributed to hadoop nodes;
>>>> hadoop1 = accumulo master ( single zookeeper instance is also running
>>>> on this node)
>>>> hadoop2 = gc
>>>> hadoop3 = monitor
>>>> hadoop4 = tablet server + tracer
>>>>
>>>> After I initialize accumulo with *"accumulo init"* command, I invoked
>>>> start_all.sh script from master machine. All services except tracer and
>>>> tablet server started working.
>>>> However on tablet server machine I see these logs :
>>>>
>>>> Thread "tracer" died Accumulo not initialized, there is no instance id
>>>> at file:/accumulo/instance_id
>>>> java.lang.RuntimeException: Accumulo not initialized, there is no
>>>> instance id at file:/accumulo/instance_id
>>>>     at
>>>> org.apache.accumulo.core.zookeeper.ZooUtil.getInstanceIDFromHdfs(ZooUtil.java:62)
>>>>     at
>>>> org.apache.accumulo.server.client.HdfsZooInstance._getInstanceID(HdfsZooInstance.java:132)
>>>>     at
>>>> org.apache.accumulo.server.client.HdfsZooInstance.getInstanceID(HdfsZooInstance.java:116)
>>>>     at
>>>> org.apache.accumulo.server.conf.ServerConfigurationFactory.<init>(ServerConfigurationFactory.java:113)
>>>>     at
>>>> org.apache.accumulo.server.conf.ServerConfiguration.<init>(ServerConfiguration.java:79)
>>>>     at org.apache.accumulo.tracer.TraceServer.main(TraceServer.java:290)
>>>>     at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
>>>>     at
>>>> sun.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:57)
>>>>     at
>>>> sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
>>>>     at java.lang.reflect.Method.invoke(Method.java:606)
>>>>     at org.apache.accumulo.start.Main$1.run(Main.java:141)
>>>>     at java.lang.Thread.run(Thread.java:745)
>>>>
>>>> I checked hdfs and it seems accumulo is initialized. Here is the output
>>>> of *"hadoop dfs -ls /accumulo/instance_id" *command*, *
>>>>
>>>> Found 1 items
>>>> -rw-r--r--   1 root supergroup          0 2014-11-19 11:19
>>>> /accumulo/instance_id/268acc40-e20b-4a35-8d8a-0e46e7859a0d
>>>>
>>>> I googled the problem, some comments stating that the problem may occur
>>>> due to missing hadoop libs in classpath, but I checked classpath of
>>>> with *"accumulo classpath"*  command, it also seems correct, both
>>>> hadoop and zookeeper libs listed in classspath.
>>>>
>>>>
>>>> Then I tried single node accumulo installation, that way all services
>>>> including tablet server seems working.
>>>> What can be the problem when I use multiple nodes? Any help is
>>>> appreciated.
>>>>
>>>> Thanks.
>>>>
>>>
>>>
>>>
>>> --
>>> Sean
>>>
>>
>>
>

Re:

Posted by Salih Kardan <ka...@gmail.com>.
Hi Sean,

I have hadoop an environment and along with Hadoop I am using some other
Apache products such Hive, Pig and HBase. Actually I don't have a specific
reason to use Hadoop 1.x but changing hadoop version require me to change
also other products' version. That's why just for now I avoid changing it
and just wanted to test If I can upgrade Accumulo version that I am using
right now due to some bugs in accumulo-1.4.2.

Before I see your mail, I have just send accumulo-env.sh script and output
of accumulo classpath command. Could u please take a look at it and how can
I force tablet server look hdfs instead of local file system?

Thanks.

