[milters] Archive

Lists Index Date Thread Search

Article: 1050
From: Sebastian Hagedorn
Date: 2006-07-19 12:08:11 -0400
Subject: Experiences with caching and milter-ahead

Removal...........: milters-request@milter.info?subject=remove
More information..: http://www.milter.info/#Support

Content-Type: text/plain; charset=iso-8859-1; format=flowed
Content-Transfer-Encoding: quoted-printable
Content-Disposition: inline


we just started using milter-ahead on our servers. We receive several=20
hundreds of thousands of mails per day and have seen a few peaks of=20
extremely high CPU load, but for the time being it looks as though this=20
might work for us. Anyway, I wanted to ask how people deal with=20
milter-ahead's cache. I decided immediately that a negative cache TTL of a=20
week is much too long for our setup. I've changed that to one day, but I'm=20
considering even smaller values, e.g. one or two hours. I'm curious what=20
other people use.

One other thing: We use four servers for the milter in a load-balanced=20
cluster. On one of of these I had a corrupted cache, i.e. all updates=20
failed. We use RHEL 3 with its db4-4.1.25-8.1 RPM. db_verify shows:

db_verify /tmp/milter-ahead.db
db_verify: Page 2053: hash page has bad prev_pgno
db_verify: Page 514: hash page has bad prev_pgno
db_verify: Page 2054: hash page has bad prev_pgno
db_verify: Page 2036: non-empty page in unused hash bucket 2032
db_verify: Page 0: page 2053 encountered a second time on free list
db_verify: DB->verify: /tmp/milter-ahead.db: DB_VERIFY_BAD: Database=20
verification failed

I just deleted it, but I wonder if this is a common problem?

Cheers, Sebastian Hagedorn
Sebastian Hagedorn - Postmaster - RZKR-R1 (Geb=E4ude 52), Zimmer 18
Zentrum f=FCr angewandte Informatik - Universit=E4tsweiter Service RRZK
Universit=E4t zu K=F6ln / Cologne University - Tel. +49-221-478-5587
Skype: shagedorn
Content-Type: application/pgp-signature
Content-Transfer-Encoding: 7bit

Version: GnuPG v1.4.3 (Darwin)



Lists Index Date Thread Search