Zabbix Elasticsearch Integration

[yarden]

Hi all,
We're evaluating several monitoring systems(Icinga2/Bosun and Zabbix).

Our basic requirement, aside from the obvious machine monitoring, is as follows:
We have a Elasticsearch cluster with a lot of data points.
We'd like to enable our dev team with a GUI to add/remove monitoring/alerts/etc but on elasticsearch query evaluation.

I know that all monitoring systems enable running external scripts but I'm trying to avoid that because the overhead for dev teams to write these scripts.

Can zabbix generate checks/alerts on a elasticsearch query via API call (e.g. /_search?fields=response:404) ?

Any other ideas?

Thanks,
Yarden

[kloczek]

Can zabbix generate checks/alerts on a elasticsearch query via API call (e.g. /_search?fields=response:404) ?
Yarden

If you will start storing values read from such url on top of such metric/item you can build what ever you want from point of view possible alarms using triggers. None of the "integrations" is needed here to organize such monitoring.
On reading values from exact URLs possibly you can use build in agent items like web.page.get[] or web.page.regexp[] so in case some simple metrics probably you will not need even some backend scripts to read data from ES.

[yarden]

Storig Values

Hi kloczek,

Can you help me with an example?

How to configure such check?

How to configure the request to elasticsearch?

Thanks,
Yarden

[kloczek]

Hi kloczek,

Can you help me with an example?

How to configure such check?

How to configure the request to elasticsearch?

Thanks,
Yarden

OK .. lets use my own apache template as kind of example.
In my template I'm parsing mod_status URL which is usually is localhost/server-status?auto. Orn this URL you can find output like:

$ curl -s localhost/server-status?auto
Total Accesses: 25569
Total kBytes: 10860
CPULoad: .00168426
Uptime: 387707
ReqPerSec: .0659493
BytesPerSec: 28.6831
BytesPerReq: 434.927
BusyWorkers: 1
IdleWorkers: 6
Scoreboard: W___.___........................................................................................................................................................................................................................................................

To extract exact values from above I'm using keys like:

web.page.regexp[localhost,server-status?auto,80,"Total Accesses: ([0-9]+)",,\1]
web.page.regexp[localhost,server-status?auto,80,"Uptime: ([0-9]+)",,\1]
web.page.regexp[localhost,server-status?auto,80,"BusyWorkers: ([0-9]+)",,\1]
web.page.regexp[localhost,server-status?auto,80,"IdleWorkers: ([0-9]+)",,\1]

To test this on host with running apache I can use zabbix_agentd binary:

# /usr/sbin/zabbix_agentd -t 'web.page.regexp[localhost,server-status?auto,80,"Total Accesses: ([0-9]+)",,\1]'
web.page.regexp[localhost,server-status?auto,80,"Total Accesses: ([0-9]+)",,\1] [s|25583]

So as you see I can use above key to store monitoring data in unsigned int metric.