| Hi,
> 3.2f dunix.
> Customer is running Polycenter Performance MOnitor 1.4 and Polycenter
> DXAdvisor on local node. HP Openview is running on the network.
> About once a month, customer sees the following error in daemon.log
> and on her console:
> Jan 22 16:26:21 ubesfin01 os_mibs[428]: os_mibs **ERROR esnmp_poll.c
> line 372: R
> ESPONSE to REGISTRATION of 'at': DPI_alreadyRegistered
This means that for some reason, os_mibs is trying to reregister
its MIB subtrees after it has already done so.
> Jan 22 16:26:21 ubesfin01 os_mibs[428]: os_mibs WARNING esnmp_poll.c
> line 201: e
> snmp_poll: Could not process a packet
> Jan 22 16:26:21 ubesfin01 os_mibs[428]: os_mibs **ERROR os_mibs.c line
> 340: eSNMP error: -2
This is expected 'unwinding', the -2 indicates a failed registration.
os_mibs now sleeps 30 seconds and reestablishes its protocol state w/snmpd.
Since it has already initiated communications, the first step will be to
'close' its current state.
> Jan 22 16:26:51 ubesfin01 snmpd[436]: Closing subagent os_mibs, reason:
> 103
And here we see the close msg being received and logged by snmpd.
The reason code of 103 is due to a bug in libesnmp.so I just discovered
thanks to this note!
It appears that things continue on their merry way, os_mibs having
reconnected w/ snmpd. Is that correct?
The question is what kicks off the whole thing?
Are they shutting down or somehow signalling os_mibs?
Is this all that's in the log?
Is anything happening on these once a month times?
Thanks,
Mike
|