[OpenAFS-devel] Re: sanity check: client vldb cache time

[Harald Barth]

> Indeed, there does not appear to be any mechanism by which a cached VLDB
> entry is invalidated if the volume is moved and the server goes down
> before some attempt to access that volume returns VOFFLINE or VMOVED or
> VNOVOL.  Well, other than manually invalidating all volcache entries,
> which _does_ happen if someone invokes VIOCCKBACK (aka fs checkv).

I think this has happened to me. 1. Move all volumes to new server and
then 2. shutdown old one. If you are too fast with 2, you have to go
around and issue fs checkv.

> I'm inclined to think that when a server goes down, something (either
> afs_MarkServerUpOrDown or CkSrv_MarkUpDown) should call afs_ResetVolumes
> for that server, so any VLDB entries get invalidated.

Yes.

> I'm also inclined to think that we ought to expire even volcache entries
> for RW volumes, so they get looked up again once in a while.

Should not impact too much to do that once in a while, like every X hours.

Harald.