You are viewing a plain text version of this content. The canonical link for it is here.
Posted to common-user@hadoop.apache.org by Ian Jackson <Ia...@trilliumsoftware.com> on 2013/11/22 23:04:58 UTC
Heterogeneous Cluster
Has anyone set up a Heterogeneous cluster, some Windows nodes and Linux nodes?
Re: Heterogeneous Cluster
Posted by Azuryy Yu <az...@gmail.com>.
RHEL and CentOS are not cross, which are all Linux kernel.
On 2013-11-29 6:39 AM, "Adam Kawa" <ka...@gmail.com> wrote:
> > "we can deploy cluster cross IDC, cross different network, but don't
> cross OS."
>
> I heard that some companies deployed Hadoop cluster, where RHEL is
> installed on masters (to get a support), and CentOs is installed on slaves.
> If it is true, then it shows that we can can deploy Hadoop cluster across
> OSes (at least for Linux).
>
>
> 2013/11/26 Azuryy Yu <az...@gmail.com>
>
>> I don't think this is a normal way, and It's not suggested. we can
>> deploy cluster cross IDC, cross different network, but don't cross OS.
>>
>> at least currently.
>>
>>
>> On Tue, Nov 26, 2013 at 6:56 AM, Andrew Machtolff <am...@askcts.com>wrote:
>>
>>> Yes, I set one up as a test. I had a windows cluster of 3 machines,
>>> and added a 4th Linux node. The Data Node was able to connect and
>>> replicate, but MR jobs failed. JobTracker/TaskTracker wasn’t translating
>>> the path to the data block. They were telling the Linux node to look in
>>> C:\.... for the data block, and that obviously didn’t work.
>>>
>>>
>>>
>>> I feel like it’s only a small change away from working, at that point, I
>>> had to shift my focus and couldn’t work on it any more.
>>>
>>>
>>>
>>> Have you had any success?
>>>
>>>
>>>
>>> Andrew
>>>
>>>
>>>
>>> *________________________________*
>>>
>>> Andrew Machtolff / Senior Consultant
>>>
>>>
>>>
>>> [image: http://images.askcts.com/images/cts_logo_email.png]<http://www.askcts.com/>
>>> www.askcts.com
>>> amachtolff@askcts.com
>>> [image: http://images.askcts.com/images/cts_logo_linkedin.png]<http://www.linkedin.com/company/cts-inc>[image:
>>> http://images.askcts.com/images/cts_logo_twitter.png]<https://twitter.com/askCTS>[image:
>>> http://images.askcts.com/images/cts_logo_facebook.png]<https://www.facebook.com/askCTS>
>>>
>>>
>>>
>>> *From:* Ian Jackson [mailto:Ian_Jackson@trilliumsoftware.com]
>>> *Sent:* Friday, November 22, 2013 4:05 PM
>>> *To:* user@hadoop.apache.org
>>> *Subject:* Heterogeneous Cluster
>>>
>>>
>>>
>>> Has anyone set up a Heterogeneous cluster, some Windows nodes and Linux
>>> nodes?
>>>
>>
>>
>
Re: Heterogeneous Cluster
Posted by Azuryy Yu <az...@gmail.com>.
RHEL and CentOS are not cross, which are all Linux kernel.
On 2013-11-29 6:39 AM, "Adam Kawa" <ka...@gmail.com> wrote:
> > "we can deploy cluster cross IDC, cross different network, but don't
> cross OS."
>
> I heard that some companies deployed Hadoop cluster, where RHEL is
> installed on masters (to get a support), and CentOs is installed on slaves.
> If it is true, then it shows that we can can deploy Hadoop cluster across
> OSes (at least for Linux).
>
>
> 2013/11/26 Azuryy Yu <az...@gmail.com>
>
>> I don't think this is a normal way, and It's not suggested. we can
>> deploy cluster cross IDC, cross different network, but don't cross OS.
>>
>> at least currently.
>>
>>
>> On Tue, Nov 26, 2013 at 6:56 AM, Andrew Machtolff <am...@askcts.com>wrote:
>>
>>> Yes, I set one up as a test. I had a windows cluster of 3 machines,
>>> and added a 4th Linux node. The Data Node was able to connect and
>>> replicate, but MR jobs failed. JobTracker/TaskTracker wasn’t translating
>>> the path to the data block. They were telling the Linux node to look in
>>> C:\.... for the data block, and that obviously didn’t work.
