We're getting reports of "554 SMTP service not available" for our local
network, but are unable to find similar messages in the mainlog. The
documentation on 'Log_level' suggests that this type of error would be
reported at level 4 but we're at the default level - 5.
The source code indicates that the DEBUG option needs to be compiled in to
see this class of error. Would it make sense to promote this kind of error
to the logs since it reflects operational problems rather than debugging
problems?
Would it be possible to have memory-based tuning data such as the maximum
number of SMTP connections since the last reset time?
We're running 1.73 on BSDI with
queue_only_load = 10
smtp_accept_max = 30
smtp_receive_timeout = 3m
Assuming that the above 554 error is due to overload,
would the following changes be a reasonable remedy?
We generate mainlog.gz files of 1.3Mb and I've never seen load factors
above 2 and have enough memory so that we don't swap.
queue_only_load =3
smtp_load_reserve = 3
smtp_reserve_nets = "198.133.237/24:\
199.4.191.0/24:\
207.149.192/22:\
199.164.194.0/24:\
207.149.192/22"
Thanks,
Ned Schumann
OlympusNet
--
*** Exim information can be found at
http://www.exim.org/ ***
