[Spacewalk-list] rhn_check / yum pid not cleaned up

[Dhaussy Alexandre]

Hello,

I have some issues since i recently upgraded rhn clients.
It seems that since this the following patch, /var/run/yum.pid is not 
cleaned up by rhn_check.

https://github.com/spacewalkproject/spacewalk/commit/63010f56bae6a794ae6317bb9870edcc14518393#diff-5b3600e835a19c6134b21a5b36bb194d
https://bugzilla.redhat.com/show_bug.cgi?id=1361185

I checked on Centos 6 and 7, Centos 7 doesn't seem to be affected.

On Centos 7, it works as expected :

   [root at wikidt ~]$ rpm -q yum yum-rhn-plugin
   yum-3.4.3-150.el7.centos.noarch
   yum-rhn-plugin-2.7.7-1.el7.noarch
   [root at wikidt ~]$ rhn_check
   [root at wikidt ~]$ ll /var/run/yum.pid
   ls: cannot access /var/run/yum.pid: No such file or directory

But on Centos 6, yum's PID stays there :

   [root at albenga ~]$ rpm -q yum yum-rhn-plugin
   yum-3.2.29-73.el6.centos.noarch
   yum-rhn-plugin-2.7.7-1.el6.noarch
   [root at albenga ~]$ rhn_check
   [root at albenga ~]$ ll /var/run/yum.pid
   -rw-r--r-- 1 root root 5 Nov 13 17:53 /var/run/yum.pid

The problem reveals itself only under unfortunate coincidences :

- rhn_check runs and writes /var/run/yum.pid
- no yum commands are launched after that (it would clean the pid)
- a (persistent) process P spaws up with the same PID number as in 
/var/run/yum.pid
- then any yum command stalls waiting for the end of P

I have the issue on production servers with >300 apache processes with 
rotatelogs :

Another app is currently holding the yum lock; waiting for it to exit...
   The other application is: httpd
     Memory : 5.8 M RSS (1.7 GB VSZ)
     Started: Sat Nov 11 03:40:09 2017 - 58:41 ago
     State  : Sleeping, pid: 18792


Regards,
Alexandre.