My original problem description follows:
===========================
System: Sun Sparcstation 1 acting as NIS server to 55 more SUN stations
12MB RAM, monochrome screen, two Quantum 104 drives internal,
one 327M WREN-IV external drive, one 1.6G Seagate Elite I external
drive, Exabyte-8200 8mm tape drive, SUN QIC-150M 1/4" tape drive
I recently had a problem with my root (/) partition of my first internal
104M hard drive where several items were lost to 'lost+found'. After fsck
did its work, mail stopped functioning for the server. All other stations
in the net can mail each other and, sometimes, the server. I say
'sometimes' because some messages have been rejected by the server.
As per the SUN System and Adminstration manual, I performed the following
test on the NIS server:
/usr/lib/sendmail -v </dev/null root
The response was:
Trying 192.9.200.10... connected.
220 ch208a.ttu Sendmail 4.1/SMI-4.1 ready at Wed, 27 Nov 91 16:41:24 CST
>>HELO ch208a.ttu
553 ch208a.ttu host name configuration error
Trying 192.9.200.10... connected.
220 ch208a.ttu Sendmail 4.1/SMI-4.1 ready at Wed, 27 Nov 91 16:41:26 CST
>>HELO ch208a.ttu
... MANY, MANY of these until...
553 ch208a.ttu host name configuration error
makeconnection: no socket: Too many open files
Too many open files
Error contacting remote server ch208a
I know I must have lost something important when my drive messed up. I
realize that I could reinstall SUNOS and everything would be OK, but I can't
[Aafford to be down just now if I can help it.
Therefore, my question is: can anyone tell me how to regenerate mail access
on a system without bringing it down? I will summarize if necessary.
=======================
There was not a clear resolution to the problem. Several people replied
that I should check /etc/sendmail.cf, /var/spool/mail, /var/spool/mqueue,
etc. I did, and everything compared perfectly with a backup made well
before the trouble began.
A band-aid fix was found thanks to a local VMS system adminstrator who had
knowledge of Internet mail's workings. In /etc/sendmail.cf, I changed the
line which read
Dj$w.$m
to read
Dj$w
Apparently, somehow, the domain name of my network was being appended to the
end of the address twice, once through $m, and again through some unknown
means. After this change, at least local mail from my server to my server
(from 'cron' processes and the like, not to mention my outgoing mail to
users) began to function again. I strongly suspect something very crucial to
mail in NIS was lost when my root partition was damaged, but all other
functions (including mail in general on all other systems in the net)
worked fine.
My final solution is going to be to reinstall SUNOS on my NIS server after
all the students have gone home for the holiday break. I guess I will never
really know what happened, but a stopgap solution is better than none at all!
Thanks to the following people for responding:
Jack Stewart (jack@laguna.CCSF.Caltech.EDU)
Rodger ???? (catsmgr@snap.ua.oz.au)
Jeff Aldrich (aldrich@sunrise.stanford.edu)
Eckhard Rueggeberg (erueg@cfgauss.uni-math.gwdg.de)
Claude-Marie Sauve (claudie@CANR.Hydro.Qc.CA)
Mike Raffety (miker@sbcoc.com)
Chuck ???? (root@jupiter.cs.uga.edu) (by far the most informative reply)
Joel L. Seber | Dry humor is wasted around here.
SUN Workstation Laboratory Manager |
Center for Manufacturing Research | -Joel L. Seber
and Technology Utilization |
Tennessee Technological University | recursive, adj.
Cookeville, TN 38505 | See 'recursive'
|
jls2013@tntech.bitnet |
This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:06:21 CDT