Hello readers,

I've googled around but I've failed to find  anything  that  resembles
what our mail gateway has suddenly started experiencing:  the  nightly
sa-learn --rebuild --force-expire stuck forever (shows with 'D'  state
in  'ps  auxw' -> uninterruptible IO wait) and using _huge_ amounts of
memory (600+ Mb). Can't strace it.

This  machine  has  been  rock  stable  for  quite  some  time  and no
significant change has been made lately.

Is  this  just  a  case of 'your Bayes DB is corrupt, delete it' or is
there something else to try first? 

Info:

- Linux RH 7.3 w/ 2.4.20 kernel and most patches
- Perl 5.6.1
- SA v2.54 (yes I know, old... but it's been so stable)
- Amavisd-new with global flat-file settings
- Postfix

The  sa-learn  --rebuild  --force-expire is run nightly at 4AM without
stopping amavisd-new (could that be the problem?). 

Our  local.cf  has  only  local  whitelist entries, a couple of simple
tune-ups probably not relevant, and:

bayes_expiry_scan_count 500000
bayes_expiry_use_scan_count 1

Any hints quite appreciated, even 'been there, zapped the whole DB'.

Greets,
_Alain_


-------------------------------------------------------
This SF.Net email sponsored by: ApacheCon 2003,
16-19 November in Las Vegas. Learn firsthand the latest
developments in Apache, PHP, Perl, XML, Java, MySQL,
WebDAV, and more! http://www.apachecon.com/
_______________________________________________
Spamassassin-talk mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/spamassassin-talk

Reply via email to