In message <[EMAIL PROTECTED]>,Kim Kimball writes:
>For us that's a win -- instead of making all volumes on the server
>unavailable for upwards of 45 minutes, well over 90% of the users are
>never aware the fileserver went down at all, and with the notifier
>script automatic salvage (my script is still defective due to the 2x
>notifier script execution or I'd paste it here) I've had exactly one
>complaint from one user after multiple spontaneous crashes (not for a
>long time, thank you all developers!) and intentional file server kills.
>
>YMMV.

i have enough brain damage to run fast-restart, namei and a local hack
which just removes f_sync().   we have the occasional unplanned outages
resulting in an unclean shutdown.  so far, no real lost data.  once,
we had a couple volumes that attached but needed salvaging because vos
dump was having a bit of trouble with them (this was before i removed
f_sync() actually).  regardless, fast restart is much nicer since i can
then schedule a full salvage at a more convienent time for the users.

you do need to salvage once in a while to clean up the .__afs* files
that escape.
_______________________________________________
OpenAFS-info mailing list
[email protected]
https://lists.openafs.org/mailman/listinfo/openafs-info

Reply via email to