Hi everybody,
I happily use Cyrus IMAPD for quite some time now. Today I had trouble
for the first time: I saw quite some lmtp processes eating all the CPU
(the host had a load of 8.3 at that time - normally it's about 0 - 0.15
as it's still not used as planned). Postfix tried to deliver one more
message to Cyrus (using lmtp over tcp). After restarting cyrus and
requing that message I have one lmtp:
23619 cyrus 25 0 32304 2940 30m R 99.2 0.3 10:18.20 lmtpd
It seems that it will stay. The message contains more headers than text
- I cannot view it very well as I only have the Postfix queue file
(which is not plain text), but I'm quite sure that it contains all the
users of my university's Novell Groupwise server in the To field and
was created using Groupwise (it is an administrative notification that
the server will be shut down), the message is about 50k in size (header
+ a very, very short plain text body as far as I see.
lmtpd keeps eating CPU even after postfix drops the connection due to
timeout - after a flush of the queue I have 2 lmtpd processes eating as
much CPU as they get.
I use cyrus-imapd-2.2.10, postfix 2.2 and a sieve script created by
Horde Ingo (I could mail it via PM if that helps, same applies to the
queue file).
Can this be a sieve script issue? A bug? A really bad message that
crashes lmtpd or postfix lmtp? Should I just delete the message and
ignore it or try to dig deeper? Any hints for the latter?
Thanx,
Baltasar
----
Cyrus Home Page: http://asg.web.cmu.edu/cyrus
Cyrus Wiki/FAQ: http://cyruswiki.andrew.cmu.edu
List Archives/Info: http://asg.web.cmu.edu/cyrus/mailing-list.html
