On Sat, Feb 18, 2006 at 09:01:07PM -0600, Ryan Underwood wrote: > > I've been observing this for several months now. The token loss almost > always corresponds with a period of high fileserver load. For example, > moving several ripped CDs to the fileserver while playing another one. > Or when several machines launch their scheduled backup scripts at the > same time. No, there is nothing running 'unlog'. I can keep the token > around for weeks if I don't bother the fileserver too much. A simple > 'aklog' gets the token again, but any AFS operation that was going on > aborts due to the lack of tokens. I have a cornucopia of Linux 2.6 > versions, but the fileserver and clients are all OpenAFS 1.4.0.
So, what would be helpful to track this down? Since the client is in charge of the tokens, it seems that the problem might be in the client somewhere, except for the correlation of the token-loss events with a loaded fileserver. I've never had to debug AFS before, and only have so many hours in the day, so bear with me. -- Ryan Underwood, <[EMAIL PROTECTED]>
signature.asc
Description: Digital signature
