ctdb (4.9.1) not starting

[AShehzad2]

Hi,

I have updated the ctdb to the latest version (ctdb-4.9.1-6.el7.x86_64) but it is NOT strating up since.
I get the following error message?

ctdbd[13172]: Starting event daemon /usr/libexec/ctdb/ctdb-eventd -P 13172 -S 14
ctdbd[13172]: Set runstate to INIT (1)
ctdbd[13172]: ctdb exiting with error: Failed to run init event
ctdbd[13172]:
ctdbd[13172]: CTDB daemon shutting down
ctdb-eventd[13174]: PID 13172 gone away, exiting

Any help would be appreciated?

Thanks.

[AShehzad2]

I managed to start it up now.
There are lot of config changes in ctdb 4.9.1.
So I migrated the old config to new style config and created couple of missing directories.
It is up now and samba has started up also.

The only issue, not sure it is by design that the Samba/ctdb VIP is not showing on the physical interface anymore.
But if I run "ctdb ip -v" then it shows that VIP is available and active.

But I am getting the following in the log file which seems like a loop. Any suggestions are welcome please. Thanks.

ctdb-recoverd[]: Trigger takeoverrun
2019/10/10 16:19:06.x ctdb-recoverd[x]: Takeover run starting
2019/10/10 16:19:06.x ctdbd[]: Takeover of IP x.x.x.x/24 on interface xxx
2019/10/10 16:19:06.x ctdb-recoverd[x]: Takeover run completed successfully
2019/10/10 16:19:07.x ctdb-recoverd[x]: Assigned IP x.x.x.x not on an interface
2019/10/10 16:19:07.x ctdb-recoverd[x]: Trigger takeoverrun
2019/10/10 16:19:07.x ctdb-recoverd[x]: Takeover run starting
2019/10/10 16:19:07.x ctdbd[]: Takeover of IP x.x.x.x/24 on interface xxx
2019/10/10 16:19:07.x ctdb-recoverd[]: Takeover run completed successf

[Thraex]

I ran into the same two issues you did and had to add the directories like /var/lib/ctdb/state. Only info I can find is at:

https://bugzilla.redhat.com/show_bug.cgi?id=1656777
https://bugs.centos.org/view.php?id=16411&nbn=4

I'll keep messing around and see if I can make any progress

[Thraex]

Alright I think I found a solution. I copied all of the legacy event scripts from the directory /usr/share/ctdb/events/legacy/ and put them into /etc/ctdb/events/legacy/. Then removed the ones I don't need like clamd, vsftpd, winbind. The ctdb log will show you if an event script fails or not. After removing the failing event scripts and starting up the ctdb service it was finally able to add the virtual ip address to the correct device and stop the takeover loop. Probably not the best way to do it, but for testing purposes it will at least get the service to work.

[Fubajaimen]

I'm having the same takeover loop problem (ctdb 4.9.1-6). CTDB will start, but public IPs are still not working. The log is full of takeover IP messages spewing every second. Only workaround at the moment has been to down rev ctdb. Still actively looking for a solution.

[Thraex]

Is it still doing that after copying the event scripts over? Every test system I had broke out of the takeover loop once I did that and restarted the service.

[pmcneil]

I had the same problem Centos 8.2. To FIX:

ln -s /usr/share/ctdb/events/legacy/10.interface.script /etc/ctdb/event/legacy/

OR

ctdb event script enable legacy 10.interface