hijack.sh: exit_handler: collect dmesg throughout the whole test#1348
hijack.sh: exit_handler: collect dmesg throughout the whole test#1348ekurdybx wants to merge 1 commit intothesofproject:mainfrom
Conversation
ekurdybx
force-pushed
the
exit-handler-on-sigterm
branch
3 times, most recently
from
5704306 to
2339704
Compare
ekurdybx
changed the title
Start collecting dmesg logs in the beggining of the test and run collection in the background. Signed-off-by: Emilia Kurdybelska <emiliax.kurdybelska@intel.com>
ekurdybx
force-pushed
the
exit-handler-on-sigterm
branch
from
2339704 to
b7d2f35
Compare
ekurdybx
requested review from
a team,
golowanow,
lgirdwood and
marc-hb
as code owners
marc-hb
left a comment
marc-hb
left a comment
There was a problem hiding this comment.
Nice idea! I remember running journalctl -b -1 many times after a crash and reboot. Not always possible when it's a CI system that not everyone has access to.
However this is currently dropping the "empty logs" safety check and the sync. I think these can just stay more or less "as is" in the same place; why remove them?
We had many "green failures" situations in the past and one of the most infamous sof-test bugs caused logs to be entirely missing (#297). Then, sof-test was proudly claiming zero error found! Best way to make all the tests pass: kill the messenger :-)
| sudo bash -c "${mtraceCmd[*]} &" >& "$clogfile" | ||
| } | ||
|
|
||
| func_dmesg_collect() { |
There was a problem hiding this comment.
| func_dmesg_collect() { | |
| func_kmsg_collect() { |
The dmesg command has a number of issues compared to journalctl and I think (and hope) hasn't been used for a long time.
4 participants
Start collecting dmesg in the beggining of the test and run collection in the background. This will ensure having dmesg logs saved in case of github exiting the test with sigkill.