SUMMARY: Short of inodes in a Netra

From: Arnaldo Mandel (am@ime.usp.br)
Date: Tue Dec 16 1997 - 05:49:08 CST


Original question:

> Our Netra NFS was installed and tuned using default parameters.
> Suddenly we found ourselves short of inodes. To give some numbers:
>
> Total inodes about 720000, 100% occupied (given by df -o i)
> (I didn't save the actual output prior to emergency action)
>
> $ df -k /export
> Filesystem kbytes used avail capacity Mounted on
> /dev/md/dsk/d0 22582354 12826798 7497326 64% /export
>
> The configuration is 12 2.1 GB disks on a single raid 5 filesystem.
>
> Is there a way of growing the number of inodes? Notice that Sun
> insists on managing the Netra through its web interface, which takes
> over most of the usual flexibility on the grounds that this is a very
> specially tuned system.

Looks like the filesystem must be treated as a normal ufs
filesystems. That means, the answer is dump/newfs/restore.
(which means, forget the WWW interface)

Thanks to:

Casper Dik <casper@holland.Sun.COM>
Rick Reineman <rick@lunger.llnl.gov>
Kevin.Sheehan@uniq.com.au (Kevin Sheehan {Consulting Poster Child})
mrs@cadem.mc.xerox.com ("Michael Salehi x22725")
bismark@alta.Jpl.Nasa.Gov (Bismark Espinoza)
craig@tuna.progroup.com (Craig W. Shaver)

Casper also pointed out the likely culprit for the misconfiguration:
> (try mkfs -m to be sure; only w/ 2.5.1 patch 104742-01 the number reported
> is correct)

Indeed, there was a discrepancy between the reports of du, mkfs -m and
quota, and the patch helped.

Right now, I am waiting for the new 1.2 release, before newfs.

-- 
Arnaldo Mandel                        
Computer Science Dept.		  
Universidade de S\~{a}o Paulo, Brazil	  
am@ime.usp.br



This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:12:11 CDT