This seems pretty important.  I thought I noticed I wasn't getting updates with 3.3.x, but never looked into it, and then ended up switching to trunk.

Is 3.3.x getting any updates?

From January first to April 13th I was running 3.3.1, running sa-update from cron daily, and got no changes, according to my backups:


# diff -r panic-2011-01-01/var/lib/spamassassin/3.003001 panic-2011-04-13/var/lib/spamassassin/3.003001
diff -r panic-2011-01-01/var/lib/spamassassin/3.003001/updates_spamassassin_org.cf panic-2011-04-13/var/lib/spamassassin/3.003001/updates_spamassassin_org.cf
1c1
< # UPDATE version 1052462
---
> # UPDATE version 1083704


For trunk (3.4.x), I got changes both of the last two days (presumably more):

# diff -r panic-2011-05-03/var/lib/spamassassin/3.004000 panic-2011-05-04/var/lib/spamassassin/3.004000 | wc -l
232

# diff -r panic-2011-05-02/var/lib/spamassassin/3.004000 panic-2011-05-03/var/lib/spamassassin/3.004000 | wc -l
410

It looks like I have backups from every day from 2011-01-01 to 2011-05-02.  Update versions:

2011-01-01 1052462
2011-03-20 1052462
2011-03-21 1083147
2011-03-22 1083377
2011-03-23 1083703
2011-03-24 1083703
2011-03-25 1083704
2011-05-02 1083704

And in all of those updates are identical except for the version number.

So I find it hard to believe that corpora starvation is the only problem.

(In reply to comment #3)
> It looks like I have backups from every day from 2011-01-01 to 2011-05-02. 
> Update versions:
> 
> 2011-01-01 1052462
> 2011-03-20 1052462
> 2011-03-21 1083147
> 2011-03-22 1083377
> 2011-03-23 1083703
> 2011-03-24 1083703
> 2011-03-25 1083704
> 2011-05-02 1083704
> 
> And in all of those updates are identical except for the version number.
> 
> So I find it hard to believe that corpora starvation is the only problem.

IIRC, late March Daryl turned off the auto-update mechanism because it caused some nasty surprises after it began working again after many months of being broken.  I proposed that we turn it back on, but with the last step that publishes the update temporarily disabled, so we could manually verify the diff of the old to new rule tarball before it is published.  They rejected that idea for reasons I disagree with, and I think we're still stuck.

I don't think there's a mass-check corpora starvation problem.

$ rsync --exclude '*~' -vaz "darxus@rsync.spamassassin.org::corpus/*-net-*.log" /home/darxus/sa/corp
$ cd ~/masscheckwork/weekly_mass_check/masses/
$ grep -l 1098073 ~/sa/corp/ham-net*.log | xargs ./log-grep-recent -m 72 > ham.log ; grep -l 1098073 ~/sa/corp/spam-net*.log | xargs ./log-grep-recent -m 2 > spam.log
$ wc -l ham.log spam.log
   199159 ham.log
   288030 spam.log

And it requires 150,000 of each, according to http://mail-archives.apache.org/mod_mbox/spamassassin-dev/201101.mbox/%3C4D41F44A.6030400@dostech.ca%3E

Hope it gets turned back on soon. Some of the changes to rules recent would be beneficial for our setup.

I'll mark this in for things to do this and next weekend.  Hopefully it materializes either one of those weekends.

I made some progress on cleaning up the update generation processes today.  I'll continue on tomorrow night.

Documentation?

sa-update for 3.3.2 works for me tested just now and got 1231867:

dns: 2.3.3.updates.spamassassin.org => 1231867

If the issue still exists, include the output of sa-update -D and we can re-open.  Otherwise considering this closed.