>>>
>>>
>>>
>>> I feel like it’s only a small change away from working, at that point, I
>>> had to shift my focus and couldn’t work on it any more.
>>>
>>>
>>>
>>> Have you had any success?
>>>
>>>
>>>
>>> Andrew
>>>
>>>
>>>
>>> *________________________________*
>>>
>>> Andrew Machtolff / Senior Consultant
>>>
>>>
>>>
>>> [image: http://images.askcts.com/images/cts_logo_email.png]<http://www.askcts.com/>
>>> www.askcts.com
>>> amachtolff@askcts.com
>>> [image: http://images.askcts.com/images/cts_logo_linkedin.png]<http://www.linkedin.com/company/cts-inc>[image:
>>> http://images.askcts.com/images/cts_logo_twitter.png]<https://twitter.com/askCTS>[image:
>>> http://images.askcts.com/images/cts_logo_facebook.png]<https://www.facebook.com/askCTS>
>>>
>>>
>>>
>>> *From:* Ian Jackson [mailto:Ian_Jackson@trilliumsoftware.com]
>>> *Sent:* Friday, November 22, 2013 4:05 PM
>>> *To:* user@hadoop.apache.org
>>> *Subject:* Heterogeneous Cluster
>>>
>>>
>>>
>>> Has anyone set up a Heterogeneous cluster, some Windows nodes and Linux
>>> nodes?
>>>
>>
>>
>
Re: Heterogeneous Cluster
Posted by Azuryy Yu <az...@gmail.com>.
RHEL and CentOS are not cross, which are all Linux kernel.
On 2013-11-29 6:39 AM, "Adam Kawa" <ka...@gmail.com> wrote:
> > "we can deploy cluster cross IDC, cross different network, but don't
> cross OS."
>
> I heard that some companies deployed Hadoop cluster, where RHEL is
> installed on masters (to get a support), and CentOs is installed on slaves.
> If it is true, then it shows that we can can deploy Hadoop cluster across
> OSes (at least for Linux).
>
>
> 2013/11/26 Azuryy Yu <az...@gmail.com>
>
>> I don't think this is a normal way, and It's not suggested. we can
>> deploy cluster cross IDC, cross different network, but don't cross OS.
>>
>> at least currently.
>>
>>
>> On Tue, Nov 26, 2013 at 6:56 AM, Andrew Machtolff <am...@askcts.com>wrote:
>>
>>> Yes, I set one up as a test. I had a windows cluster of 3 machines,
>>> and added a 4th Linux node. The Data Node was able to connect and
>>> replicate, but MR jobs failed. JobTracker/TaskTracker wasn’t translating
>>> the path to the data block. They were telling the Linux node to look in
>>> C:\.... for the data block, and that obviously didn’t work.
>>>
>>>
>>>
>>> I feel like it’s only a small change away from working, at that point, I
>>> had to shift my focus and couldn’t work on it any more.
>>>
>>>
>>>
>>> Have you had any success?
>>>
>>>
>>>
>>> Andrew
>>>
>>>
>>>
>>> *________________________________*
>>>
>>> Andrew Machtolff / Senior Consultant
>>>
>>>
>>>
>>> [image: http://images.askcts.com/images/cts_logo_email.png]<http://www.askcts.com/>
>>> www.askcts.com
>>> amachtolff@askcts.com
>>> [image: http://images.askcts.com/images/cts_logo_linkedin.png]<http://www.linkedin.com/company/cts-inc>[image:
>>> http://images.askcts.com/images/cts_logo_twitter.png]<https://twitter.com/askCTS>[image:
>>> http://images.askcts.com/images/cts_logo_facebook.png]<https://www.facebook.com/askCTS>
>>>
>>>
>>>
>>> *From:* Ian Jackson [mailto:Ian_Jackson@trilliumsoftware.com]
>>> *Sent:* Friday, November 22, 2013 4:05 PM
>>> *To:* user@hadoop.apache.org
>>> *Subject:* Heterogeneous Cluster
>>>
>>>
>>>
>>> Has anyone set up a Heterogeneous cluster, some Windows nodes and Linux
>>> nodes?
