Postfix 20051103 regular snapshot
Wietse Venema
connection caching and more detailed delay logging. These are for
people who want to squeeze more peformance out of Postfix.
Victor's change in the connection caching strategy solves a problem
while delivering mail to sites with multiple MX hosts of which some
are slow. This change causes Postfix to make better use of the
connections to the faster MX hosts. See RELEASE_NOTES for the fine
details of this solution.
The more detailed delay logging can help to locate performance
bottle necks better. It shows how much time is lost in the queue
manager, how much is lost making connections, and how much goes
into actual mail delivery.
The new logfile record looks like this:
Nov 3 16:04:31 bristle postfix/smtp[30840]: 19B6B2900FE:
to=<wie...@test.example.com>, orig_to=<wietse@test>,
relay=mail.example.com[1.2.3.4], conn_use=2, delay=0,
delays=0/0.01/0.05/0.1, dsn=2.0.0, status=sent (250 2.0.0 Ok)
The meaning of the "delays=a/b/c/d" fields is as follows: a=time
before queue manager, b=time in queue manager, c=connection setup
time, d=message transmission. Numbers smaller than 0.01 seconds
are truncated to 0, to reduce the noise level in the logfile.
The following two new logfile fields may or may not be present:
orig_to This is omitted when the address did not change.
conn_use This is omitted when a connection is used once.
The connection use count logging is needed because Postfix may now
reuse the same connection hundreds of times. This may trigger bugs
in servers that have a memory leak or some other resource leak.
Errors that correlate with large reuse counts can be remediated by
less agressive connection caching.
Wietse
Pascal Maes
Hello,
Should it be possible to provide a bounce message instead of the message
which is hard coded in bounce_notify_util.c ?
Regards,
--
-- Pascal --
Jussi Silvennoinen
>
>
> Should it be possible to provide a bounce message instead of the message
> which is hard coded in bounce_notify_util.c ?
Only by modifying the source.
--
Jussi