[ClusterLabs Developers] bundle/docker: zombie process on resource stop

[Klaus Wenninger]

On 07/24/2017 12:37 PM, Valentin Vidic wrote:
> On Mon, Jul 24, 2017 at 12:19:20PM +0200, Klaus Wenninger wrote:
>> Wasn't aware of the different behavior of pacemaker_remoted as pidone.
>> If it is really spawned as pidone - which might be worth verifying -
>> it should indeed create a zombie catcher pid1.
> You are right about PID 1, it looks like this from the inside when running:
>
>   PID TTY      STAT   TIME COMMAND
>     1 ?        Ss     0:00 /usr/sbin/pacemaker_remoted
>    55 ?        Ss     0:00 /sbin/httpd -DSTATUS -f /etc/httpd/conf/httpd.conf -c PidFile /var/run//httpd.pid
>    57 ?        S      0:00  \_ /sbin/httpd -DSTATUS -f /etc/httpd/conf/httpd.conf -c PidFile /var/run//httpd.pid
>    58 ?        S      0:00  \_ /sbin/httpd -DSTATUS -f /etc/httpd/conf/httpd.conf -c PidFile /var/run//httpd.pid
>    61 ?        S      0:00  \_ /sbin/httpd -DSTATUS -f /etc/httpd/conf/httpd.conf -c PidFile /var/run//httpd.pid
>    64 ?        S      0:00  \_ /sbin/httpd -DSTATUS -f /etc/httpd/conf/httpd.conf -c PidFile /var/run//httpd.pid
>    65 ?        S      0:00  \_ /sbin/httpd -DSTATUS -f /etc/httpd/conf/httpd.conf -c PidFile /var/run//httpd.pid
>
> and from the outside:
>
>   PID TTY      STAT   TIME COMMAND
> 15738 ?        Ssl    9:12 /usr/sbin/dockerd -H fd://
> 15742 ?        Ssl    6:50  \_ containerd -l unix:///var/run/docker/libcontainerd/containerd.sock --metrics-interval=0 --start-timeout 2m --state-dir /var/run/docker/libcontainerd/containerd --shim container
> 21583 ?        Sl     0:00  |   \_ containerd-shim 6827b7ead8fc5c199d0b5c9f3cc2c37013147e6d96dda25eaacad7d7e2ca5031 /var/run/docker/libcontainerd/6827b7ead8fc5c199d0b5c9f3cc2c37013147e6d96dda25eaacad7d7e2ca5
> 21598 ?        Ss     0:00  |   |   \_ /usr/sbin/pacemaker_remoted
> 21770 ?        Ss     0:00  |   |       \_ /sbin/httpd -DSTATUS -f /etc/httpd/conf/httpd.conf -c PidFile /var/run//httpd.pid
> 21772 ?        S      0:00  |   |           \_ /sbin/httpd -DSTATUS -f /etc/httpd/conf/httpd.conf -c PidFile /var/run//httpd.pid
> 21773 ?        S      0:00  |   |           \_ /sbin/httpd -DSTATUS -f /etc/httpd/conf/httpd.conf -c PidFile /var/run//httpd.pid
> 21776 ?        S      0:00  |   |           \_ /sbin/httpd -DSTATUS -f /etc/httpd/conf/httpd.conf -c PidFile /var/run//httpd.pid
> 21778 ?        S      0:00  |   |           \_ /sbin/httpd -DSTATUS -f /etc/httpd/conf/httpd.conf -c PidFile /var/run//httpd.pid
> 21780 ?        S      0:00  |   |           \_ /sbin/httpd -DSTATUS -f /etc/httpd/conf/httpd.conf -c PidFile /var/run//httpd.pid
>

Interesting... Looking at the code I would expect 2 processes. Some
'pcmk-init' with pidone and
'pacemaker_remoted' that is spawned out as child.
So looks as if the spawning thing isn't working. Strange thing is that
the only criterion to do
that is that it checks for being pidone.
The code seems to be there since April 19th.

Klaus