>>>
>>
>>
>
Re: Heterogeneous Cluster
Posted by Azuryy Yu <az...@gmail.com>.
RHEL and CentOS are not cross, which are all Linux kernel.
On 2013-11-29 6:39 AM, "Adam Kawa" <ka...@gmail.com> wrote:
> > "we can deploy cluster cross IDC, cross different network, but don't
> cross OS."
>
> I heard that some companies deployed Hadoop cluster, where RHEL is
> installed on masters (to get a support), and CentOs is installed on slaves.
> If it is true, then it shows that we can can deploy Hadoop cluster across
> OSes (at least for Linux).
>
>
> 2013/11/26 Azuryy Yu <az...@gmail.com>
>
>> I don't think this is a normal way, and It's not suggested. we can
>> deploy cluster cross IDC, cross different network, but don't cross OS.
>>
>> at least currently.
>>
>>
>> On Tue, Nov 26, 2013 at 6:56 AM, Andrew Machtolff <am...@askcts.com>wrote:
>>
>>> Yes, I set one up as a test. I had a windows cluster of 3 machines,
>>> and added a 4th Linux node. The Data Node was able to connect and
>>> replicate, but MR jobs failed. JobTracker/TaskTracker wasn’t translating
>>> the path to the data block. They were telling the Linux node to look in
>>> C:\.... for the data block, and that obviously didn’t work.
>>>
>>>
>>>
>>> I feel like it’s only a small change away from working, at that point, I
>>> had to shift my focus and couldn’t work on it any more.
>>>
>>>
>>>
>>> Have you had any success?
>>>
>>>
>>>
>>> Andrew
>>>
>>>
>>>
>>> *________________________________*
>>>
>>> Andrew Machtolff / Senior Consultant
>>>
>>>
>>>
>>> [image: http://images.askcts.com/images/cts_logo_email.png]<http://www.askcts.com/>
>>> www.askcts.com
>>> amachtolff@askcts.com
>>> [image: http://images.askcts.com/images/cts_logo_linkedin.png]<http://www.linkedin.com/company/cts-inc>[image:
>>> http://images.askcts.com/images/cts_logo_twitter.png]<https://twitter.com/askCTS>[image:
>>> http://images.askcts.com/images/cts_logo_facebook.png]<https://www.facebook.com/askCTS>
>>>
>>>
>>>
>>> *From:* Ian Jackson [mailto:Ian_Jackson@trilliumsoftware.com]
>>> *Sent:* Friday, November 22, 2013 4:05 PM
>>> *To:* user@hadoop.apache.org
>>> *Subject:* Heterogeneous Cluster
>>>
>>>
>>>
>>> Has anyone set up a Heterogeneous cluster, some Windows nodes and Linux
>>> nodes?
>>>
>>
>>
>
Re: Heterogeneous Cluster
Posted by Adam Kawa <ka...@gmail.com>.
> "we can deploy cluster cross IDC, cross different network, but don't
cross OS."
I heard that some companies deployed Hadoop cluster, where RHEL is
installed on masters (to get a support), and CentOs is installed on slaves.
If it is true, then it shows that we can can deploy Hadoop cluster across
OSes (at least for Linux).
2013/11/26 Azuryy Yu <az...@gmail.com>
> I don't think this is a normal way, and It's not suggested. we can deploy
> cluster cross IDC, cross different network, but don't cross OS.
>
> at least currently.
>
>
> On Tue, Nov 26, 2013 at 6:56 AM, Andrew Machtolff <am...@askcts.com>wrote:
>
>> Yes, I set one up as a test. I had a windows cluster of 3 machines, and
>> added a 4th Linux node. The Data Node was able to connect and replicate,
>> but MR jobs failed. JobTracker/TaskTracker wasn’t translating the path to
>> the data block. They were telling the Linux node to look in C:\.... for the
>> data block, and that obviously didn’t work.
>>
>>
>>
>> I feel like it’s only a small change away from working, at that point, I
>> had to shift my focus and couldn’t work on it any more.
>>
>>
>>
>> Have you had any success?
