snmpd lost connection

[Jean-Christophe Valiere]

Hi,

I've installed a rhel2.1as on a hp proliantBLp20. I've installed hpasm-7.00 on it and configured the snmp services. Everything is working great and I can access to the machine from the web to see if the snmp agent is working but after a while the agent doesn't work and it seems that the snmp agent can't contact the submib any more.
In the /var/log/snmpd.log I got:
cmaX: no subagent matches the subMIB
and in the /var/spool/compaq/cma.og I got:
0x40058020Timeout: No Response from127.0.0.1

Does any body know where the probleme come from ??

[Don_89]

Your probably using the default snmp package that comes with AS 2.1.Goto http://h18023.www1.hp.com/support/files/server/us/download/19020.html and get the ucd-snmp package from HP and type rpm -ivh ucd-snmp* --force, this should install the proper snmp extensions for the blade.

[Steven E. Protter]

He's right. The only MIB thats going to work with that server is the one HP(Compaq) produced.

Better yet,not all versions of red hat are supported by HP on the proliant.

SEP

[Jean-Christophe Valiere]

Unfortunately not.
I'm already using ucd-snmp hp package.
But what is strange is that when I start the service it works, but after about 10 minutes it doesn't and the snmpd services takes 100% of the CPU.

[Don_89]

Ahhh,

I've seen this before. Uninstall version 7.0 of the hp agents and load 6.40. Your problems will go away..

I had ver 7.0 of those agents on a BL10, one of the foundation agents was polling something to do with the disk controller and started generating errors which were filling up my syslog. Not exactly the same problem but ver 7 of those agents have given my nothing but problems..

[Jean-Christophe Valiere]

Thanks a lot for the informations.

As the server is in production for the moment I can't do any changes for a while (my boss refuses ..). As there are some other servers where snmpd is working properly I am very astonished. I have compare the configurations of two servers where one was working and the other not.
I have just found one difference for the moment, the server where snmpd doesn't work has no reverse in the DNS.
Could the probleme come from here ??
The probleme is that the server where the service doesn't work were not installed by me, so I'm discovering the environment very slowly...

Thanks in advance

[Jean-Christophe Valiere]

Ok,

Some people are going to think I'm crazy or stupid but as I have some time today, I've done some more investigation about the snmp probleme I have.
I've installed all the necessary package on 4 HP ProLiant BL20p G2 and it works great (all sub agent and the snmpd service too).
When I do the same installation on a HP DL380 G3, i got the following error:
0x40058020Timeout: No Response from127.0.0.1
and the snmpd process takes 100% of the CPU.
13281 root 25 0 2832 2824 1712 R 99.0 0.0 1:32 snmpd

I've decided to debug the running process to see what failed (I should normally be stopped on the 'select' loop). But It fails before this instruction and here is the error:
0x403d3b00 in __sigprocmask (how=1, set=0xbffe7fd0, oset=0xbffe7f50) at ../sysdeps/unix/sysv/linux/sigprocmask.c:48
48 ../sysdeps/unix/sysv/linux/sigprocmask.c: No such file or directory.
in ../sysdeps/unix/sysv/linux/sigprocmask.c

It is very curious that the snmpd process failed but I was first thinking that the ucd-snmp package I've installed was too 'new' (Yes It's crazy) so I decided to install the previous package which I download from the HP Web Site of course, and naturally I fails too.
So if any body has an idee or any help files (except changing from DL 380 to ProLiant BL20p G2 ;)), I would be very happy.

PS: I've tried an older version of hpasm (v6.40), but it doesn't work too.

Thanks in advance.