Coda File System

Re: more problems seen

From: Jan Harkes <jaharkes_at_cs.cmu.edu>
Date: Tue, 22 Jan 2002 12:18:41 -0500
On Sun, Jan 20, 2002 at 09:20:45AM +0100, Ivan Popov wrote:
> Worse than that, I cannot trigger server-server resolution,
> a replicated volume is updated on scm but the changes do not propagate to
> the other server unless I shut down scm and then the client begins to talk
> to the other server.

This sound a lot like we are still seeing the similar problem whenever
we hit '0' in the host addresses somewhere. I just grepped the coda-src
tree for uses strcmp and strncmp but I haven't come up with anything
yet.

> [often I get "ValidateVolumes ... failed!" in the logs (about single
> replicated volumes) - what does it mean? It doesn't look like causing any
> trouble.

That typically fails, it is part of the 'rapid cache validation'. Venus
has a conservative estimate of the VolumeVersionVector and as soon as
anything in a volume is updated this volume version vector is changed.

If the ValidateVolumes succeeds then the client knows it doesn't have to
revalidate any of the files in that volume.

> While playing with replicated volumes i get
> "ViceValidateAttrs ... failed!" - is it harmful or not?]

This is an indication that either a file has been updated, or that one
of the servers might have missed some updates and needs some better
scrutiny (i.e. a GetAttr, and possibly resolution).

Jan
Received on 2002-01-22 12:18:43