Martijn Brinkers (List) wrote:
It's sending now. But I still get lock ups for 10 minutes at a time. It
is
apparently locking up on a bad server, then trying the next server in the
list, etc. But the world halts 10 minutes at a time, and James does
nothing
at all outbound during that time.
I can see if I got a bunch of spam bounces to bad servers, with perhaps 3
or
4 mail servers listed for the domain, I could be hung up 30-40 minutes per
single email. This can't be right, can it?
Questions:
1) Is it working as designed that when a server is down, the outbound
processing halts until a 10-minute timeout?
2) Where is that 10 minute timeout specified? I can't find it anywhere in
the config.
It seemed that your server was not able to connect to *.mail.com or
*.outblaze.com and that the connection had to timeout every time it tried to
connect. On Linux the default timeout is, I think, 180 sec (=3 min).
mail.com has two mx records which can be resolved to 6 IP addresses. Worse
case scenario would be that all 6 of the SMTP servers timeout which will
take 6 * 3 = 18 min. James remote delivery mailet tries all SMTP servers one
after the other until all of them are tried once. This can take 6 * 3 = 18
min / mail. During this 18 min. connection attempt an outgoing connection is
taken and can therefore not be used to send mail to other domains. A
possible solution would be for James to try just one SMTP server and if that
fails skip this email and try another mail from the spool. Next time it
should try the next SMTP server etc. This keeps sending mail to other
domains without much interruption.
Martijn Brinkers
You may be interested in a patch I wrote:
http://issues.apache.org/jira/browse/JAMES-358
Unfortunately the patch has not been accepted.
I still think that it should be applied at least as an option.
Stefano
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
