Bug 991473

Created attachment 785691 [details]
Proposed fix

Fix dnsmasq initscript to correctly identify the system dnsmasq instance.

In testing this from our end we cannot seem to reproduce this:

[root@testsystem-63vm ~]# uname -a
Linux testsystem-63vm 2.6.32-358.11.1.el6.x86_64 #1 SMP Wed May 15 10:48:38 EDT 2013 x86_64 x86_64 x86_64 GNU/Linux
[root@testsystem-63vm ~]# service libvirtd status
libvirtd is stopped
[root@testsystem-63vm ~]# service dnsmasq status
dnsmasq is stopped
[root@testsystem-63vm ~]# service dnsmasq start
Starting dnsmasq:                                          [  OK  ]
[root@testsystem-63vm ~]# service dnsmasq status
dnsmasq (pid  3134) is running...
[root@testsystem-63vm ~]# service dnsmasq stop
Shutting down dnsmasq:                                     [  OK  ]
[root@testsystem-63vm ~]# service libvirtd restart
Stopping libvirtd daemon:                                  [FAILED]
Starting libvirtd daemon:                                  [  OK  ]
[root@testsystem-63vm ~]# service dnsmasq status
dnsmasq is stopped
[root@testsystem-63vm ~]# 

The user has seen that the 'pidof' code in /etc/init.d/functions appears to grab the pid of the process from the process table if it can't find a file in /var/run.

There is no file in /var/run so it's (correctly) using the pid from the libvirt instance and from what the user has investigated in the code, it seems to back up what he's seeing. 

I feel that this could be a holdover from the following bug as they seem to be fairly closely related.

https://bugzilla.redhat.com/show_bug.cgi?id=850944

The user's version of the package matches the errata that was released for this bug shown here however:

http://rhn.redhat.com/errata/RHSA-2013-0277.html

Output from the user's system:
[root@test ~]# rpm --verify dnsmasq
.......T.  c /etc/dnsmasq.conf

[root@test ~]# rpm -q dnsmasq
dnsmasq-2.48-13.el6.x86_64

[root@test ~]# cat /etc/redhat-release
Red Hat Enterprise Linux Server release 6.4 (Santiago)

Finally, the customer's /var/lib/libvirt/dnsmasq has three files:

default.addnhosts
default.hostsfile
default.leases

All of these are empty. Otherwise, user is not using dnsmasq and all files from the package verify cleanly (no changes).

(In reply to Robert McSwain from comment #2)
> In testing this from our end we cannot seem to reproduce this:

You can reproduce this as follows:

[root@localhost ~]# uname -a
Linux localhost.localdomain 2.6.32-358.el6.x86_64 #1 SMP Tue Jan 29 11:47:41 EST 2013 x86_64 x86_64 x86_64 GNU/Linux
[root@localhost ~]# ps -C dnsmasq
  PID TTY          TIME CMD
[root@localhost ~]# service dnsmasq status
dnsmasq is stopped
[root@localhost ~]# service libvirtd status
libvirtd (pid  2450) is running...
[root@localhost ~]# virsh net-list
Name                 State      Autostart     Persistent
--------------------------------------------------

[root@localhost ~]# virsh net-start default
Network default started

[root@localhost ~]# virsh net-list
Name                 State      Autostart     Persistent
--------------------------------------------------
default              active     yes           yes

[root@localhost ~]# ps -C dnsmasq
  PID TTY          TIME CMD
 2785 ?        00:00:00 dnsmasq  -> *** started by libvirt ***
[root@localhost ~]# service dnsmasq status
dnsmasq (pid 2785) is running...  -> *** should be "dnsmasq is stopped" ***
[root@localhost ~]# service dnsmasq start
Starting dnsmasq:                                          [  OK  ]
[root@localhost ~]# service dnsmasq status
dnsmasq (pid  2826) is running...
[root@localhost ~]# service dnsmasq stop
Shutting down dnsmasq:                                     [  OK  ]
[root@localhost ~]# service dnsmasq status
dnsmasq (pid 2785) is running...  -> *** should be "dnsmasq is stopped" ***

> The user has seen that the 'pidof' code in /etc/init.d/functions appears to
> grab the pid of the process from the process table if it can't find a file
> in /var/run.
> 
> There is no file in /var/run so it's (correctly) using the pid from the
> libvirt instance and from what the user has investigated in the code, it
> seems to back up what he's seeing.

initscript uses function "status" from /etc/init.d/functions if
'service dnsmasq status' is called. 

> I feel that this could be a holdover from the following bug as they seem to
> be fairly closely related.
>
> https://bugzilla.redhat.com/show_bug.cgi?id=850944

This issue is different from what I've fixed in Bug #850944. It was unidentified
at that time. But it is related to process pids, pidfiles in that the initscript
is incorrectly identifying a process that was not started using it.

Hope this and the reproducer helps you. Anyway if you have a customer ticket
for this issue I'm recommending adding it to this Bug so there is a higher chance
of getting it fixed in the future RHEL-6 releases.

I've added the customer ticket to this issue. Thank you!

This request was evaluated by Red Hat Product Management for
inclusion in the current release of Red Hat Enterprise Linux.
Because the affected component is not scheduled to be updated
in the current release, Red Hat is unable to address this
request at this time.

Red Hat invites you to ask your support representative to
propose this request, if appropriate, in the next release of
Red Hat Enterprise Linux.

Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.

For information on the advisory, and where to find the updated
files, follow the link below.

If the solution does not work for you, open a new bug report.

http://rhn.redhat.com/errata/RHBA-2014-0757.html