history file is munged after hard reboot

Kim Moir Kim_Moir at ca.ibm.com
Tue Jul 22 19:40:34 UTC 2003


Hi 

KM>Thanks for all your  suggestions.

I think we can work it out, if it's important, provided me don't mind 
some clients redisplaying some previously-read articles.

First step would be to walk through the spool peering at Xref headers, and
see what sort of scale of duplication we're talking about.  I'm not 
comfortable with File::Find to do it off the top of my head, but if your 
spool is small, try

    find . -type f | xargs grep -h ^Xref: | sort | uniq -c | sort -n

in $patharticles (or some subdirectory thereof).  If you've got a grep 
that supports a max-count feature, add -m1 to the grep flags, and you can 
probably run this happily on a larger spool.

That will give you a sense of how many article numbers are duplicated (or 
more than duplicated, but hopefully not).

KM> Thanks.  We have had several ongoing intermittent hardware problems 
with this server which have caused additional issues with the indexes. 
Since we have a news to mail gateway for all the newsgroups, we've decided 
that people can read the archives and the problems with the indexes will 
have to be tolerated until we resolve the hardware issue.  Could this 
problem stem from the issue that we never expire messages?   I know it's a 
bad idea to configure the server to do this, however, this is how it was 
originally configured and it was requested that this setting remain.  For 
instance, one of our newgroups has over 60,000 messages.



More information about the inn-workers mailing list