problem with php/lsws restart

[ts77]

Hi there,

my problem is only partly lsws-related but I guess you can help most in this case.

First my configuration, I'm running lsws with php-fcgi on one machine and the database-server on another one.
Two days ago my database-server went down and needed a reboot.
Lsws/PHP seems to have timed out while that was the case.

Still even as the db-server got up again it still returned the timeout.
Seems like the php-children have still been busy/hanging in some strange state.
I needed to stop lsws, manually kill the php-processes (lsws wasn't able to do this and was hanging too because of that) and restart lsws to get it running again.

Anyone ever seen such a problem? Any idea on how to work around it?

Thanks,

Thomas

[Test]

Originally Posted by ts77

H
I needed to stop lsws, manually kill the php-processes (lsws wasn't able to do this and was hanging too because of that) and restart lsws to get it running again.

the same problem,
Fast CGI instances are not terminated (sig TERM is a must) by litespeed after shutdown/restart.

[ts77]

Well, I can't say that it never works.
If the fcgi-processes are spawned by lsws *and* correctly working, then they are killed for me.
Its just that in the case mentioned above (the php-processes are hanging) they are not getting killed or at least not successfully killed.

[mistwang]

Thomas,

Do you have to kill PHP processes with "SIGKILL" or just "SIGTERM"?
LSWS always try to stop all fcgi processes stated by itself with "SIGTERM" before it quits. So, unless LSWS lost tracking to PID of those processes, or those processes does not respond to "SIGTERM", those processes should be killed.

I guess for some reason lsws lost tracking to those PID.

I will take a closer look on this issue.

[ts77]

a "killall php" was enough which is SIGTERM as far as I know.