From: Fabio Ferreira (Fabio.Ferreira@markway.com.br)
Date: Wed Jul 16 2003 - 12:41:00 EDT
Managers,
Many thanks to Alan Rollow and Chris Bryant for their responses.
Alan,
many gratefulness for its deep analysis.
############################################################################
#######################
The message listing included indicates multiple problems
over about three weeks. They could be unrelated. Some
of the highlights.
June 23rd, /oracle817 being full. Odds are, the file
system was full. Migrate it to a larger device, remove
unused files or find a way to give it more space.
July 6th, AdvFS I/O error. Probably a real I/O error
on the underlying device. Could merely be a bad block
or two, but it could be problem particular to the
device; rz13.
July 8th, System and Process corrected errors. These
are probably CPU and system problems.
July 10th, Adapter level I/O errors. Could be the
device out at the end of the particular I/O adapter,
or it could be the adapter itself. The I/O error on
rz11 could be related, since it is on the same bus
as the reported bus/target/LUN, but it is a different
device (bus 1, target 5, LUN 6 would rzg13, in V4
naming convention).
Find the error log translation/analysis tool that works
best for your system and take a close look at the events
that were logged around the time of the event listed
in the messages file. If the system is on a service
contract or under warranty, contact the appropriate
service vendor. The usual choices are uerf(8), DECevent
and Compaq Analyze. Which works for your system, depends
on the model of system and I didn't see that listed.
If suspect a close association of the problems, look at
the environment where the system is sitting. Make sure
the temperature and humidity are within the range allowed
for the system. Lots of problems can be heat induced if
it is summer where you are.
Alan Rollow
############################################################################
#######################
Sounds like you have a corrupt domain. You may need to run verify against
it verify -a domain5_domain is the command. If this domain comes back
corrupt then you will need to rebuild the domain and restore from tape.
HTH,
Chris Bryant
UNIX Administrator
Dollar Thrifty Automotive Group
############################################################################
#######################
Thanks,
Fabio Ferreira - Consultor de TI
MarkWay Business & Informática Ltda.
Tel/Fax: 55-21-2262-4312
<http://www.markway.com.br/>
"11 ANOS CRIANDO NEGÓCIOS"
This archive was generated by hypermail 2.1.7 : Sat Apr 12 2008 - 10:49:27 EDT