/usr/sbin/mib2agt

[Joe Profaizer]

What is this? And why does it take up so much memory?

1 S root 944 1 0 154 20 40363200 25452 9b051c Ja
n 15 ? 2:07 /usr/sbin/mib2agt


..Joe

[William Baines · Accepted solution]

mib2agt is the SNMP mib agent process

There is a known memory leak problem - see patch PHSS_21046 for HP-UX 11.0 system

[Joe Profaizer]

Thanks!! Is there a way to stop and restart this process after installing the patch or must I reboot?

[Berlene Herren]

You can stop the service and restart it using the #/sbin/init.d/ stop
#/sbin/init.d/ start

scripts.

Berlene

[John Palmer]

Applying the patch will automatically restart the mib2agt process. A reboot is not required.

Regards,
John

[Tan Shirley]

Dear all,

I had the same problem as Joe regarding the mib2agt taking up lots of memory. I've tried installing the patch but seems like it's not of too much help.

Instead, I've totally removed the link in /sbin/rc2.d and /sbin/rc1.d.
Now, my system has got so much more free memory !

But I'm worried... will my system be OK without the mib2agt? Thought it's just an Snmp agent... shouldn't be a big issue, right?

Please advise... Thanks !

Regards,
Shirley

[Sandy Orr]

I am running HP-UX 11.0 on a 6 way N-Class with 2GB RAM, 1GB swap, swapmem_on=1 and dbc_max/min set to 50/5.

Sept 2000 XSWGR/XSWHWCR bundles are installed with oracle 8.1.6.2 and our companies products. Afer a few days memory runs out and login is impossible. The initial assumption was a product memory leak, but investigations have shown the problem to be a more general one:

- Product shared memeory segments not released
- Oracle shared memory not released
- Glance shared memory not released
- buf cache grows to 1GB

There is no swapping

swapinfo shows 'memory' % growing.
vmstat shows free pages decreasing (possibly co-inciding with ftp sessions)

o Am I missing more recent patches?
o Will setting the swap area to a better size resolve this and, if so, is this not a defect in memory management based on the above configuration?

I have been investigating this for two weeks with no single symptom and several red herrings including mib2agt high memory usage resolved by patch PHSS_21046

Please can someone show me the light?

This is the third recent hp IT resource forum I have found today around very similar memory management issues. See also:
- HP-UX 10.20 memory choking
- Memory Util in GlancePlus

Thanks
Sandy

[Dennis Raudenbush]

This is responding to Sandy's memory issue:

We too have a ton of HP-UX 11.0 servers running Oracle 8x. Unless this nclass is also acting as a busy NFS server (i.e. exporting filesystems to other UNIX NFS client server), I'd decrease dbc_max_pct from default 50% to 10-15%. Oracle defines it's own disk memory buffers. We too have psuedo swap enabled with swapmem_on=1 - however I've seen some other posts stating that Oracle/Sybase recommend turning this off. I too caught the mib2agt memory leak problem which I hope this patch resolves it, otherwise I'll be disabling this also. After downsizing dbc_max_pct I'd take ps -elf snapshots to find who is eating the memory. The only time I've experienced problems with share memory segments not releasing is when a process gets hung on I/O, HP-UX will mark it for deletion but will never let it go until the process dies. Unfortunately one of my biggest complaints with HP-UX is you can't kill a process waiting on I/O completion (really sucks HP - so much for production quality OS that requires a reboot for this situation that no other SVID V UNIX requires). For instance Oracle LOCAL= procs oftentimes will stop a server shutdown unless you kill them off. We use shutdown immediate - but even with that Oracle LOCAL= procs still sometimes hang-up. I'd also check 8.1.6.2 - chances are Oracle has reports/fixes for process memory leaks. We've also applied the Sept2000 GR/HW bundles. Make sure after these you apply the Oracle recommended patches.
Hope this helps.

john creighton

[Bill Hassell]

To answer the mib2agt question, SNMP processes are not needed if you are not using network monitoring tools such as OpenView Node Manager somewhere on your network. To disable the SNMP processes, there are 4 SNMP files to edit in /etc/rc.config.d. But before you edit the files, stop all the processes like this:

/sbin/init.d/SnmpHpunix stop
/sbin/init.d/SnmpMaster stop
/sbin/init.d/SnmpMib2 stop
/sbin/init.d/SnmpTrpDst stop

The reason to stop these first is that trying to stop the processes using the above commands when the files below state that they were not supposed to be run in the first place, sometimes doesn't work.

In each of the files below, there will be an enabling env variable set =1, as in: SNMP_MIB2_START=1, so edit the file and change it to SNMP_MIB2_START=0. In /etc/rc.config.d:

SnmpHpunix: SNMP_HPUNIX_START=0
SnmpMaster: SNMP_MASTER_START=0
SnmpMib2: SNMP_MIB2_START=0
SnmpTrpDst: SNMP_TRAPDEST_START=0

Now, none of the SNMP processes are running and won't be started after you reboot.