when kiiled parent process， all child process become orphan process

[liujinyang-highgo]

version
ppgool 4.6.2

description

start pgppool and running successfully.
then use "kiill -9 pid" kill parent process. parent process was killed and all its child process become orphan process.

[root@test1 bin]# ps -ef | grep pgpool
root 21487 1 4 13:04 ? 00:00:00 ./pgpool -f ../etc/pgpool.conf -D
root 21489 21487 0 13:04 ? 00:00:00 pgpool: wait for connection request
root 21490 21487 0 13:04 ? 00:00:00 pgpool: wait for connection request
root 21491 21487 0 13:04 ? 00:00:00 pgpool: wait for connection request
root 21492 21487 0 13:04 ? 00:00:00 pgpool: wait for connection request
root 21493 21487 0 13:04 ? 00:00:00 pgpool: wait for connection request
root 21495 21487 0 13:04 ? 00:00:00 pgpool: PCP: wait for connection request
root 21496 21487 0 13:04 ? 00:00:00 pgpool: worker process
root 21497 21487 0 13:04 ? 00:00:00 pgpool: health check process(0)
root 21498 21487 0 13:04 ? 00:00:00 pgpool: health check process(1)
root 21499 21487 0 13:04 ? 00:00:00 pgpool: health check process(2)
root 21514 15621 0 13:04 pts/2 00:00:00 grep --color=auto pgpool
[root@test1 bin]# kill -9 21487
[root@test1 bin]# ps -ef | grep pgpool
root 21489 1 0 13:04 ? 00:00:00 pgpool: wait for connection request
root 21490 1 0 13:04 ? 00:00:00 pgpool: wait for connection request
root 21491 1 0 13:04 ? 00:00:00 pgpool: wait for connection request
root 21492 1 0 13:04 ? 00:00:00 pgpool: wait for connection request
root 21493 1 0 13:04 ? 00:00:00 pgpool: wait for connection request
root 21495 1 0 13:04 ? 00:00:00 pgpool: PCP: wait for connection request
root 21496 1 0 13:04 ? 00:00:00 pgpool: worker process
root 21497 1 0 13:04 ? 00:00:00 pgpool: health check process(0)
root 21498 1 0 13:04 ? 00:00:00 pgpool: health check process(1)
root 21499 1 0 13:04 ? 00:00:00 pgpool: health check process(2)
root 21546 15621 0 13:04 pts/2 00:00:00 grep --color=auto pgpool

[pengbo0328]

Thank you for reporting this issue.

Setting child_life_time is an effective way to mitigate this issue. However, please note that this mechanism only applies to child processes that have handled at least one client connection.

If Pgpool-II is managed by systemd, the processes should be properly handled by systemd, so this issue is unlikely to occur in typical environments.