Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!uwm.edu!cs.utexas.edu!usc!snorkelwacker!mit-eddie!uw-beaver!ubc-cs!eric!zaphod.mpr.ca!parker
From: parker@zaphod.mpr.ca (Ross Parker)
Newsgroups: comp.unix.ultrix
Subject: Re: what are gnodes?
Keywords: gnodes inodes
Message-ID: <2049@eric.mpr.ca>
Date: 2 Feb 90 23:30:23 GMT
References: <2048@eric.mpr.ca> <2915@decuac.DEC.COM> <821@larry.sal.wisc.edu>
Sender: news@eric.mpr.ca
Reply-To: parker@zaphod.mpr.ca (Ross Parker)
Organization: Microtel Pacific Research Ltd. (MPR)
Lines: 28

> The problem may actually be that you're running into the *system* limit
> on gnodes, rather than the individual filesystem's inode limit (depends
> on where, and under what circumstances the error message is appearing).
> 

I should probably expand on this...

What I've called here the 'system' limit is implying that you're running
out of in-core gnodes... i.e. you have more open files than the system
can currently handle.

The filesystem limit defines how many files can exist on the disk. Both
limits may need to be adjusted. As the problem is happening in your 'news'
partition, it's quite conceivable that your are running into the hard
filesystem limit and will need to rebuild your disk (ugh!).

For your sake, I hope you're just running out of in-core gnodes. On
our systems at least, popping in a new kernel is a much easier procedure!
BTW - this is the only case that I personally have seen the 'out of gnodes'
message - I seem to recall in the distant past that when a filesystem
ran out of physical inodes, the message was quite different, but then, that
was aeons ago, and the messages have quite likely changed.

Just do a 'df -i' to check your physical inode allocation.


Ross Parker				parker@mpre.mpr.ca
(604)293-5495				uunet!ubc-cs!mpre!parker