SNMP traps for disk failure?

[Martin Wolff]

Hi, we have a TRU64 5.1B system running for 5 years now. We have LSM managing a disk array of 14 scsi disks. As time goes by probability of failure increases, and we want to be prepare in case of a disk failure. I have searched a lot and did not find any way to configure a trap that can signal a scsi disk failure, nor a LSM trouble. Is this posible?
Thanks in advance.

[Martin Wolff]

Sorry i did not tell that it´s a DS-25 cluster.

[Ivan Ferreira]

WEBES is a product that you can download and install. WEBES will report any problem that it can find on the system binary error log, like hardware errors (including disks). Notifications can be sent via email.

I will suggest you to install WEBES for Tru64.

[Martin Wolff]

Thank you very much for that information.
I was setting up a script that checks "voldisk list" output and send an email if the status if any of the disks change. But this won´t work for the local disks of each machine, so i guess i will study WEBES and try it out. Unless there is a SNMP solution, as we have all aplications using SNMP for sending alarms, and also the filedomain % ocupation are sent via SNMP.

[Geert Van Pamel]

The EVM utility sends e-mail to the root account for any event with priority greater than 600.

I assume a disk or hardware error will have a priority code greater than 600?

In that case it should be sufficient to set a forward in the root e-mail account (could be a distribution list, or an address pointing to a shared mailbox) e.g. with:

vi ~/.forward

You could trigger a test message with:

evmpost -p 600 -u "Simulate disk error"

Normally EVM is enabled by default. More info with "man 5 evm".

I have read somewhere that EVM could also run a script, so this would allow to send an SMS message? If somebody could give more details on this...

[marsh_1]

hi,

you might also try the systems insight manager agents they can be configured to trap out to whatever snmp management software you have (best to go to hp software such as hp sim server as you don't have to load in the mibs :-) ).
this can also give you a lot of statistical info on cpu , disk io etc see here for download etc

http://h30097.www3.hp.com/cma/

[marsh_1]

hi,

you might also try the systems insight manager agents they can be configured to trap out to whatever snmp management software you have (best to go to hp software such as hp sim server as you don't have to load in the mibs :-) ).
this can also give you a lot of statistical info on cpu , disk io etc see here for download etc

http://h30097.www3.hp.com/cma/

[Martin Wolff]

Thank you very much, tommorrow i will use your ideas to test in the lab.

[Martin Wolff]

Hi, the maint window could not be opened yet!!
I was wondering about the following command:

hwmgr -status component

That shows lines like the following:

HWID:HOSTNAME SUMMARY STATE STATE LEVEL NAME
----------------------------------------
95: do online available dsk15

If this command can give me the status info maybe i can make a monitor script that sends an email if the state changes.
What do you think about this?
Does the state changes? Has anyone in the Forum used this command?

Thank you very much in advance,
Martin.

[Martin Wolff]

I changed the focus away from SNMP, and made a script that check every 5 minutes the status of the disks. If there is a failure it sends an SMS to the people specified at the script.
This was the easiest way to solve my problem.
Thank you all for the sugestions.