Fix the pid occupied check when use bookkeeper-daemon.sh start or stop #3113
Conversation
|
on which platforms did you test this ? |
|
Ci is not running for some reason |
|
perhaps: https://www.githubstatus.com/incidents/fpk08rxnqjz2 can you try to close and reopen the PR ? |
@eolivelli I test the case in on Linux and MacOS. On Linux the 'kill -0 $pid' and 'ps -p $pid' performance diff as the discussion (https://stackoverflow.com/questions/30958964/why-do-kill-0-pid-echo-and-ps-ppid-echo-sometimes-differ);On MacOs, the both exhibit the same result; |
hangc0276
requested review from
hangc0276,
dlg99,
eolivelli,
nicoloboschi,
shoothzj and
zymap
This comment was marked as duplicate.
This comment was marked as duplicate.
Sorry, something went wrong.
|
fix old workflow,please see #3455 for detail |
|
@Nicklee007 Could you please rebase the code? I will push on this forward. Thanks :) |
Nicklee007
force-pushed
the
fix-pid-occupied-check
branch
from
ccd5f7d
to
529d616
Compare
|
@StevenLuMT Currently we don't have shell test suite, and it's difficult to built. So I think manual testing is enough |
ok,LGTM |
apache#3113) Master Issue: apache#3112 ### Motivation Fix the failed pid occupied check. we'll fail when use bookkeeper-daemon.sh start or stop bookie, after the last time we exit the bookie direct kill or the bookie occurred non-normal exit, then the bin/bookkeeper-bookie.pid are retained, and the pid in the file is occupied by the thread in other progress. ### Changes Change the pid occupied check from 'kill -0 $pid' to 'ps -p $pid'. The both will return true when the pid is occupied by one progress, but 'kill -0 $pid' will return true and the 'ps -p $pid' will return false when the pid is occupied by one progress's sub thread. StackOverflow discussion: https://stackoverflow.com/questions/30958964/why-do-kill-0-pid-echo-and-ps-ppid-echo-sometimes-differ Co-authored-by: nicklixinyang <nicklixinyang@didiglobal.com>
Master Issue: #3112
Motivation
Fix the failed pid occupied check. we'll fail when use bookkeeper-daemon.sh start or stop bookie, after the last time we exit the bookie direct kill or the bookie occurred non-normal exit, then the bin/bookkeeper-bookie.pid are retained, and the pid in the file is occupied by the thread in other progress.
Changes
Change the pid occupied check from 'kill -0 $pid' to 'ps -p $pid'. The both will return true when the pid is occupied by one progress, but 'kill -0 $pid' will return true and the 'ps -p $pid' will return false when the pid is occupied by one progress's sub thread.
StackOverflow discussion:
https://stackoverflow.com/questions/30958964/why-do-kill-0-pid-echo-and-ps-ppid-echo-sometimes-differ