User pressure forced me to simply reboot the machine, whence I
could easily umount the disc and sort out the problem.
email@example.com: pointed out that I should also check for swap,
using 'swap -l'. So I tried:
# swap -l | grep c0t1d0s6 # check for a swap partition
# swap -l | grep export1 # check for a swap file
both of which reported no swap on my filesystem.
James.E.Coby.Jr@cdc.com: suggested I should also have used
I had in fact tried this (and also lsof, both with no joy), but omitted to
list it in my original message.
firstname.lastname@example.org: suggested I try using 'lockfs -h /export1'.
Unfortunately I had rebooted before this suggestion, so couldn't try it
(this time). However, beware this from the man page:
"Using lockfs to lock a file system is discouraged because
this requires extensive knowledge of SunOS internals to be
used effectively and correctly"
Next time it happens I will try the lockfs, but otherwise, still no
solution other than a reboot. 8-(
> Had this sort of problem a number of times and usually end up sorting
> it by rebooting the machine, which is obviously not good.
> Current machine is ultra2 running solaris7.
> As far as I can tell the file system (/export1) is not busy (evidence below),
> yet I still cannot umount it.
> # mount | grep export1
> /export1 on /dev/dsk/c0t1d0s6 quota/setuid/read/write/largefiles/logging
> on Fri Jan 22 19:53:02 1999
> # share | grep export1
> # fuser -cu /export1
> # lsof -V /export1
> lsof: no file system use located: /export1
> (lsof had previously identified a user csh and rpc.ttdbs processes, which
> I duly killed)
> # automount # in case automounter was confused
> # umount /export1
> umount: /export1 busy
> Seems to me that either:
> - the info recording that the filesystem is busy has got confused, or
> - there is a process using the filesystem that fuser, lsof have failed
> to identify?
This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:13:16 CDT