Re: Limiting scheduler retries...

[David Potter <potter@csuite.ns.ca>]

Hi Andy... thanks for the reply. 

Although most of the entries indicate retries at an interval of _about_
1hr (unusual word to be using when talking about a computer process) there
are several instances where the message was retried in only a few minutes
and other examples from the logs show similar behaviour.

I've had a look through scheduler/diagnostics.c but I'm afraid I'm no
closer to unraveling the mystery of the 'retryat diagnostic'.

Another factor may be that we kill and restart zmailer processes as part
of our log rotation just after midnight. I'm suspicious that the retries
sequence is starting over every night...?


david potter

Version info...
ZMailer router (2.99.49p10-s6 #12: Sat Feb 21 13:44:45 AST 1998)


On Mon, 13 Jul 1998, Andy Poling wrote:

> On Mon, 13 Jul 1998, David Potter wrote:
> > We're regularly getting large numbers of local retries that are failing
> > because of quota problems.
> > 
> > We're loging to syslog with the following entry typical:
> > 
> > ===== Syslog ===========================
> > Jul 13 00:02:58 halifax mailbox[16493]: S.pe.pv42981: 
> > to=<az999@chebucto.ns.ca>, delay=25:50:31, xdelay=00:00:00, 
> > mailer=local, stat=retryat +3600 header write to "/var/mail/az999"
> -----------------------------^^^^^
> 
> Here is the reason: 3600 seconds is one hour. 
> 
> > failed
> > ========================================
> > 
> > Other entries for S.pe.pv42981 show the following retries...
> > 
> >         	Interval		
> > 		hh:mm
> > 12:02:58 AM		delay=25:50:31	
> > 01:58:42 AM	01:55	delay=27:46:15	
> > 02:03:00 AM	00:04	delay=27:50:33	
> > 03:02:59 AM	00:59	delay=28:50:32	
> > 04:03:02 AM	01:00	delay=29:50:35	
> > 05:03:05 AM	01:00	delay=30:50:38	
> > 06:03:01 AM	00:59	delay=31:50:34	
> > 07:07:59 AM	01:04	delay=32:55:32	
> > 08:52:59 AM	01:45	delay=34:40:32	
> > 09:23:38 AM	00:30	delay=35:11:11	
> > 11:43:08 AM	02:19	delay=37:30:41	
> > 01:26:24 PM	01:43	delay=39:13:57	
> > 
> > This does not seem to correspond to what we're hoping to get 
> > from scheduler.conf
> 
> The retryat diagnostic takes precedence.  What you probably want to do (what
> I did when faced with this problem) is to issue a different retryat
> diagnostic in the case of a quota failure... I think I used 8 hours for
> instance.
> 
> -Andy
> 
> Global Auctions
> http://www.globalauctions.com
>