Heartbeat fails to start after Dapper upgrade

[Tore Anderson]

Binary package hint: heartbeat-2

Dapper appears to mount /var/run as a tmpfs, which precludes Heartbeat from working correctly, as it
needs the directories /var/run/heartbeat/ccm and /var/run/heartbeat/crm, which is overmounted. So it fails:

Jul 5 10:28:59 hiro heartbeat: [5275]: ERROR: socket_wait_conn_new: trying to create in /var/run/heartbeat/register bind:: No such file or directory
Jul 5 10:28:59 hiro heartbeat: [5275]: debug: Cannot open registration socket at /var/run/heartbeat/register
Jul 5 10:28:59 hiro heartbeat: [5279]: CRIT: Emergency Shutdown: Master Control process died.
Jul 5 10:28:59 hiro heartbeat: [5279]: CRIT: Killing pid 5275 with SIGTERM
Jul 5 10:28:59 hiro heartbeat: [5279]: CRIT: Killing pid 5280 with SIGTERM
Jul 5 10:28:59 hiro heartbeat: [5279]: CRIT: Killing pid 5281 with SIGTERM
Jul 5 10:28:59 hiro heartbeat: [5279]: CRIT: Killing pid 5282 with SIGTERM
Jul 5 10:28:59 hiro heartbeat: [5279]: CRIT: Killing pid 5283 with SIGTERM
Jul 5 10:28:59 hiro heartbeat: [5279]: CRIT: Killing pid 5284 with SIGTERM
Jul 5 10:28:59 hiro heartbeat: [5279]: CRIT: Killing pid 5285 with SIGTERM
Jul 5 10:28:59 hiro heartbeat: [5279]: CRIT: Emergency Shutdown(MCP dead): Killing ourselves.

From heartbeat-2.postinst:

        for i in /var/lib/heartbeat/ccm /var/lib/heartbeat/crm \
                        /var/run/heartbeat/ccm /var/run/heartbeat/crm; do
                chmod 750 $i
                chown hacluster $i
                chgrp haclient $i
        done

Adding similar code to /etc/init.d/heartbeat to create and chmod/own the missing directories on start alleviates the problem.

Tore

[Daniel Pittman (daniel-rimspace)]

I can confirm this problem: we have exactly the same situation. To remedy it I modified the system to create the relevant directories prior to starting heartbeat, which works.

[Jordan Curzon (curzonj)]

The patch below solved it for me.

[Jordan Curzon (curzonj)]

The patch I posted before actually doesn't work. I hadn't restarted since I had fixed the problem and that is why it seemed to work. I have tested this one and it does work.

[blufox (2500-pradeep-chail)]

Hi all, here is my fix.
Works for sure.
I tried it and it is working perfectly fine.

[Alvaro Figueroa (fede2)]

Could this be fixed in feisty?

I use heartbeat on a number or applications in quite a lot of servers. This bug is quite important for us who use heartbeat.

Please let me know if I can do anything to help. Thanks.

[Barry deFreese (bddebian)]

Do any of you happen to know if this is still an issue with 2.0.8-2 that is in Gutsy? Thanks.

[Emmet Hikory (persia)]

Thanks for your work on this issue. As previously noted, please check to see if the fix also applies to gutsy. I have added the patch tag to draw the attention of packagers. I have unsubscribed ubuntu-universe-sponsors, as this does not represent a request of sponsorship of a new revision (patch is not a debdiff). If a debdiff is created to address this in gutsy, please resubscribe ubuntu-universe-sponsors to request upload.

[Launchpad Janitor (janitor)]

[Expired for heartbeat-2 (Ubuntu) because there has been no activity for 60 days.]