On Fri, Sep 14, 2001 at 10:25:28AM -0700, Darren Dunham wrote:
> After having some problems with some clients in 4.0, we've recompiled
> and installed 3.1.2.
>
> We're running it in debug mode and these messages are popping up
> every once in a while..
>
> Sep 13 23:40:47.109 [22366] new child for connection
> Sep 13 23:40:47.109
> Sep 13 23:40:47.116 [3852] forked() for new connection; pid=22366
> Sep 13 23:40:47.116
> Sep 13 23:40:47.597 [22366] file type of /var/mail/welcome.lock is S_IFLNK or more
>than 1 link
> Sep 13 23:40:47.597
> Sep 13 23:40:47.598 [22366] welcome at dbapp.company.com (xx.xx.xx.xx): -ERR
>[SYS/TEMP] maillock error 'Other lock error' (5) on '/var/mail/welcome': Bad file
>number (9)
> Sep 13 23:40:47.598
> Sep 13 23:40:57.598 [22366] exiting after Qpopper returned
> Sep 13 23:40:57.598
>
> What can cause such an error?
Sounds to me like a different locking scheme between the MTA and
qpopper since you rebuilt it. I think normally you would indeed not
want to see links on the lock file (though some locking schemes work
that way.) It's a good thing that popper uses some sanity checks on the
lock or you'd be getting seriously corrupted mail spools.
Which 4.0 were you running?
The first versions had compatibility problems with some clients, but
4.0.3 seems to have no compatibility problems with anything our
customers run, which includes not just Mac and Windows, but Linux,
PDAs, old systems like Windows 3.1, pretty much you name it. (Of
course someone will now pop up and give a whole list of clients they've
had compatibility troubles with. Such is life.)
Nothing is running on the box except for
> sendmail/procmail for MTA and delivery and qpopper for pickup. Users
> cannot log into the box, and I'm pretty certain that the user cannot be
> logging in twice. It's actually picked up by an application that runs
> on a single box, and it pops down mail from a series of mailboxes in
> order.
>
> The box is running Solaris 8, and all the filesystems are local (UFS).
> The filesystem is nearly empty, both for blocks and inodes.
> /dev/vx/dsk/mail 11212288 17952 11082214 1% /var/mail (blocks)
> /dev/vx/dsk/mail 427 1365333 0% /var/mail (inodes)
>
> Has anyone else seen these messages occur frequently? I can't reproduce
> them at will, but the application is bombing out when it does occur, and
> I get the above message in the debug file.
BTW, thank you for including a lot of relevant info in your post. Us
mailing list types appreciate having enough info to make a relevant
guess.
-- Clifton
--
Clifton Royston -- LavaNet Systems Architect -- [EMAIL PROTECTED]
WWJD? "JWRTFM!" - Scott Dorsey (kludge) "JWG" - Eddie Aikau