Salih Kardan

On Wed, Nov 19, 2014 at 8:33 PM, Christopher <ct...@apache.org> wrote:

> On Wed, Nov 19, 2014 at 1:23 PM, Sean Busbey <bu...@cloudera.com> wrote:
>
>> the error from the tablet server claims to be looking at a local
>> filesystem path (the part with " file:/accumulo/instance_id" should
>> start with hdfs).
>>
>> Are your hadoop client configs in the classpath?
>>
>> Also, the Accumulo PMC *strongly* recommends using Hadoop 2 for Accumulo
>> 1.5+. Is there a particular reason you're using 1.x?
>>
>>
> Just to clarify...
>
> The default for 1.5 is still Hadoop 1.x. It supports Hadoop 2.x, and I'd
> personally agree with the recommendation to use Hadoop 2.x when possible,
> even when running 1.5 (report any bugs), but the default build and testing
> and release packaging environment for 1.5 is still Hadoop 1.x, so I don't
> know that the it's "strongly" recommended. Still, probably a good idea.
>
> --
> Christopher L Tubbs II
> http://gravatar.com/ctubbsii
>
>
>
>
>> On Wed, Nov 19, 2014 at 5:03 AM, Salih Kardan <ka...@gmail.com> wrote:
>>
>>> Hello everyone,
>>>
>>> Currently I am testing Accumulo 1.6.1 with Hadoop 1.2.1 and Zookeeper
>>> 3.4.5 with 4 hadoop nodes. I gave a role to each node in hadoop cluster;
>>> specifically
>>> assume my hadoop nodes like this: hadoop1, hadoop2, hadoop3 and hadoop4.
>>>
>>> The accumulo roles are distributed to hadoop nodes;
>>> hadoop1 = accumulo master ( single zookeeper instance is also running on
>>> this node)
>>> hadoop2 = gc
>>> hadoop3 = monitor
>>> hadoop4 = tablet server + tracer
>>>
>>> After I initialize accumulo with *"accumulo init"* command, I invoked
>>> start_all.sh script from master machine. All services except tracer and
>>> tablet server started working.
>>> However on tablet server machine I see these logs :
>>>
>>> Thread "tracer" died Accumulo not initialized, there is no instance id
>>> at file:/accumulo/instance_id
>>> java.lang.RuntimeException: Accumulo not initialized, there is no
>>> instance id at file:/accumulo/instance_id
>>>     at
>>> org.apache.accumulo.core.zookeeper.ZooUtil.getInstanceIDFromHdfs(ZooUtil.java:62)
>>>     at
>>> org.apache.accumulo.server.client.HdfsZooInstance._getInstanceID(HdfsZooInstance.java:132)
>>>     at
>>> org.apache.accumulo.server.client.HdfsZooInstance.getInstanceID(HdfsZooInstance.java:116)
>>>     at
>>> org.apache.accumulo.server.conf.ServerConfigurationFactory.<init>(ServerConfigurationFactory.java:113)
>>>     at
>>> org.apache.accumulo.server.conf.ServerConfiguration.<init>(ServerConfiguration.java:79)
>>>     at org.apache.accumulo.tracer.TraceServer.main(TraceServer.java:290)
>>>     at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
>>>     at
>>> sun.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:57)
>>>     at
>>> sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
>>>     at java.lang.reflect.Method.invoke(Method.java:606)
>>>     at org.apache.accumulo.start.Main$1.run(Main.java:141)
>>>     at java.lang.Thread.run(Thread.java:745)
>>>
>>> I checked hdfs and it seems accumulo is initialized. Here is the output
>>> of *"hadoop dfs -ls /accumulo/instance_id" *command*, *
>>>
>>> Found 1 items
>>> -rw-r--r--   1 root supergroup          0 2014-11-19 11:19
>>> /accumulo/instance_id/268acc40-e20b-4a35-8d8a-0e46e7859a0d
>>>
>>> I googled the problem, some comments stating that the problem may occur
>>> due to missing hadoop libs in classpath, but I checked classpath of
>>> with *"accumulo classpath"*  command, it also seems correct, both
>>> hadoop and zookeeper libs listed in classspath.
>>>
>>>
>>> Then I tried single node accumulo installation, that way all services
>>> including tablet server seems working.
>>> What can be the problem when I use multiple nodes? Any help is
>>> appreciated.
>>>
>>> Thanks.
>>>
>>
>>
>>
>> --
>> Sean
>>
>
>

Re:

Posted by Christopher <ct...@apache.org>.
On Wed, Nov 19, 2014 at 1:23 PM, Sean Busbey <bu...@cloudera.com> wrote:

> the error from the tablet server claims to be looking at a local
> filesystem path (the part with " file:/accumulo/instance_id" should start
> with hdfs).
>
> Are your hadoop client configs in the classpath?
>
> Also, the Accumulo PMC *strongly* recommends using Hadoop 2 for Accumulo
> 1.5+. Is there a particular reason you're using 1.x?
>
>
Just to clarify...

