On 7/18/2014 10:07 AM, Kevin Golding wrote:
Fair enough, although I would note that if, for example, I split my
main corpus into subsets (which is tempting at times) then some
messages would show in both the obsolete masscheck and the current ones.
Also if we're really using 2012 dated log files should I stop worrying
about keeping everything within the past 6-7 months?
We are NOT using those files to my knowledge. I moved all the old logs
to
/home/updatesd/svn/new-rule-score-gen/corpus/corpus-logs-pre-2014-07-18/
and will delete if all looks good at 10PM~ EDT with the
do-stable-update-with-scores job
One thing I will say is I've gradually moved my (working) masscheck
later and later in the day. I know originally it said 9am UTC but I
have to run mine after 10am UTC for some odd reason - perhaps that
could explain some of the problems?
I think the key point is do you see the logs you are submitting
listed as used? I didn't see your logs as being used with the
correct revision. Looking at http://rsync.spamassassin.org/, I don't
see your logs at all which means you haven't uploaded in quite a long
time. Can you check your rsync output easily to see if you are
uploading ok?
Otherwise, I think the last files we have from you are from 2012!
Yikes! I will check but I showed up as expected in the list you posted:
Checking corpus/usable-corpus-set1/spam-net-kpg-core.log for SVN
1609892...
# SVN revision: 1609892
And this was the output I got this morning:
rsync -qPcvz ham-kpg-core.log spam-kpg-core.log
[email protected]::corpus/
This rsync lacks old-style --compress due to its external zlib. Try -zz.
Continuing without compression.
I've been seeing myself on the ruleqa site and I am in the rsync
listing for today too, with my huge 32MB uncompressed file!
Ahh, you had uploaded some logs in 2012 under kgolding so that's what I
was looking for. Sorry for the false alarm.
Regards,
KAM
