I had an issue which the vendor support needed messages from the messages log file.
There of course are messages which are frequent and could hide the needed debug messages.
I just needed to filter them out.
cat messages | grep -v "message not wanted" | grep "message repeated" > textfile
Smaller and more succinct detail for the vendor support people.
Legibility could be an issue for some so if you want to be nice; run the file through a conversion:
unix2dos <file> <outputfile>
A trouble call reported an old server running Solaris 10 with a full root partition.
Looking around I noticed the FMD log files were quite large and the active file was growing. To ease the space condition, I deleted the rotated files (ie .0, .1, etc.) and started looking into what was the problem.
The Solaris Fault Management Facility was created to provide a self-healing capability. It through the fmd daemon monitors various aspects of system health and as in this case, logged many messages for system issues.
The first obvious check was to use the fmadm faulty command to see if anything was flagged as faulty. In this situation; there was a bad dimm.
This wasn’t enough to fill log files so I had a look at /var/fm/fmd file and it had several entries for a processor.
The fmstat command which will report statistics logged by fmd and it’s modules confirmed the log activity.
Since it was an old server with no warranty; the hardware people were notified to look at the server and retire it if they couldn’t repair it.
Though Solaris is slowly disappearing(at least at my current company), there are still legacy machines which will remain and sometimes you have to debug them especially over network speed. Solaris has a legacy of having problems with auto-negoting. Sometimes you may think you are running full-duplex and you are not.
If you have access to the network switch, it’s easy to check. If you don’t you have to rely on solaris commands.
The NDD command will give you configuration information on the interface.
For example:
# ndd /dev/bge0 link_duplex
2
# ndd /dev/bge0 link_autoneg
1
# ndd /dev/bge0 link_status
1
# ndd /dev/bge0 link_speed
1000
Here duplex is full with auto-negotiating off and the link is down.
NDD can do much more but I only needed to check the interface settings.
techdork 