Thanks to the people who replied. Turns out that we hit the
situation Alan describes, where advfs does not correctly update
info that "df" uses. Problem was that "used" space shown in
"df -k" was much lower than it should have been, based on
"du -k" output plus the "available" space info from "df -k".
A quick-witted sysadmin here ran quotacheck and the space was
recovered.
Alan_at_nabeth says:
1. The files on the domain are sparse and not using nearly
as many blocks as they are bytes.
2. A long running problem with AdvFS is that it doesn't
maintain correct data for the system call interface
used by df(1). I don't know the particulars, but it
apparently has something to do with disk quotas.
Dennis Peacock suggested:
I have seen this type of error on Tru64 4.0F. Just remember to always use
the "showfdmn -k" and you will get a properly reported disk usage on you
server.
Received on Mon May 21 2001 - 17:18:18 NZST