>>
>>
>>
>> Andrew
>>
>>
>>
>> *________________________________*
>>
>> Andrew Machtolff / Senior Consultant
>>
>>
>>
>> [image: http://images.askcts.com/images/cts_logo_email.png]<http://www.askcts.com/>
>> www.askcts.com
>> amachtolff@askcts.com
>> [image: http://images.askcts.com/images/cts_logo_linkedin.png]<http://www.linkedin.com/company/cts-inc>[image:
>> http://images.askcts.com/images/cts_logo_twitter.png]<https://twitter.com/askCTS>[image:
>> http://images.askcts.com/images/cts_logo_facebook.png]<https://www.facebook.com/askCTS>
>>
>>
>>
>> *From:* Ian Jackson [mailto:Ian_Jackson@trilliumsoftware.com]
>> *Sent:* Friday, November 22, 2013 4:05 PM
>> *To:* user@hadoop.apache.org
>> *Subject:* Heterogeneous Cluster
>>
>>
>>
>> Has anyone set up a Heterogeneous cluster, some Windows nodes and Linux
>> nodes?
>>
>
>
Re: Heterogeneous Cluster
Posted by Adam Kawa <ka...@gmail.com>.
> "we can deploy cluster cross IDC, cross different network, but don't
cross OS."
I heard that some companies deployed Hadoop cluster, where RHEL is
installed on masters (to get a support), and CentOs is installed on slaves.
If it is true, then it shows that we can can deploy Hadoop cluster across
OSes (at least for Linux).
2013/11/26 Azuryy Yu <az...@gmail.com>
> I don't think this is a normal way, and It's not suggested. we can deploy
> cluster cross IDC, cross different network, but don't cross OS.
>
> at least currently.
>
>
> On Tue, Nov 26, 2013 at 6:56 AM, Andrew Machtolff <am...@askcts.com>wrote:
>
>> Yes, I set one up as a test. I had a windows cluster of 3 machines, and
>> added a 4th Linux node. The Data Node was able to connect and replicate,
>> but MR jobs failed. JobTracker/TaskTracker wasn’t translating the path to
>> the data block. They were telling the Linux node to look in C:\.... for the
>> data block, and that obviously didn’t work.
>>
>>
>>
>> I feel like it’s only a small change away from working, at that point, I
>> had to shift my focus and couldn’t work on it any more.
>>
>>
>>
>> Have you had any success?
>>
>>
>>
>> Andrew
>>
>>
>>
>> *________________________________*
>>
>> Andrew Machtolff / Senior Consultant
>>
>>
>>
>> [image: http://images.askcts.com/images/cts_logo_email.png]<http://www.askcts.com/>
>> www.askcts.com
>> amachtolff@askcts.com
>> [image: http://images.askcts.com/images/cts_logo_linkedin.png]<http://www.linkedin.com/company/cts-inc>[image:
>> http://images.askcts.com/images/cts_logo_twitter.png]<https://twitter.com/askCTS>[image:
>> http://images.askcts.com/images/cts_logo_facebook.png]<https://www.facebook.com/askCTS>
>>
>>
>>
>> *From:* Ian Jackson [mailto:Ian_Jackson@trilliumsoftware.com]
>> *Sent:* Friday, November 22, 2013 4:05 PM
>> *To:* user@hadoop.apache.org
>> *Subject:* Heterogeneous Cluster
>>
>>
>>
>> Has anyone set up a Heterogeneous cluster, some Windows nodes and Linux
>> nodes?
>>
>
>
Re: Heterogeneous Cluster
Posted by Adam Kawa <ka...@gmail.com>.
> "we can deploy cluster cross IDC, cross different network, but don't
cross OS."
I heard that some companies deployed Hadoop cluster, where RHEL is
installed on masters (to get a support), and CentOs is installed on slaves.
If it is true, then it shows that we can can deploy Hadoop cluster across
OSes (at least for Linux).
2013/11/26 Azuryy Yu <az...@gmail.com>
> I don't think this is a normal way, and It's not suggested. we can deploy
> cluster cross IDC, cross different network, but don't cross OS.
>
> at least currently.
>
>
> On Tue, Nov 26, 2013 at 6:56 AM, Andrew Machtolff <am...@askcts.com>wrote:
>
>> Yes, I set one up as a test. I had a windows cluster of 3 machines, and
>> added a 4th Linux node. The Data Node was able to connect and replicate,
>> but MR jobs failed. JobTracker/TaskTracker wasn’t translating the path to
>> the data block. They were telling the Linux node to look in C:\.... for the
>> data block, and that obviously didn’t work.
