From: [email protected] [mailto:[email protected]] 
On Behalf Of Nigel Kukard
Sent: 31 October 2011 06:06
To: [email protected]
Subject: Re: [policyd-users] Policyd Failed

On 10/25/11 08:19, Ian Mordey wrote:
Yes I use logrotate for cbpolicyd.log and syslog.. It looks like this caused 
the problem due to policyd restarting while syslog was restarting and being 
unable to talk to syslog. Is there any option to retry after this type of 
failure?

Cheers

From: [email protected]<mailto:[email protected]> 
[mailto:[email protected]] On Behalf Of Nigel Kukard
Sent: 24 October 2011 17:22
To: [email protected]<mailto:[email protected]>
Subject: Re: [policyd-users] Policyd Failed

On 10/24/11 08:30, Ian Mordey wrote:
Hi there
We’ve been running policyd for a few months with no problems however it looks 
like we had a failure last night. I’ve seen this in the logs:

[2011/10/24-03:48:02 - 28715] [CORE] NOTICE: 2011/10/24-03:48:02 Server closing!
[2011/10/24-03:48:02 - 28715] [CORE] ERROR: 2011/10/24-03:48:02 HUP'ing server

Then the log rotated and this in the next log:
[2011/10/24-03:48:03 - 28715] [CBPOLICYD] NOTICE: Policyd v2 / Cluebringer - 
v2.0.11RC1
[2011/10/24-03:48:03 - 28715] [CBPOLICYD] NOTICE: Initializing system modules.
[2011/10/24-03:48:03 - 28715] [CBPOLICYD] NOTICE: System modules initialized.
[2011/10/24-03:48:03 - 28715] [CBPOLICYD] NOTICE: Module load started...
[2011/10/24-03:48:03 - 28715] [CORE] NOTICE:   => AccessControl: enabled
[2011/10/24-03:48:03 - 28715] [CORE] NOTICE:   => Quotas: enabled
[2011/10/24-03:48:03 - 28715] [CORE] NOTICE:   => Protocol(Postfix): enabled
[2011/10/24-03:48:03 - 28715] [CBPOLICYD] NOTICE: Module load done.
no connection to syslog available
                - /dev/log is not a socket at /usr/sbin/cbpolicyd line 274

My config for syslogging has:
log_mail=mail@syslog:unix


Any ideas what caused this?  We have two servers running policyd and only one 
failed.

Just out of curiosity, do you have any log rotation going for cbpolicyd 
logfiles?

-N

From what it looks like above the server got a kill -HUP , but seeing as most 
of policyd config is in the DB it doesn't support being killed by HUP to reload 
config.

-N

So my logrotate config for cbpolicyd looks like this:
/var/log/cbpolicyd.log {
   missingok
    daily
    rotate 7
    create
    compress
    start 0
    postrotate
                /bin/kill -HUP `cat /var/run/cbpolicyd.pid 2> /dev/null` 2> 
/dev/null || true
    endscript
}

Are you saying I don’t need the kill –HUP? Would it be advisable to remove this?

Cheers
Ian
_______________________________________________
Users mailing list
[email protected]
http://lists.policyd.org/mailman/listinfo/users

Reply via email to