[OpenAFS-devel] --fast-restart?

[Jeffrey Hutzelman]

On Tuesday, August 30, 2005 03:42:17 PM -0400 chas williams - CONTRACTOR 
<chas@cmf.nrl.navy.mil> wrote:

> In message <AC490E8C5769CDC04F55B22B@sirius.fac.cs.cmu.edu>,Jeffrey
> Hutzelman w rites:
>> So, the idea here is that you don't force a full salvage of every volume
>> on  a partition just because the fileserver has crashed.  If the
>> fileserver  detects a problem with a volume, it will still mark the
>> volume "needs  salvaged" and take it offline, and since that flag is
>> persistent, that  volume will _stay_ offline until it gets salvaged --
>> presumably by an admin
>
> right.  but i dont really see many cases where fileserver might actually
> do this.  we have crashed/wedged the local fileserver several due to
> some hardware trouble and i have never seen a volume come up as needing
> salvaged.  usually, we get a complaint at some later date from a user
> about strange problems with files in his volume.

The fileserver will do this when it actually has a problem accessing a 
volume.  For example, if the vnode index points to an underlying inode that 
doesn't exist, or if there's an error accessing the underlying file, or a 
problem accessing a directory, or whatever.  It doesn't do it just because 
the volume wasn't detached cleanly by the last server.

Note that I don't particularly see this as useful either, but someone 
wanted it, and in some circumstances it could be useful (for example, 
consider a rep server with only RO clones - that doesn't need a salvage on 
every fileserver restart)