dolf@nikhefh.nikhef.nl (Dolf Starreveld) (02/01/90)
For the last couple of weeks our news operations have been hampered by a
shortage of inodes on the /var/spool/news partition. The devices will be
about 70% full, but cannot allocate space for any more files, leaving us
with a loss of about 40 megabytes.
Now, the obvious solution is newfs on the partition, using parameters to
deliver about twice as many inodes, since inodes are relatively cheap.
The problem is, using the sun newfs ( or mkfs for that matter ), I cannot
realise more than 2048 inodes per cylinder group, and cannot increase the
number of cylinder groups, since mkfs will only accept values as multiples
of sixteen.
The following is default:
mkfs /dev/rxd0e 301500 67 15 8192 1024 16 10 60 2048 t 0 -1
/dev/rxd0e: 301500 sectors in 300 cylinders of 15 tracks, 67 sectors
154.4Mb in 19 cyl groups (16 c/g, 8.23Mb/g, 2048 i/g)
Changing the parameter 'bytes per inode' should have the same effect. As
default this has the value 2048, and reducing it to 1024 or 512
bytes/inode has no effect on the number of inodes per cylinder group.
I used a block size of 4K once and newfs/mkfs configured twice as many
i-nodes (because I got twice as many cylinder groups). So the file system
was how I liked it, but our machine did not like it! After I mounted this
new file system the system panicked:
Jan 22 14:16:15 carol vmunix: panic: segmap_unlock
Why is this? Is this a SUN kernel bug? Can anybody offer a better/other
solution to increase the number of i-nodes on my file system?
Since I had to temporarily shut down news facilities, please respond by
e-mail also.
Regards.
Dolf Starreveld Phone: +31 20 592 5056/5022 (FAX: 5155), TELEX: 10262 HEF NL
EMAIL: dolf@fwi.uva.nl (dolf%fwi.uva.nl@hp4nl.nluug.nl)
SNAIL: Dept. of Math. and Computing Science, University of Amsterdam,
Kruislaan 409, NL-1098 SJ Amsterdam, The Netherlands