>>
>>
>>
>> I feel like it’s only a small change away from working, at that point, I
>> had to shift my focus and couldn’t work on it any more.
>>
>>
>>
>> Have you had any success?
>>
>>
>>
>> Andrew
>>
>>
>>
>> *________________________________*
>>
>> Andrew Machtolff / Senior Consultant
>>
>>
>>
>> [image: http://images.askcts.com/images/cts_logo_email.png]<http://www.askcts.com/>
>> www.askcts.com
>> amachtolff@askcts.com
>> [image: http://images.askcts.com/images/cts_logo_linkedin.png]<http://www.linkedin.com/company/cts-inc>[image:
>> http://images.askcts.com/images/cts_logo_twitter.png]<https://twitter.com/askCTS>[image:
>> http://images.askcts.com/images/cts_logo_facebook.png]<https://www.facebook.com/askCTS>
>>
>>
>>
>> *From:* Ian Jackson [mailto:Ian_Jackson@trilliumsoftware.com]
>> *Sent:* Friday, November 22, 2013 4:05 PM
>> *To:* user@hadoop.apache.org
>> *Subject:* Heterogeneous Cluster
>>
>>
>>
>> Has anyone set up a Heterogeneous cluster, some Windows nodes and Linux
>> nodes?
>>
>
>
Re: Heterogeneous Cluster
Posted by Adam Kawa <ka...@gmail.com>.
> "we can deploy cluster cross IDC, cross different network, but don't
cross OS."
I heard that some companies deployed Hadoop cluster, where RHEL is
installed on masters (to get a support), and CentOs is installed on slaves.
If it is true, then it shows that we can can deploy Hadoop cluster across
OSes (at least for Linux).
2013/11/26 Azuryy Yu <az...@gmail.com>
> I don't think this is a normal way, and It's not suggested. we can deploy
> cluster cross IDC, cross different network, but don't cross OS.
>
> at least currently.
>
>
> On Tue, Nov 26, 2013 at 6:56 AM, Andrew Machtolff <am...@askcts.com>wrote:
>
>> Yes, I set one up as a test. I had a windows cluster of 3 machines, and
>> added a 4th Linux node. The Data Node was able to connect and replicate,
>> but MR jobs failed. JobTracker/TaskTracker wasn’t translating the path to
>> the data block. They were telling the Linux node to look in C:\.... for the
>> data block, and that obviously didn’t work.
>>
>>
>>
>> I feel like it’s only a small change away from working, at that point, I
>> had to shift my focus and couldn’t work on it any more.
>>
>>
>>
>> Have you had any success?
>>
>>
>>
>> Andrew
>>
>>
>>
>> *________________________________*
>>
>> Andrew Machtolff / Senior Consultant
>>
>>
>>
>> [image: http://images.askcts.com/images/cts_logo_email.png]<http://www.askcts.com/>
>> www.askcts.com
>> amachtolff@askcts.com
>> [image: http://images.askcts.com/images/cts_logo_linkedin.png]<http://www.linkedin.com/company/cts-inc>[image:
>> http://images.askcts.com/images/cts_logo_twitter.png]<https://twitter.com/askCTS>[image:
>> http://images.askcts.com/images/cts_logo_facebook.png]<https://www.facebook.com/askCTS>
>>
>>
>>
>> *From:* Ian Jackson [mailto:Ian_Jackson@trilliumsoftware.com]
>> *Sent:* Friday, November 22, 2013 4:05 PM
>> *To:* user@hadoop.apache.org
>> *Subject:* Heterogeneous Cluster
>>
>>
>>
>> Has anyone set up a Heterogeneous cluster, some Windows nodes and Linux
>> nodes?
>>
>
>
Re: Heterogeneous Cluster
Posted by Azuryy Yu <az...@gmail.com>.
I don't think this is a normal way, and It's not suggested. we can deploy
cluster cross IDC, cross different network, but don't cross OS.
at least currently.
