Weirdness with duplicate message-id and hash calculations
Christopher P. Lindsey
isc.org at spamfodder.com
Sun Jan 26 05:24:16 UTC 2003
Hi everyone,
I'm struggling with an issue involving supposedly duplicate Message-IDs
in messages going into the server.
I'm trying to feed a message in with a header like this:
Message-ID: <5BD18FBC01F1D211990600A0C9F495DC04FF9843 at ecs-ms1.ecsu.ctstateu.edu>
inews comes back and says that this is a duplicate. Sure enough,
grephistory says that too:
bash$ bin/grephistory -l '5BD18FBC01F1D211990600A0C9F495DC04FF9843 at ecs-ms1.ecsu.ctstateu.edu'
[1F95BCB4187B2AFD14A5CA732EBC5ECB] 1039977459~-~1039977458 @02003DFCCBF300F300000000000000000000@
But if I tell grephistory to just generate a hash, it doesn't match
that record at all:
bash$ bin/grephistory -h '5BD18FBC01F1D211990600A0C9F495DC04FF9843 at ecs-ms1.ecsu.ctstateu.edu'
[7411B9D7A2122BBAAE31E5A917D5F4D2]
And checking the articles themselves doesn't reveal any messages with
that Message-ID:
bash$ cd /var/news/spool/articles
bash$ grep --recursive '5BD18FBC01F1D211990600A0C9F495DC04FF9843 at ecs-ms1.ecsu.ctstateu.edu' .
bash$ echo $?
1
Now here's the really weird part. I have a quick little perl script
that translates the token into a path for ovdb. I'll feed it the
token returned by my original grephistory:
bash$ /usr/local/scripts/hashtofilename @02003DFCCBF300F300000000000000000000@
/var/spool/news/articles/time-00/fc/cb/00f3-3df3
Now let's get the Message-ID out of that message:
bash$ grep Message-ID: `/mnt/scratch/hashtofilename @02003DFCCBF300F300000000000000000000@`
Message-ID: <NGBBLKJNELBFHOMCKPMIAELGCJAA.tchessie at yahoo.com>
Prepare for the freakiness... If I do a grephistory on that message-id,
it supposedly doesn't exist. But if I tell it to give me a hash, it
gives me the one that was returned for that other Message-ID: when
I ran grephistory -l!
bash$ bin/grephistory 'NGBBLKJNELBFHOMCKPMIAELGCJAA.tchessie at yahoo.com'
/dev/null
bash$ bin/grephistory -h 'NGBBLKJNELBFHOMCKPMIAELGCJAA.tchessie at yahoo.com'
[1F95BCB4187B2AFD14A5CA732EBC5ECB]
So for some reason queries to the history database are getting the wrong
results, and I'm not sure why.
I don't know if this problem was there before or not, but I noticed
it after running 'makehistory -b -F -O -I' with a throttled server.
I'm running inn 2.3.4, using ovdb for my overview datbase. The
server is running RedHat 6.2 with BerkeleyDB 3.2.9a.
Does anyone have any ideas?
Thanks,
Chris
More information about the inn-workers
mailing list