Linux 2.6.22-3-amd64 #1 SMP Mon Nov 12 17:53:18 UTC 2007 x86_64 GNU/Linux
All agents run in active mode

To resolve this, I run a simple /etc/init.d/zabbix-server restart.

I have also looked through the logs, and have not found anything. I'll attach logs next time I catch them before they are over-written.

This problem is somehow caused by the active agent handling in the server code. I changed my agent configuration from passive to active. Then the server randomly (2-48h) stops collecting data.

some logfile excerpts:

zabbix_agentd.log
28311:20080319:141202 Timeout while answering request
28311:20080319:141202 Getting list of active checks failed. Will retry after 60 seconds
28311:20080319:141305 Timeout while answering request
28311:20080319:141305 Getting list of active checks failed. Will retry after 60 seconds
28311:20080319:141408 Timeout while answering request
28311:20080319:141408 Getting list of active checks failed. Will retry after 60 seconds

zabbix_server.log
9366:20080318:200821 Active parameter [net.if.in[eth0,bytes]] is not supported by agent on host [ZABBIX-Server]
9366:20080318:200821 Active parameter [net.if.out[eth0,bytes]] is not supported by agent on host [ZABBIX-Server]
9371:20080318:203743 Executing housekeeper
9371:20080318:203746 Deleted 3631 records from history and trends
9371:20080318:213847 Executing housekeeper
9371:20080318:213850 Deleted 3652 records from history and trends
(only housekeeper messages after this)

I'll attach some server logs in a next post with debuglevel=5

Which might be interesting is, that I use a virtual server for zabbix (virtuozzo enviroment). Don't know if this is important.

This is already fixed in pre-1.4.5 code.

Alexei

Logfile (debuglevel=4) of 1.4.4 zabbix_server with active agents causing the server to stop collecting data (error still exist in pre-1.4.5 from 17.3.2008)

I opened this ticket awhile ago once the server stopped accepting (or possibly recording) data from agents.

I sent Alexei my logs and after closer inspection, the only thing I can relate this to is a busy mysql server. It happens at about 4 AM almost every night. Almost every day, zabbix needs to be restarted and I can't enable actions, because all of the high severity nodata actions are triggered when data is no longer collected.

The server stopped this night again. Actually I doubt that the MySQL Server is the problem. Zabbix is the only process, that uses the database. I tried the nightly build from 19.3.2008 for the 1.4 branch.
Before 'make install' I cleaned /usr from all files named zabbix* just to be sure.

I will attach new server logs, if someone is willing to look into them...

zabbix_server.log from pre-1.4 (19.3.2008) with debuglevel=4
The Server died between 22:04 and 22:06 so I have uploaded only this timeslot.

Torsten- Can you correlate your problem with high server load or high disk IO? I'm 99% sure that when my server disk io is very busy at night, it causes zabbix to malfunction.

I installed SAR, and right when it gets very busy disk IO, is about the time zabbix stops working properly. From the data below, it zabbix broke at about 4 AM.

SAR busy disk from march 18:
12:00:01 AM tps rtps wtps bread/s bwrtn/s
03:55:01 AM 466.96 14.74 452.22 206.62 5890.57
04:05:25 AM 1457.06 84.53 1372.53 1255.55 15686.59
04:15:01 AM 1331.55 591.61 739.93 8005.62 9049.32
04:25:01 AM 1657.70 613.01 1044.69 6307.10 13085.33

SAR normal disk from march 18:
02:05:01 PM 306.73 3.16 303.57 43.45 3491.67
02:15:01 PM 281.48 3.44 278.04 34.28 3145.71
02:25:01 PM 286.40 4.95 281.45 45.20 3155.84
02:35:01 PM 277.97 3.60 274.37 34.95 3162.64

Actually I have problems getting this IO statistics because zabbix server is running on a vserver (virtuozzo). I'll give it another try next week.

The strange thing is, that this problem only arises when I use active agents. Is this the same situation for you?

I encounter this behavior almost once per day with 1.5 from early March. It happens when our backup process starts and put pressure on disk. Something breaks at this point. I have to restart Zabbix server to get rid of that.

seems fixed in 1.4.5. - thanks!

I close it.

Solved? I'm still experiencing this issue...

I am pretty sure the problem no longer exists in 1.4.5. I need evidence if you do not agree

Alexei