On Tue, Nov 26, 2013 at 6:56 AM, Andrew Machtolff <am...@askcts.com>wrote:
> Yes, I set one up as a test. I had a windows cluster of 3 machines, and
> added a 4th Linux node. The Data Node was able to connect and replicate,
> but MR jobs failed. JobTracker/TaskTracker wasn’t translating the path to
> the data block. They were telling the Linux node to look in C:\.... for the
> data block, and that obviously didn’t work.
>
>
>
> I feel like it’s only a small change away from working, at that point, I
> had to shift my focus and couldn’t work on it any more.
>
>
>
> Have you had any success?
>
>
>
> Andrew
>
>
>
> *________________________________*
>
> Andrew Machtolff / Senior Consultant
>
>
>
> [image: http://images.askcts.com/images/cts_logo_email.png]<http://www.askcts.com/>
> www.askcts.com
> amachtolff@askcts.com
> [image: http://images.askcts.com/images/cts_logo_linkedin.png]<http://www.linkedin.com/company/cts-inc>[image:
> http://images.askcts.com/images/cts_logo_twitter.png]<https://twitter.com/askCTS>[image:
> http://images.askcts.com/images/cts_logo_facebook.png]<https://www.facebook.com/askCTS>
>
>
>
> *From:* Ian Jackson [mailto:Ian_Jackson@trilliumsoftware.com]
> *Sent:* Friday, November 22, 2013 4:05 PM
> *To:* user@hadoop.apache.org
> *Subject:* Heterogeneous Cluster
>
>
>
> Has anyone set up a Heterogeneous cluster, some Windows nodes and Linux
> nodes?
>
Re: Heterogeneous Cluster
Posted by Azuryy Yu <az...@gmail.com>.
I don't think this is a normal way, and It's not suggested. we can deploy
cluster cross IDC, cross different network, but don't cross OS.
at least currently.
On Tue, Nov 26, 2013 at 6:56 AM, Andrew Machtolff <am...@askcts.com>wrote:
> Yes, I set one up as a test. I had a windows cluster of 3 machines, and
> added a 4th Linux node. The Data Node was able to connect and replicate,
> but MR jobs failed. JobTracker/TaskTracker wasn’t translating the path to
> the data block. They were telling the Linux node to look in C:\.... for the
> data block, and that obviously didn’t work.
>
>
>
> I feel like it’s only a small change away from working, at that point, I
> had to shift my focus and couldn’t work on it any more.
>
>
>
> Have you had any success?
>
>
>
> Andrew
>
>
>
> *________________________________*
>
> Andrew Machtolff / Senior Consultant
>
>
>
> [image: http://images.askcts.com/images/cts_logo_email.png]<http://www.askcts.com/>
> www.askcts.com
> amachtolff@askcts.com
> [image: http://images.askcts.com/images/cts_logo_linkedin.png]<http://www.linkedin.com/company/cts-inc>[image:
> http://images.askcts.com/images/cts_logo_twitter.png]<https://twitter.com/askCTS>[image:
> http://images.askcts.com/images/cts_logo_facebook.png]<https://www.facebook.com/askCTS>
>
>
>
> *From:* Ian Jackson [mailto:Ian_Jackson@trilliumsoftware.com]
> *Sent:* Friday, November 22, 2013 4:05 PM
> *To:* user@hadoop.apache.org
> *Subject:* Heterogeneous Cluster
>
>
>
> Has anyone set up a Heterogeneous cluster, some Windows nodes and Linux
> nodes?
>
Re: Heterogeneous Cluster
Posted by Azuryy Yu <az...@gmail.com>.
I don't think this is a normal way, and It's not suggested. we can deploy
cluster cross IDC, cross different network, but don't cross OS.
at least currently.
On Tue, Nov 26, 2013 at 6:56 AM, Andrew Machtolff <am...@askcts.com>wrote:
> Yes, I set one up as a test. I had a windows cluster of 3 machines, and
> added a 4th Linux node. The Data Node was able to connect and replicate,
> but MR jobs failed. JobTracker/TaskTracker wasn’t translating the path to
> the data block. They were telling the Linux node to look in C:\.... for the
> data block, and that obviously didn’t work.
>
>
>
> I feel like it’s only a small change away from working, at that point, I
> had to shift my focus and couldn’t work on it any more.
>
>
>
> Have you had any success?
