This will let you dump the Lustre log from a kernel vmcore:
https://github.com/Xyratex/crash-tools/tree/master/cr-ext
This is an extended backtrace plugin/extension for crash that decodes the parameters in a nice way, and maybe works:
https://wiki.hpdd.intel.com/display/~jhammond/XBT+-+extended+backtrace+extension+for+crash
Here’s how to load the extension:
sh> mod -S
MODULE NAME SIZE OBJECT FILE
MODULE NAME SIZE OBJECT FILE
ffffffffa000f460 dm_mod 84337 /lib/modules/2.6.32-431.20.3.el6_lustre.g5a7c614.x86_64/kernel/drivers/md/dm-mod.ko
crash> extend /root/lustre-ext.so
/root/lustre-ext.so: shared object loaded
crash> lustre -l /tmp/foo.dl
lustre_walk_cpus(0, 5, 1)
cmd: p (*cfs_trace_data[0])[0].tcd.tcd_cur_pages
p (*cfs_trace_data[0])[0].tcd.tcd_pages.next
…
lustre: Dumped 355 total debug pages from 2 CPUs to /tmp/foo.dl
I am getting this error while getting lustre logs using
crash> lustre -l lustre.bin
lustre_walk_cpus(0, 5, 1)
cmd: p (*cfs_trace_data[0])[0].tcd.tcd_cur_pages
p (*cfs_trace_data[0])[0].tcd.tcd_pages.next
lustre: gdb request failed: “p (*cfs_trace_data[0])[0].tcd.tcd_cur_pages”
Is it possible that gdb isn’t installed? I think the crash tool relies on that to dump the data. What version of Red Hat/Centos are you running?