From: Vladislav B. <vs...@vl...> - 2009-12-04 12:06:51
|
Kristof Van Doorsselaere, on 12/04/2009 11:32 AM wrote: > Vlad, > > Attached file should be a full log (as found in kern.log) > > starting from: > > scst start untill the crash Yes, this is what I need. It looks like a weird memory corruption somewhere. Could you enable additional logging by: # echo "add scsi" >/proc/scsi_tgt/trace_level # echo "add entryexit" >/proc/scsi_tgt/iscsi/trace_level # echo "add debug" >/proc/scsi_tgt/iscsi/trace_level Then reproduce the crash, then send me the same logs? The amount of logs can be huge, so you may need to rebuild your kernel to increase CONFIG_LOG_BUF_SHIFT up to the maximum possible value (20?) and put your syslog daemon in the async mode. On RedHat systems it can be done by adding '-' symbol before path in the corresponding line in /etc/syslog.conf. And don't forget to compress the logs by bzip2 -9 before send ;) Thanks, Vlad |