IPMI issues

[nickp]

Hi

IPMI monitoring is setup and working, however it seems Zabbix "looses" the connection or the ipmi interface drops the connection.

For example
2012.Aug.17 12:22:11 3.1413
2012.Aug.17 12:21:11 3.1413
2012.Aug.17 12:21:00 3.1413
2012.Aug.17 12:16:21 3.1413
2012.Aug.17 12:11:11 3.1413
2012.Aug.17 12:08:11 3.1413
2012.Aug.17 12:06:25 3.1413

It should check every minute.

Anybody experiencing the same issues?

[nickp]

After changing the pollers to 1, still not solved and getting:


1632:20120817:172721.843 IPMI item [DIMM-2A_temperature] on host [ESXI01] failed: first network error, wait for 15 seconds
1601:20120817:172736.437 IPMI item [DIMM-2A_temperature] on host [ESXI01] failed: another network error, wait for 15 seconds
1601:20120817:172741.450 enabling IPMI checks on host [DB]: host became available

I found something in de irmc log's (Fujitsu servers uses IRMC) that the ipmi connects with version 1.5? Could that be the issue, it should also support 2.0.

regards

[nickp]

Version of ipmitool is: OpenIPMI-2.0.16-12.el6.x86_64
Zabbix 2.0.2 compiled with ipmi

[Chal]

Looks like the problem is in Zabbix, not in IRMC.... I wrote a shell script, that flooding a Fuji. IRMC IPM service (with ipmitool command). The script run 1 ipmi query per second, and I not see any problem, IRMC always responding correctly.

I found a relevant bug:

https://support.zabbix.com/browse/ZBX-3243

Thereafter I try reduced ipmi pollers in Zabbix config to 1 (dont ask me why, I just testing it . Result: no problems encountered the last 10 hours, and the "IPMI item failed: another network error..." messsages have disappeared from the log, and no more gap in graphs. The question is, that one poller enough for the all IPMI hosts...