Quoting Marilyn Davis (marilyn@???):
> Thank you Sander,
>
> > Could it perhaps be a problem with spamassassin's spamd not
> > accepting the connection from spamc?
> > I think it might well be.
> I don't think so. I just tested by killing spamd and sending a message.
> So, that makes it run fine. It just doesn't get checked for spam content.
> I don't know what the PANIC error is about. What broke the pipe?
AFAIK a broken pipe occurs if you try to write on a socket that isn't
connected anymore. So for instance, the initial connection to spamd
succeeded, but then spamd died / restarted or the connection got
terminated otherwise, and when you tried to write to the socket it
failed because there was nothing to write to.
> I don't think the messages that were temporarily rejected reached as
> far as the sa-exim code. There were no PANIC's.
There *WAS*. sa-exim PANIC'ed, and because of that, sa-exim didn't know
what to do with the message and told exim to temp-reject it.
> I don't know why the first one that made it past RCPT hit a PANIC, but
> maybe it's a clue of some sort.
sa-exim runs after the DATA section has been closed.
> >> ret=write(writefd[1],hl->text,strlen(hl->text));
> So. I still don't know.
I do.
I'm 99.9% sure it had to do with spamd / spamc not functioning correctly.
You see the line that caused the error is in fact a write() to a
filedescriptor. If you look up on where that filedescriptor is defined
you'll most certainly find it's a connection to spamd.
Take care,
Sander.
--
| You are not getting older, you're just getting closer to death.
| 1024D/08CEC94D - 34B3 3314 B146 E13C 70C8 9BDB D463 7E41 08CE C94D
