[30978] in Hotline Meeting
tree-killer became a pacifist
daemon@ATHENA.MIT.EDU (Operator)
Tue Nov 21 23:41:22 1995
From: root@MIT.EDU (Operator)
Date: Tue, 21 Nov 1995 23:41:18 -0500
To: hotline@MIT.EDU, op@MIT.EDU
Lou informed me this evening that users weren't able to print to Ceres
(which is served by Tree-Killer). The queue appeared (to lpq) to have
no entries, even though users were sending print jobs there. I found
out from folks at SIPB that all the other printers fed by the print
server Tree-Killer were acting the same way. Matt and I investigated.
It turned out Tree-Killer's /var partition had filled. This appeared
to be the cause of the failure. Since all the print queues were
appearing to be empty (and not printing), and therefor any queued
print jobs had already been dropped on the floor, we did:
lpc down all
lpc clean all
lpc up all
[There was an "lpc restart all" in there before we remembered to do
the "lpc up all".]
This reset the normal state of the print queues, and cleared out half
the disk usage, restoring Tree-Killer to normal operation.
Our new (since 1991...) model of print service associates many more
printers per server. As a result, we need bigger spool disks.
Putting the spool (or spool/printer) directory on a 1+ gig external
hard disk should give us enough room to last another few years. The
partition used currently has only about 100 Meg free for spool space,
but I'm sure we'll be getting a steady increase in the rate of
FrameMaker documents and other large documents with imbedded images.
These all consume a lot more disk space than traditional (text) jobs.
