Co-locate elasticsearch with hadoop/yarn


(Ronny Vaningh-2) #1

Hi

I have some beefy boxes with 512 Gb ram and I would like to co-locate
yarn/hadoop with elasticsearch

Does anyone have experience in doing the same ?
How did you split the resources (memory/disk) across both functions ?

Hdfs like jbod while E.S. like raid10

Thank

Reg

--
You received this message because you are subscribed to the Google Groups "elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send an email to elasticsearch+unsubscribe@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/elasticsearch/CACA5U5mW8zn0BLWryeePdR%2B0yiBGhd_9pQWVy-Uufj3H-oeQhA%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.


#2

I have not tried this but my initial thoughts would be

  • Set ES_HEAP_SIZE = 30 GB, give Hadoop an appropriate amount, leave the rest for the OS cache.
  • Set the filesystem paths where ES and Hadoop store data to separate physical disk(s). You don't want them contending for bandwidth.
  • You don't have to use RAID for ES, you can use multiple data paths if you have multiple disks.
  • At this size, many people choose to run multiple instances of ES on a single physical. Give each instance 30GB and point to different disks.

A

On Sep 4, 2014, at 11:32 AM, Ronny Vaningh ronny@guard-it.be wrote:

Hi

I have some beefy boxes with 512 Gb ram and I would like to co-locate yarn/hadoop with elasticsearch

Does anyone have experience in doing the same ?
How did you split the resources (memory/disk) across both functions ?

Hdfs like jbod while E.S. like raid10

Thank

Reg

--
You received this message because you are subscribed to the Google Groups "elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send an email to elasticsearch+unsubscribe@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/elasticsearch/CACA5U5mW8zn0BLWryeePdR%2B0yiBGhd_9pQWVy-Uufj3H-oeQhA%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send an email to elasticsearch+unsubscribe@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/elasticsearch/0A8CF809-7F38-40B2-AE23-508F1477B823%40elasticsearch.com.
For more options, visit https://groups.google.com/d/optout.


(Ronny Vaningh-2) #3

Thanks Andrew

If I'm correct e.s. will first fill up the first disk, and so on if you
specify multiple data paths vs scattering over them like hdfs.. correct?
You overcome this by giving easch e.s. instanve a data path on a sepparate
disk

Thanks

Regards

Ronny
On 5 Sep 2014 02:30, "Andrew Selden" andrew.selden@elasticsearch.com
wrote:

I have not tried this but my initial thoughts would be

  • Set ES_HEAP_SIZE = 30 GB, give Hadoop an appropriate amount, leave the
    rest for the OS cache.
  • Set the filesystem paths where ES and Hadoop store data to separate
    physical disk(s). You don’t want them contending for bandwidth.
  • You don’t have to use RAID for ES, you can use multiple data paths if
    you have multiple disks.
  • At this size, many people choose to run multiple instances of ES on a
    single physical. Give each instance 30GB and point to different disks.

A

On Sep 4, 2014, at 11:32 AM, Ronny Vaningh ronny@guard-it.be wrote:

Hi

I have some beefy boxes with 512 Gb ram and I would like to co-locate
yarn/hadoop with elasticsearch

Does anyone have experience in doing the same ?
How did you split the resources (memory/disk) across both functions ?

Hdfs like jbod while E.S. like raid10

Thank

Reg

--
You received this message because you are subscribed to the Google Groups
"elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send an
email to elasticsearch+unsubscribe@googlegroups.com.
To view this discussion on the web visit
https://groups.google.com/d/msgid/elasticsearch/CACA5U5mW8zn0BLWryeePdR%2B0yiBGhd_9pQWVy-Uufj3H-oeQhA%40mail.gmail.com
https://groups.google.com/d/msgid/elasticsearch/CACA5U5mW8zn0BLWryeePdR%2B0yiBGhd_9pQWVy-Uufj3H-oeQhA%40mail.gmail.com?utm_medium=email&utm_source=footer
.
For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups
"elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send an
email to elasticsearch+unsubscribe@googlegroups.com.
To view this discussion on the web visit
https://groups.google.com/d/msgid/elasticsearch/0A8CF809-7F38-40B2-AE23-508F1477B823%40elasticsearch.com
https://groups.google.com/d/msgid/elasticsearch/0A8CF809-7F38-40B2-AE23-508F1477B823%40elasticsearch.com?utm_medium=email&utm_source=footer
.
For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send an email to elasticsearch+unsubscribe@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/elasticsearch/CACA5U5%3DUD_HuLL_DEM4J-kKL330pvnex5QLjwJkF-f_ac1Sehg%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.


(David Pilato) #4

f you give 2 disks to Elasticsearch, it will fill both at the same time.
Not one after the other.

--
David :wink:
Twitter : @dadoonet / @elasticsearchfr / @scrutmydocs

Le 5 sept. 2014 à 08:19, Ronny Vaningh ronny@guard-it.be a écrit :

Thanks Andrew

If I'm correct e.s. will first fill up the first disk, and so on if you specify multiple data paths vs scattering over them like hdfs.. correct?
You overcome this by giving easch e.s. instanve a data path on a sepparate disk

Thanks

Regards

Ronny

On 5 Sep 2014 02:30, "Andrew Selden" andrew.selden@elasticsearch.com wrote:

I have not tried this but my initial thoughts would be

  • Set ES_HEAP_SIZE = 30 GB, give Hadoop an appropriate amount, leave the rest for the OS cache.
  • Set the filesystem paths where ES and Hadoop store data to separate physical disk(s). You don’t want them contending for bandwidth.
  • You don’t have to use RAID for ES, you can use multiple data paths if you have multiple disks.
  • At this size, many people choose to run multiple instances of ES on a single physical. Give each instance 30GB and point to different disks.

A

On Sep 4, 2014, at 11:32 AM, Ronny Vaningh ronny@guard-it.be wrote:

Hi

I have some beefy boxes with 512 Gb ram and I would like to co-locate yarn/hadoop with elasticsearch

Does anyone have experience in doing the same ?
How did you split the resources (memory/disk) across both functions ?

Hdfs like jbod while E.S. like raid10

Thank

Reg

--
You received this message because you are subscribed to the Google Groups "elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send an email to elasticsearch+unsubscribe@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/elasticsearch/CACA5U5mW8zn0BLWryeePdR%2B0yiBGhd_9pQWVy-Uufj3H-oeQhA%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send an email to elasticsearch+unsubscribe@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/elasticsearch/0A8CF809-7F38-40B2-AE23-508F1477B823%40elasticsearch.com.
For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send an email to elasticsearch+unsubscribe@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/elasticsearch/CACA5U5%3DUD_HuLL_DEM4J-kKL330pvnex5QLjwJkF-f_ac1Sehg%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send an email to elasticsearch+unsubscribe@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/elasticsearch/2B91F6B4-1A18-46FF-97DA-14CD55B63D51%40pilato.fr.
For more options, visit https://groups.google.com/d/optout.


(system) #5