Hello,

I am currently running slackware 9.0 on a Intel dual Xeon 2.4 GHz server with a Adaptec SCSI card Model: 2110S FW:380E using the Adaptec I2O RAID: Version 2.4 Build 5 module.

The system locks up constantly - I have disabled the screen saver to try and see some console messages. The screen shows the standard login prompt but the system is totally frozen.

The keyboard does not work all - no num lock, no caps -- nothing

The serial port which is connected via a null modem cable for heart no longer works causing a fail over.

The network card light is on but no connect.

I have checked every log file (I have syslog.conf setup to log everything) and there nothing close the time of the crash that I can tell. There are no entries with in 5 minutes before and there are a few around 5-10 minutes before but they look like 5 logins only from valid users.

The is running:
2.4.26 kernel

dnotify-0.16.0.tar.gz
heartbeat-1.2.0.tar.gz
libnet.tar.gz
logcheck-1.1.1.tar.gz
nail-10.5-i486-1.tgz
perl-5.8.0-i486-5.tgz
pine-4.58-i486-2.tgz
popa3d-0.6.3-i486-1.tgz
postfix-2.0.18-20040209.tar.gz
proftpd-1.2.9.tar.gz
ssmtp-2.88.tgz
sysstat-5.0.5.tar.gz

Plus a in house java app that connects to the FTP and POP accounts using a polling method. Now all the above software (excluding the java app) we have running on other servers with out any problems.

Any suggestions on how to further trouble shoot this issue ? Is it possible for the java app to cause the system crash / lock up ?? Our developers seem very reluctant to look at the java app or even consider that it could be causing the crash because the JVM runs in a virtual machine ?

P.S we have two boxes -- exact same hardware and both are crashing so I do not think it is a hardware problem. Could be a hardware compatibility problem ??

Thanks ..


Michael.

The Java program is unlikely to be causing the fault. You could try running it under a different JVM to see if it that clears the problem, but it's unlikely.

The problem is most likely to be being caused by some problem in the kernel configuration. Selecting a wrong processor version, or adding untested optimisations to the Makefile could cause this kind of behaviour. I would suggest you try recompiling the kernel (and updating it at the same time).

Another option you could try would be to remount the /var partition with the “sync” option. This disables write-aheach caching, so log messages have a much better chance of being written to the disk.

Hope that helps,

— Robert J. Lee

Hello,

Thanks for the reply ... I do not know about the kernel problem ... originally the box was running 2.4.23 and was up and running for 5 months with out a problem but during this time it was not being activitly used. Then we started our Java app and started using the box ... every since then about every 15days it crashes. We have tried upgrading the kernel to 2.4.25 and then to 2.4.26. Each time a fresh kernel source is downloaded from kernel.org. We have not made and changs to the kernel code. But we have imported our own config file ... but it is the same config file on the rest of the servers (15+ of them).

The var parition is not separately mounted

I will see what I can do about rebuilding the kernel.

Michael.

Just a thought but it could be that you've missed out something like bugfix support for your particular motherboard/chipset configuration.

Also, at the cost of a bigger performance hit, it may be worth remounting the / partition with the sync option; this will help to prevent any filesystem corruption in the event of a crash, and help to keep /var/log/messages up to date with “I'm about to crash”-type errors

Hello,

I looked in to the bug fix / suppor for the server hardware and it seems ok ... the server is actually about 1-2 years old. I will definitly look into getting the / partition remounted with sync.

Thanks for all the responses so far.

Michael.