solved: what to do if you do a bad overview size estimate initially

John Kozubik john_kozubik at hotmail.com
Mon Mar 13 05:29:53 UTC 2000


Obviously, if you mistake your spool/overview size ratio (as I did) and need 
to solve this problem, the _correct_ course of action is to expand /var or 
reload the OS with a larger /var or otherwise address the root of the 
problem, which is incorrect provisioning of disk space.

However, this is easier said than done - especially on a production server - 
you may need to buy yourself some time before the disks fill up.

Here is what I did - nothing too complicated, but I didn't know it 
initially, and it may help others.

1. set an expiration date for all articles in 'expire.ctl' that is very 
draconian.  For example:

*:A:1:5:never

(5 day lifetime for all articles in this case)

2. execute the command:

/usr/local/news/bin/news.daily notdaily flags='-v1 -n -N'

(you may not need the -n)

Explanation:

news.daily will run expire and clean everything up - the problem is, if your 
storage mechanism is self-expiring, then the expire command will not pay 
attention to the 5-day lifetime specified in expire.ctl - however, the -N 
switch tells it to obey it.  the 'notdaily' switch on 'news.daily' tells it 
to skip the daily-only logging stuff ... the 'flags=' switch is for 
specifying flags on the expire command - again, we specify -N, and -v1 
because that is what news.daily would have specified by default, and we also 
do a -n in case inn is not running when we run this (as in my case, it 
crashed because I ran out of /var)

This procedure cleaned out my overview nicely, and although it is not the 
_correct_ fix to the problem, it buys me time indefinitely (I can run this 
once in a while in addition to my nightly news.daily run that does _not_ 
follow expire.ctl)

-john
______________________________________________________
Get Your Private, Free Email at http://www.hotmail.com




More information about the inn-workers mailing list