The default for 1.5 is still Hadoop 1.x. It supports Hadoop 2.x, and I'd
personally agree with the recommendation to use Hadoop 2.x when possible,
even when running 1.5 (report any bugs), but the default build and testing
and release packaging environment for 1.5 is still Hadoop 1.x, so I don't
know that the it's "strongly" recommended. Still, probably a good idea.

--
Christopher L Tubbs II
http://gravatar.com/ctubbsii




> On Wed, Nov 19, 2014 at 5:03 AM, Salih Kardan <ka...@gmail.com> wrote:
>
>> Hello everyone,
>>
>> Currently I am testing Accumulo 1.6.1 with Hadoop 1.2.1 and Zookeeper
>> 3.4.5 with 4 hadoop nodes. I gave a role to each node in hadoop cluster;
>> specifically
>> assume my hadoop nodes like this: hadoop1, hadoop2, hadoop3 and hadoop4.
>>
>> The accumulo roles are distributed to hadoop nodes;
>> hadoop1 = accumulo master ( single zookeeper instance is also running on
>> this node)
>> hadoop2 = gc
>> hadoop3 = monitor
>> hadoop4 = tablet server + tracer
>>
>> After I initialize accumulo with *"accumulo init"* command, I invoked
>> start_all.sh script from master machine. All services except tracer and
>> tablet server started working.
>> However on tablet server machine I see these logs :
>>
>> Thread "tracer" died Accumulo not initialized, there is no instance id at
>> file:/accumulo/instance_id
>> java.lang.RuntimeException: Accumulo not initialized, there is no
>> instance id at file:/accumulo/instance_id
>>     at
>> org.apache.accumulo.core.zookeeper.ZooUtil.getInstanceIDFromHdfs(ZooUtil.java:62)
>>     at
>> org.apache.accumulo.server.client.HdfsZooInstance._getInstanceID(HdfsZooInstance.java:132)
>>     at
>> org.apache.accumulo.server.client.HdfsZooInstance.getInstanceID(HdfsZooInstance.java:116)
>>     at
>> org.apache.accumulo.server.conf.ServerConfigurationFactory.<init>(ServerConfigurationFactory.java:113)
>>     at
>> org.apache.accumulo.server.conf.ServerConfiguration.<init>(ServerConfiguration.java:79)
>>     at org.apache.accumulo.tracer.TraceServer.main(TraceServer.java:290)
>>     at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
>>     at
>> sun.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:57)
>>     at
>> sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
>>     at java.lang.reflect.Method.invoke(Method.java:606)
>>     at org.apache.accumulo.start.Main$1.run(Main.java:141)
>>     at java.lang.Thread.run(Thread.java:745)
>>
>> I checked hdfs and it seems accumulo is initialized. Here is the output
>> of *"hadoop dfs -ls /accumulo/instance_id" *command*, *
>>
>> Found 1 items
>> -rw-r--r--   1 root supergroup          0 2014-11-19 11:19
>> /accumulo/instance_id/268acc40-e20b-4a35-8d8a-0e46e7859a0d
>>
>> I googled the problem, some comments stating that the problem may occur
>> due to missing hadoop libs in classpath, but I checked classpath of
>> with *"accumulo classpath"*  command, it also seems correct, both hadoop
>> and zookeeper libs listed in classspath.
>>
>>
>> Then I tried single node accumulo installation, that way all services
>> including tablet server seems working.
>> What can be the problem when I use multiple nodes? Any help is
>> appreciated.
>>
>> Thanks.
>>
>
>
>
> --
> Sean
>