>
>
>
> Andrew
>
>
>
> *________________________________*
>
> Andrew Machtolff / Senior Consultant
>
>
>
> [image: http://images.askcts.com/images/cts_logo_email.png]<http://www.askcts.com/>
> www.askcts.com
> amachtolff@askcts.com
> [image: http://images.askcts.com/images/cts_logo_linkedin.png]<http://www.linkedin.com/company/cts-inc>[image:
> http://images.askcts.com/images/cts_logo_twitter.png]<https://twitter.com/askCTS>[image:
> http://images.askcts.com/images/cts_logo_facebook.png]<https://www.facebook.com/askCTS>
>
>
>
> *From:* Ian Jackson [mailto:Ian_Jackson@trilliumsoftware.com]
> *Sent:* Friday, November 22, 2013 4:05 PM
> *To:* user@hadoop.apache.org
> *Subject:* Heterogeneous Cluster
>
>
>
> Has anyone set up a Heterogeneous cluster, some Windows nodes and Linux
> nodes?
>
Re: Heterogeneous Cluster
Posted by Azuryy Yu <az...@gmail.com>.
I don't think this is a normal way, and It's not suggested. we can deploy
cluster cross IDC, cross different network, but don't cross OS.
at least currently.
On Tue, Nov 26, 2013 at 6:56 AM, Andrew Machtolff <am...@askcts.com>wrote:
> Yes, I set one up as a test. I had a windows cluster of 3 machines, and
> added a 4th Linux node. The Data Node was able to connect and replicate,
> but MR jobs failed. JobTracker/TaskTracker wasn’t translating the path to
> the data block. They were telling the Linux node to look in C:\.... for the
> data block, and that obviously didn’t work.
>
>
>
> I feel like it’s only a small change away from working, at that point, I
> had to shift my focus and couldn’t work on it any more.
>
>
>
> Have you had any success?
>
>
>
> Andrew
>
>
>
> *________________________________*
>
> Andrew Machtolff / Senior Consultant
>
>
>
> [image: http://images.askcts.com/images/cts_logo_email.png]<http://www.askcts.com/>
> www.askcts.com
> amachtolff@askcts.com
> [image: http://images.askcts.com/images/cts_logo_linkedin.png]<http://www.linkedin.com/company/cts-inc>[image:
> http://images.askcts.com/images/cts_logo_twitter.png]<https://twitter.com/askCTS>[image:
> http://images.askcts.com/images/cts_logo_facebook.png]<https://www.facebook.com/askCTS>
>
>
>
> *From:* Ian Jackson [mailto:Ian_Jackson@trilliumsoftware.com]
> *Sent:* Friday, November 22, 2013 4:05 PM
> *To:* user@hadoop.apache.org
> *Subject:* Heterogeneous Cluster
>
>
>
> Has anyone set up a Heterogeneous cluster, some Windows nodes and Linux
> nodes?
>
RE: Heterogeneous Cluster
Posted by Andrew Machtolff <am...@askcts.com>.
Yes, I set one up as a test. I had a windows cluster of 3 machines, and added a 4th Linux node. The Data Node was able to connect and replicate, but MR jobs failed. JobTracker/TaskTracker wasn't translating the path to the data block. They were telling the Linux node to look in C:\.... for the data block, and that obviously didn't work.
I feel like it's only a small change away from working, at that point, I had to shift my focus and couldn't work on it any more.
Have you had any success?
Andrew
________________________________
Andrew Machtolff / Senior Consultant
[http://images.askcts.com/images/cts_logo_email.png]<http://www.askcts.com/>
www.askcts.com<http://www.askcts.com/>
amachtolff@askcts.com<ma...@askcts.com>
[http://images.askcts.com/images/cts_logo_linkedin.png]<http://www.linkedin.com/company/cts-inc>[http://images.askcts.com/images/cts_logo_twitter.png]<https://twitter.com/askCTS>[http://images.askcts.com/images/cts_logo_facebook.png]<https://www.facebook.com/askCTS>
From: Ian Jackson [mailto:Ian_Jackson@trilliumsoftware.com]
Sent: Friday, November 22, 2013 4:05 PM
To: user@hadoop.apache.org
Subject: Heterogeneous Cluster
Has anyone set up a Heterogeneous cluster, some Windows nodes and Linux nodes?
