Agent 1.6.1 problem (?) proc.num

[DeXTeR-ITA]

Hi all!

I have a one little problem with zabbix agent 1.6.1 on debian etch/sarge.

I upgrade the server from 1.4.4 to 1.6.1 and I found no problem, then I tried to upgrade agents in some debian system but the key proc.num[heartbeat], in the linux template, report the wrog value (0 instead of 4).

ps -ef |grep heartbeat

root     30119     1  0 Nov11 ?        00:00:43 heartbeat: heartbeat: master control process
nobody   30121 30119  0 Nov11 ?        00:00:00 heartbeat: heartbeat: FIFO reader
nobody   30122 30119  0 Nov11 ?        00:00:00 heartbeat: heartbeat: write: ucast bond0
nobody   30123 30119  0 Nov11 ?        00:00:02 heartbeat: heartbeat: read: ucast bond0
root     22096 22073  0 10:00 pts/0    00:00:00 grep heartbeat

with agent 1.6.1

zabbix_get -s xx.xx.xx.211 -p 10050 -k proc.num[heartbeat]
0

with agent 1.4.4

zabbix_get -s xx.xx.xx..212 -p 10050 -k proc.num[heartbeat]
4

Someone have any idea?
Thanks a lot!

[SMB]

I'm having the same problem monitoring a python script. proc.num[script.py] returns 0. There is one running.

[DeXTeR-ITA]

I hold the same problem, don't find a solution...

[krimson]

Looking in the source, it seems zabbix looks up the processname in /proc/<pid>/status. The first line says Name: <processname>. Lookup the pid of your heartbeat or blah.py process and see what is says for processname in this file. Try sticking that into proc.num[...].

Chances are it says heartbeat: heartbeat ... and python in the case of the python script. Not too sure the zabbix linux agent is doing a proper job here on processname matching, cos the string in the status file seems to be limited to just 15 (?) characters.