[vps-mail] SA 2.63 upgrade issues -

[Abigail Marshall <abigail@xxxxxxxxxxxx>]

SA 2.63 beta is running so great on my VPS 1 that I had
forgotten that the vinstall was still in beta. I have not
experienced any memory problems on a very busy VPS1 C - I
promised Jonathan on the Iserver list that I'd post a few
caveats for the vinstall here (so it's available for others
to refer to in list archives). [Scott - feel free to copy
any info for any documentation or file you want to create at
perlcode.org]:

1. procmailrc file:

These lines should be in the file, and will be added by the
vinstall if you opt to overwrite the old procmailrc file:

DROPPRIVS=yes

:0fw
|/usr/local/bin/spamc -U /var/run/spamd.sock

When I used the vinstall, the program inadvertantly
overwrote custom lines I had in my procmail file, to invoke
post Spamassassin filtering. So while email was properly
tagged, it was not directed to appropriate spam files and
went directly to users instead.

This was MY fault, because I had inserted my custom filter
above this line:

 ## end spamassassin vinstall (do not remove these comments)

However, I'm not the only one capable of making mistakes, so
it is important to
  a) Save old procmailrc file
  b) Compare new & old after vinstall to make sure that any
     custom recipes are not

Note re options:

-s max_size
  Default Message Size:
  spamc defaults to 250K which means messages larger than
  250K will not be passed to spamd.  If you frequently
  receive spam with larger attachments, you might want to
  experiment with increasing this. Alternatively, if you
  seem to have memory or server load problems, you may want
  to set it smaller.



2. Bayes database:

With PREVIOUS versions of SA, I have had problems on
occasion with the Bayes database becoming corrupted if the
bayes_seen file grows too large. I assume this was because
of a memory or lock file issue associated with the sa-learn
process.

To avoid this, I set this option in the local.cf file:

bayes_expiry_max_db_size 110000

(The default is 150,000).

I have not had any problems with this setting. This still
leaves me with a fairly large Bayes database (5.5 Mg) and
Bayes is working very effectively.

Documentation available online here:
http://www.spamassassin.org/doc/sa-learn.html


-Abigail

======================================================================
This is <vps-mail@xxxxxxxxxxxx>       <http://www.perlcode.org/lists/>
Before posting a question, please search the archives (see above URL).