Sendmail cant deliver DSN's to postmaster

[les]

Hi,

On all my scalix installs (10+) i am using sendmail with spamass-milter as per the standard setup instructions.
Running on CentOS 4 mainly but also FC5.
And running various versions i.e. 10.0.2 11.0.0 11.0.1 and 11.0.2

I've seen a few posts on the forums, read them, but found no hard and fast answers or resolutions.

Basically....

When a message is classed as spam and is beyond the level you define in /etc/sysconfig/spamass-milter (-r) a DSN is sent to the sender.
I actually would prefer not to do that (like amavis can be told to do so) but with spamass-milter you dont get a choice.

Now when the DSN cant be sent to the remote user (usually because its a dud address which is why i'd rather not send one in the first place) it is then sent to postmaster.

However this delivery fails and you get a savemail panic log entry.
These servers are all default setup, so postmaster in /etc/aliases goes to root.

Ive googled around without much help and have verified that i can send to postmaster with the sendmail checks
sendmail -bv MAILER-DAEMON and sendmail -bv postmaster

I can also send to postmaster as root.

i supsect that sending to postmaster is not the real problem. However i will say that i do get some MAILER-DAEMON messages appear in my administrator account (redirected from local root).

My main concern is that the savemail panic causes the message in /var/spool/clientmqueue to be left behind, but renamed with an uppercase Q. thus if you do a mailq -Ac you never see it. Go into the directory and you'll have a bunch of old messages.

I guess i could just do a find or tmpwatch to clean up that directory of any Q* files older than x days but i'd rather know why this occurs and how to properly solve it instead of just hacking a workaround.

Can anybody shed some light on these savemail panics and non delivery of DSN's to postmaster that fail and leave dud messages in /var/spool/clientmqueue?

TIA