[Freeipa-users] named exits

[Petr Spacek]

On 02/21/2012 07:17 PM, Jimmy wrote:
> I have checked and there is no cron job scheduled to do anything with named.
>
> On Tue, Feb 21, 2012 at 12:04 PM, Jimmy <g17jimmy at gmail.com
> <mailto:g17jimmy at gmail.com>> wrote:
>
>     This has happened a couple times in the past few weeks and I thought
>     it was an admin error before, but looking at /var/log/messages the
>     named daemon does something every evening after 10pm local time(3am
>     UTC.) As you see in this log clip DNS reloads at 03:34UTC on 2/19-
>
>     Feb 19 03:34:01 csp-idm named[20761]: received control channel
>     command 'reload'
>     Feb 19 03:34:01 csp-idm named[20761]: loading configuration from
>     '/etc/named.conf'
>
>     But then on 2/20 the reload doesn't happen, it exits.
>
>     Feb 20 03:48:01 csp-idm systemd[1]: named.service: control process
>     exited, code=exited status=1
>
>     I don't see any reference to named in other log files. Where else
>     should I look to debug this issue?
>     Thanks

I suspect log rotating system from reloading BIND. Usual behaviour is:

1) rename log file
2) restart/reload daemon (to force closing old and opening new log file)

Do you have some log rotating system installed? If so, please look into 
it's configuration to confirm this suspicion.


Does BIND crash only after reload command is received? (Or at "usual" 
reload time?)

It will be very handy, if you can provide more details from log or 
coredump/files collected with ABRT.

Is you DNS server under heavy load? Or it does nothing at reload/crash time?

-- 
Best regards,
Petr Spacek