Very Computer

Does anybody have a good script or even just a rule of thumb about the best way
to monitor system error messages on AIX, as reported by errpt and taking the
appropriate action ?

Capturing the output from errpt is easy . The tricky part is identifying which
messages are urgent and require immediate attention ( like disk or SCSI errors
) and which ones are just warnings.

Thanks, PP

Quote:> Does anybody have a good script or even just a rule of thumb about the
best way
> to monitor system error messages on AIX, as reported by errpt and taking
the
> appropriate action ?

> Capturing the output from errpt is easy . The tricky part is identifying
which
> messages are urgent and require immediate attention ( like disk or SCSI
errors
> ) and which ones are just warnings.

> Thanks, PP

There is a errnotify class in ODM that runs a defined command when an error
occurs.
This class is standard in AIX4.
For AIX3 I have a couple of files that implement this errnotify class and
add a smit interface. I don't know if this smit interface works with AIX4
(not tested yet).

Some major problems to watch already implemented in the
/usr/lib/ras/notifymeth which is the default method (program) called.

I found out that watching DISK_ERR2 (disconnect of disk) and DISK_ERR4(bad
block relocation) is useful.
BBR should be responded to if it occurs more often than 2 times per GB disk
capacity and year.

There may be other events you want to watch (TAPE_ERR s to clean the tape
drive).

I'm sorry but it looks like you have to deal with odmget, odmadd, odmdelete
to add your tailored notification.

--
Burkhard Weeber
viastore systems GmbH
P/O Box 300668
D-70446 Stuttgart

If you're running AIX4 and are interested in H/W errors, you might
look into diagela. Try "info -s diagela"