Some traps not being sent

[David Livingstone]

I have HPSIM 5 loaded on RH ES 3.0. loaded on
a DL380G3. Managed clients are also RH ES 3.0
on DL380 loaded with latest RH updates and
running psp-7.40.rhel3.linux.en.

If I pull a nic(ie eth0 bcm5700) on one of the
managed nodes a msg is generated to /var/log/messages and an email to root. A trap
is also sent and received by the SIM server.
However when a disk event occurs(ie Smart array
cciss raid) such as pulling and reinserting a
disk the trap is not sent and nothing
is recieved on the server. I do get an entry
in the messages file and an email to root is
generated ie

Subject: HP Insight Management Agents Trap Alarm

Trap-ID=3047

Spare Drive Status Change: Slot 0 SCSI Bus 2 Target 2.
Status is now Inactive.


Why is no trap sent ? I have used tcpdump
to verify.


Thanks

[David Claypool]

Are you sure it's not being sent or is it possible that it's not received? Is it received and discarded, maybe, because of a configuration issue?

Turn on 'Receive unregistered traps' in HP SIM and see if you are in fact geting the trap and not being able to interpret it. Also, double check to make sure that after you installed HP SIM that you applied the 7.4 MIB update to it.

[David Livingstone]

- Thanks for the reply.
- The Server is set to receive unregistered traps.
- On the HPSIM server I only had the base
5.0 install. I downloaded and installed
successfully upd740mib and HP SIM 5.0
Update 1 for Linux.
- unfortunately this hasn't changed anything.
- To verify transmission reception I set up
a test system hatest2(RH EL 3, psp 7.40, Proliant DL380G4).
- On the HPSIM server I then ran tcpdump for
all packets coming from hatest2.
- I then pulled the eth0 nic and reconnected
on hatest2. On HPSIM I do get the event
for the nic being disconnected(but not
reconnected). Here is what I received
from tcpdump :
[root@rtcedm1 log]# tcpdump -v -i eth1 host hatest2
tcpdump: listening on eth1
15:13:21.668514 arp who-has rtcedm1 tell hatest2.cn.ca
15:13:21.668538 arp reply rtcedm1 is-at 0:b:cd:9c:d5:d6
15:13:21.668821 hatest2.cn.ca.32789 > rtcedm1.snmptrap: { SNMPv1 { Trap(35)
E:232 165.115.204.196 enterpriseSpecific[specific-trap(18006)!=0] 275995
.iso.org=[|snmp]} } (DF) [tos 0x50] (ttl 64, id 0, len 306)
15:13:23.368448 hatest2.cn.ca.32789 > rtcedm1.snmptrap: { SNMPv1 { Trap(35)
E:232 192.168.81.33 enterpriseSpecific[specific-trap(18005)!=0] 276165
.iso.org=[|snmp]} } (DF) [tos 0x50] (ttl 64, id 0, len 306)

- If I now pull the spare drive as above and
then re-insert tcpdump does log connections
however hpsim shows nothing ie

15:13:44.136065 hatest2.cn.ca.32791 > rtcedm1.snmptrap: { SNMPv1 { Trap(35)
E:232 192.168.81.33 enterpriseSpecific[specific-trap(3046)!=0] 278238
[|snmp]}
} (DF) [tos 0x50] (ttl 64, id 0, len 451)
15:13:44.170979 hatest2.cn.ca.32794 > rtcedm1.snmptrap: { SNMPv1 { Trap(35)
E:232 192.168.81.33 enterpriseSpecific[specific-trap(3047)!=0] 278243
[|snmp]}

What is curious is that the trap which did
showup was the one with the correct
ip address(165.115.204.196 - eth1 on hatest2) while the ones which did not appear
show the eth0 hatest2 interface 192.168.81.33)

Ideas ?

Thanks

[Pedro VERA DE LA ROCHA]

Hi David.

Please, may I ask you something related to your tests?

I am doing tests remotely, so I cannot pull a cable, instead I am testing with:

'ifconfig eth1 down'

I see the corresponding messages show up in /var/log/messages but no trap is sent to the external SNMP Manager.

However, if I manually send a trap like this:

'snmptrap -v2c -c SNMP-trap snmpman '' .1.3.6.1.6.3.1.1.5.3 ifIndex i 2 ifAdminStatus i 1 ifOperStatus i 1'

then the trap is indeed sent to the SNMP Manager.

I guess then my snmpd.conf file is well configured, so why the trap would not be sent?

If you take down an interface as I am doing do you get a trap?

[David Livingstone]

Pedro,

The problem I am dealing with is that
the trap is sent however it is sent with
the wrong interface address in the
trap message. The end result is that
the trap is not seen by HPSIM.

- suggest you run ethereal or tcpdump
from your HPSIM and verify what
you recieve if anything.
- I have a case open with HP on this
one and supposedly a fix is coming soon.
This only appears to happen when you have
multiple ip addresses/interfaces on the
machine.

[Pedro VERA DE LA ROCHA]

Hi, David, thanks for your reply.

My problem is actually I don't get any trap when I shutdown the interface. I use 'tcpdump' and I am possitive: no trap gets out from the machine. In my understanding, this should generate a trap, right?

I am working remotely, I can try when I will be on site to pull an ethernet connector and check if the trap is generated then.

By the way, do you know how I could do to generate traps because of low disk space or other thresholds? I mean by setting some known read-write MIB variable, not by using HPSIM because I don't use this software, I am only managing some Proliant nodes which have 'hpasm' installed.

Regards.