More Apache Monitoring

[gpmidi]

Problem With Type Of Check

hi paul,

i sent you the debug log since the server logs dont show anything.
Maybe my setup is just weird or the agentd isnt correctly configured?

We have X servers - each running agentd.
The server "MONITORING" (DNS via /etc/hosts: "MONITORING", local IP Adress: 192.168.0.10) runs the zabbix_server which uses the DB on "TS" (DNS via /etc/hosts: "TS", local IP Adress: 192.168.0.11) to store the information. The web frontend is also located on "TS".

Heres the configuration of the zabbix_server on "MONITORING"


Heres the configuration of the zabbix_agentd on "TS"

Try using mass-update to change the type of check from "Zabbix Agent (Active)" to "Zabbix Trapper". I --think-- that will fix the problem you're having. In either case I'll be updating the XML Import with the right check type in the next couple days.
-Paul

[FireMail]

sadly it didnt remove the problem.

The items now stay activated, but still "failed: 20"

[gpmidi]

Edit the script and send resulting file

sadly it didnt remove the problem.

The items now stay activated, but still "failed: 20"

Let's gather some debug info:

1. Edit the script and change the line with "os.remove(tempfilepath)" to "print tempfilepath". Make sure that the number of spaces before the "os.re..." is the same as before "print..." (Python uses spacing instead of {}).
2. Run the script
3. At the end of the script it will print out a path. Send me the contents of that file (paul at gpmidi dot ignorethis net).
4. Revert the script back to normal.

If you have problems with editing the script let me know and I'll post a debug version or something. Also, I'll be in the Zabbix IRC channel (#zabbix on freenode) today. So just say my name a few times to get my attention or, better yet, directly send me a message via IRC.
-Paul (gpmidi_wrk on IRC)

[FireMail]

contacted you via IRC, hope we find the problem

[gpmidi]

contacted you via IRC, hope we find the problem

Firemail and I worked together via IRC and figured out that it's the keys. The key in the template is "apache[localhost,BusyWorkers]" and the template is sending it as "apache[TS,BusyWorkers]". TS is the hostname of the Apache instance that is being monitored.

This raises the question about how to fix this. This leaves up with a couple of options:

1. Alter the script so that items are sent up like "apache[BusyWorkers]". This would remove the ability to monitor multiple Apache instances with a single Zabbix host.
2. Alter the script so that (by default) items are sent up like "apache[,BusyWorkers]" but there is a flag that will cause it to send up with the hostname of the Apache server it's checking (ie "apache[somehostname,BusyWorkers]"
3. Use host level macros to do something like "apache[{$APACHEHOST},BusyWorkers]". This should work in theory but am having problems getting this working.
4. Use something like "apache[{HOSTNAME},BusyWorkers]". I'm not even sure if this will work. FireMail's testing indicates that it doesn't. This method would be the ideal solution as it will always work but allow users to override the checked host.

[fraenki]

I'd like to suggest to use the Apache ScoreBoardFile to gather information. While the /server-status page may be unresponsive due to high server load, the ScoreBoardFile is always available.

The script "ftss" is able to present the ScoreBoardFile in plaintext to make processing with Zabbix easy:

# ftss /var/run/apache_status
13762   K       127.0.0.1       127.0.0.1       GET / HTTP/1.1
13763   _
13764   _
13765   _
13766   _
13767   _

It's easy to use and to implement... just take a look at the Website for a 3-step manual:

http://fabletech.com/ftss

Bye
- fraenki

[FireMail]

hi paul,

i know its not directly related to you but something i wanted to ask. is it possible that the "kb/s" value of the apache status is already an average over the time the apache is running?
after 2 days it finally reached 95kb/s and doesnt even move slightly between peak hours and the night where it should be lower.

[bashman]

If you don't want to use a python script you can use zapache

[gpmidi]

hi paul,

i know its not directly related to you but something i wanted to ask. is it possible that the "kb/s" value of the apache status is already an average over the time the apache is running?
after 2 days it finally reached 95kb/s and doesnt even move slightly between peak hours and the night where it should be lower.

Sounds like it is averaged. But I don't know for sure.
-Paul

[gpmidi]

I'd like to suggest to use the Apache ScoreBoardFile to gather information. While the /server-status page may be unresponsive due to high server load, the ScoreBoardFile is always available.

The script "ftss" is able to present the ScoreBoardFile in plaintext to make processing with Zabbix easy:

# ftss /var/run/apache_status
13762   K       127.0.0.1       127.0.0.1       GET / HTTP/1.1
13763   _
13764   _
13765   _
13766   _
13767   _

It's easy to use and to implement... just take a look at the Website for a 3-step manual:

http://fabletech.com/ftss

Bye
- fraenki

I'm going to look into doing this. Thanks for the tip!
-Paul

[gpmidi]

If you don't want to use a python script you can use zapache

The downside is that you make a call to mod_status for each value you pull. This adds significant overhead if mod_status is showing the state of every process. Plus it doesn't parse state of all of the processes.
-Paul

[bashman]

The downside is that you make a call to mod_status for each value you pull. This adds significant overhead if mod_status is showing the state of every process. Plus it doesn't parse state of all of the processes.
-Paul

You're right Paul, but I don't want to use neither a perl scrip nor a zabbix_sender. Is there another way to do it?.

Thanks.

[gpmidi]

You're right Paul, but I don't want to use neither a perl scrip nor a zabbix_sender. Is there another way to do it?.

Thanks.

You could, in theory, do all of what my Python script does in Bash. However it would be difficult.
-Paul

[radamand]

is there any way to get this to work for a host using SSL with a self signed cert?

[gpmidi]

Depends

is there any way to get this to work for a host using SSL with a self signed cert?

It may work with a self-signed cert. Depending on the version of Python it may or may not do certificate validation.
-Paul