What are your monitors for elastic search in production?


(Tony Chong) #1

For those that are running elastic search in production environments, what
exactly are you monitoring? We are using nagios and I was thinking a simple
port check could verify if the service is running, but how about the
indexes? Is there any way to check to see if the indexes are corrupted or
not?


(David Pilato) #2

You could have a look at this thread talking about that. http://elasticsearch-users.115913.n3.nabble.com/Monitor-ES-td2635070.html http://elasticsearch-users.115913.n3.nabble.com/Monitor-ES-td2635070.html

There is a link to a nagios pl script : http://elasticsearch-users.115913.n3.nabble.com/attachment/2636406/0/elasticCheck.pl

Hope this helps.

David.

De : elasticsearch@googlegroups.com [mailto:elasticsearch@googlegroups.com] De la part de Tony Chong
Envoyé : lundi 8 août 2011 19:58
À : elasticsearch@googlegroups.com
Objet : What are your monitors for elastic search in production?

For those that are running elastic search in production environments, what exactly are you monitoring? We are using nagios and I was thinking a simple port check could verify if the service is running, but how about the indexes? Is there any way to check to see if the indexes are corrupted or not?


(Tony Chong) #3

Awesome. Thanks alot!


(James Cook) #4

There is a health check you can do. Does Nagios let you parse JSON response
data?
http://www.elasticsearch.org/guide/reference/api/admin-cluster-health.html

Status is green, yellow or red.

I'd also like to know if there is any technique to determine whether split
brain has occurred in a very dynamic environment (non-unicast, new nodes
started and shutdown based on CPU utilization)?

-- jim

On Mon, Aug 8, 2011 at 1:58 PM, Tony Chong tonyjchong@gmail.com wrote:

For those that are running elastic search in production environments, what
exactly are you monitoring? We are using nagios and I was thinking a simple
port check could verify if the service is running, but how about the
indexes? Is there any way to check to see if the indexes are corrupted or
not?


(Otis Gospodnetić) #5

Hi,

We have and use our SPM for Solr:

If there is interest, we can add support for various ES metrics -
please let us know!

Otis

Sematext is hiring Search Engineers -- http://sematext.com/about/jobs.html

On Aug 8, 1:58 pm, Tony Chong tonyjch...@gmail.com wrote:

For those that are running elastic search in production environments, what
exactly are you monitoring? We are using nagios and I was thinking a simple
port check could verify if the service is running, but how about the
indexes? Is there any way to check to see if the indexes are corrupted or
not?


(Lukáš Vlček) #6

Hi Otis,

just out of curiosity, what load does this monitoring put on the search
machine/cluster? (2-10% of CPU in avg?)

Regards,
Lukas

On Tue, Aug 9, 2011 at 9:06 AM, Otis Gospodnetic <otis.gospodnetic@gmail.com

wrote:

Hi,

We have and use our SPM for Solr:
http://sematext.com/spm/index.html

If there is interest, we can add support for various ES metrics -
please let us know!

Otis

Sematext is hiring Search Engineers -- http://sematext.com/about/jobs.html

On Aug 8, 1:58 pm, Tony Chong tonyjch...@gmail.com wrote:

For those that are running elastic search in production environments,
what
exactly are you monitoring? We are using nagios and I was thinking a
simple
port check could verify if the service is running, but how about the
indexes? Is there any way to check to see if the indexes are corrupted or
not?


(Shay Banon) #7

On Mon, Aug 8, 2011 at 11:08 PM, James Cook jcook@tracermedia.com wrote:

There is a health check you can do. Does Nagios let you parse JSON response
data?
http://www.elasticsearch.org/guide/reference/api/admin-cluster-health.html

Status is green, yellow or red.

I'd also like to know if there is any technique to determine whether split
brain has occurred in a very dynamic environment (non-unicast, new nodes
started and shutdown based on CPU utilization)?

If you know which nodes are currently up, then you can go one by one, and
ask for the cluster state, and check that all see the same master node.

-- jim

On Mon, Aug 8, 2011 at 1:58 PM, Tony Chong tonyjchong@gmail.com wrote:

For those that are running elastic search in production environments, what
exactly are you monitoring? We are using nagios and I was thinking a simple
port check could verify if the service is running, but how about the
indexes? Is there any way to check to see if the indexes are corrupted or
not?


(Otis Gospodnetić) #8

Hi Lukas,

Non-scientific, observation-based estimate is that it's on the lower
side of that range.

Otis

Sematext is hiring Search Engineers -- http://sematext.com/about/jobs.html

On Aug 9, 4:36 am, Lukáš Vlček lukas.vl...@gmail.com wrote:

Hi Otis,

just out of curiosity, what load does this monitoring put on the search
machine/cluster? (2-10% of CPU in avg?)

Regards,
Lukas

On Tue, Aug 9, 2011 at 9:06 AM, Otis Gospodnetic <otis.gospodne...@gmail.com

wrote:
Hi,

We have and use our SPM for Solr:
http://sematext.com/spm/index.html

If there is interest, we can add support for various ES metrics -
please let us know!

Otis

Sematext is hiring Search Engineers --http://sematext.com/about/jobs.html

On Aug 8, 1:58 pm, Tony Chong tonyjch...@gmail.com wrote:

For those that are running elastic search in production environments,
what
exactly are you monitoring? We are using nagios and I was thinking a
simple
port check could verify if the service is running, but how about the
indexes? Is there any way to check to see if the indexes are corrupted or
not?


(system) #9