Monitor a Log File

[Trips2007]

Is it possible to monitor for keywords (like FATAL or ERROR) in a log file?

How?

[ableev]

Try this:

• Create item key like

  system.run["cat /var/log/file | grep FATAL"]

• Store value: Delta (Simple Change)
• Trigger with formula item_key.last(0)>0 (or smthng)

If your item will update one time per minute, you will get count of fatal errors in file.
You can do the same for ERROR.

[Colttt]

take a look at this:
http://www.zabbix.com/documentation/...ypes/log_items

[NOB]

Logfile monitoring with passive agent ?

Hi

I remember a discussion about changing the logfile monitoring to
passive instead of "agent (active)" only.

This would help a lot for logfile monitoring in a clustered environment,
i.e. the logfile "moves" with the application and filesystem / logical voiumes
from server to server.
In that case the zabbix server would "poll" the logfile via the virtual IP
address of the service from the agent and the current position is
stored independent of the agent (and log[...] item) running on the both monitored servers.

So my question is:

Was this feature finally implemented (in 1.8.x ?) and the documentation was not updated
or is this still not available ?

Kind regards

Norbert.

[NOB]

Answer found in source

Hi

obviously, the log file monitoring is still just possible for an active agent.

Kind regards

Norbert.

[Trips2007]

Hi

obviously, the log file monitoring is still just possible for an active agent.

Many things about Zabbix are not obvious to people who are just getting started. There is a great deal of conflicting information out there (probably due to changes between different versions). Unfortunately the official docs leave a lot out.