RE: Heterogeneous Cluster
Posted by Andrew Machtolff <am...@askcts.com>.
Yes, I set one up as a test. I had a windows cluster of 3 machines, and added a 4th Linux node. The Data Node was able to connect and replicate, but MR jobs failed. JobTracker/TaskTracker wasn't translating the path to the data block. They were telling the Linux node to look in C:\.... for the data block, and that obviously didn't work.
I feel like it's only a small change away from working, at that point, I had to shift my focus and couldn't work on it any more.
Have you had any success?
Andrew
________________________________
Andrew Machtolff / Senior Consultant
[http://images.askcts.com/images/cts_logo_email.png]<http://www.askcts.com/>
www.askcts.com<http://www.askcts.com/>
amachtolff@askcts.com<ma...@askcts.com>
[http://images.askcts.com/images/cts_logo_linkedin.png]<http://www.linkedin.com/company/cts-inc>[http://images.askcts.com/images/cts_logo_twitter.png]<https://twitter.com/askCTS>[http://images.askcts.com/images/cts_logo_facebook.png]<https://www.facebook.com/askCTS>
From: Ian Jackson [mailto:Ian_Jackson@trilliumsoftware.com]
Sent: Friday, November 22, 2013 4:05 PM
To: user@hadoop.apache.org
Subject: Heterogeneous Cluster
Has anyone set up a Heterogeneous cluster, some Windows nodes and Linux nodes?
RE: Heterogeneous Cluster
Posted by Andrew Machtolff <am...@askcts.com>.
Yes, I set one up as a test. I had a windows cluster of 3 machines, and added a 4th Linux node. The Data Node was able to connect and replicate, but MR jobs failed. JobTracker/TaskTracker wasn't translating the path to the data block. They were telling the Linux node to look in C:\.... for the data block, and that obviously didn't work.
I feel like it's only a small change away from working, at that point, I had to shift my focus and couldn't work on it any more.
Have you had any success?
Andrew
________________________________
Andrew Machtolff / Senior Consultant
[http://images.askcts.com/images/cts_logo_email.png]<http://www.askcts.com/>
www.askcts.com<http://www.askcts.com/>
amachtolff@askcts.com<ma...@askcts.com>
[http://images.askcts.com/images/cts_logo_linkedin.png]<http://www.linkedin.com/company/cts-inc>[http://images.askcts.com/images/cts_logo_twitter.png]<https://twitter.com/askCTS>[http://images.askcts.com/images/cts_logo_facebook.png]<https://www.facebook.com/askCTS>
From: Ian Jackson [mailto:Ian_Jackson@trilliumsoftware.com]
Sent: Friday, November 22, 2013 4:05 PM
To: user@hadoop.apache.org
Subject: Heterogeneous Cluster
Has anyone set up a Heterogeneous cluster, some Windows nodes and Linux nodes?
RE: Heterogeneous Cluster
Posted by Andrew Machtolff <am...@askcts.com>.
Yes, I set one up as a test. I had a windows cluster of 3 machines, and added a 4th Linux node. The Data Node was able to connect and replicate, but MR jobs failed. JobTracker/TaskTracker wasn't translating the path to the data block. They were telling the Linux node to look in C:\.... for the data block, and that obviously didn't work.
I feel like it's only a small change away from working, at that point, I had to shift my focus and couldn't work on it any more.
Have you had any success?
Andrew
________________________________
Andrew Machtolff / Senior Consultant
[http://images.askcts.com/images/cts_logo_email.png]<http://www.askcts.com/>
www.askcts.com<http://www.askcts.com/>
amachtolff@askcts.com<ma...@askcts.com>
[http://images.askcts.com/images/cts_logo_linkedin.png]<http://www.linkedin.com/company/cts-inc>[http://images.askcts.com/images/cts_logo_twitter.png]<https://twitter.com/askCTS>[http://images.askcts.com/images/cts_logo_facebook.png]<https://www.facebook.com/askCTS>
From: Ian Jackson [mailto:Ian_Jackson@trilliumsoftware.com]
Sent: Friday, November 22, 2013 4:05 PM
To: user@hadoop.apache.org
Subject: Heterogeneous Cluster
Has anyone set up a Heterogeneous cluster, some Windows nodes and Linux nodes?