The following are my personal observations (not an official ISU
position) after a *very* frustrating week with AFS. I suspect
you've probably heard some of these before.
* When I send e-mail requesting help to the AFS support address,
I would expect at least a "we got your message" reply by
noon the next working day. I would not expect to call our
support person and find he's not even heard of our problem.
* Upgrade documentation should clearly indicate known problems
with upgrading. It is *very* disturbing to have an upgrade
result in a big problem and be told (after 2 days of your
cell disfunctioning) it's a "known problem". It is worse when
the advice is not quite right and it takes 4 more days of
Transarc and the OS vendor pointing fingers to get a useful
answer.
* Upgrade documentation should clearly indicate when and how
you can reverse an upgrade.
* If a new release requires more resources, this should be clearly
indicated in the documentation.
* Vfsck should recognize fifos/named-pipes (see sample below)
it is very annoying to have our servers *always* require
manual attention during a reboot.
* The fs flush/flushv commands do not work reliably. Our cell looks
like: (root.afs)(root.cell)(users)(user.username). We recently
replicated the volume users and *nothing* short of a reboot got
clients to recognize this properly (in some cases, the afs
cache had to be "rm -rf"ed too).
* With 18 drives/partitions per server and several hundred volumes per
partition it takes an awfully long time for the fileserver to get started.
Why can't it attach volumes from multiple partitions in parallel (like
the salvager works in parallel)?
* Why is the whole fileserver shutdown to salvage a partition?
* The ability to control what happens when a fileserver and/or volume is
not available (like the NFS "hard", "soft", "intr" mount flags) would
be very useful.
* The per-cell setuid/no-setuid is not very useful. This should
be a per-volume flag too.
* The bos status command should show if a fileserver is just
starting (attaching volumes) as opposed to saying "running".
* It should be possible to specify multiple partitions (less than
-all) on bos salvage.
* Having an "@cell" like "@sys" would be very nice at times.
* Why are quotas set/shown with fs and not vos?
* A -verbose on fs checkservers and checkvolumes listing servers/volumes
as they are checked would be useful.
John Hascall
[EMAIL PROTECTED]
---------------- sample output from vfsck ------------------------------
** Phase 1 - Check Blocks and Sizes
UNKNOWN FILE TYPE I=28744
CLEAR? [yn] y
** Phase 2 - Check Pathnames
UNALLOCATED I=28744 OWNER=root MODE=0
SIZE=0 MTIME=Dec 31 18:00 1969
NAME=/var/tmp/screens/S-backup/3811.console.fs-9
REMOVE? [yn]
