On Wed, Mar 21, 2007 at 11:15:02AM +0100, Enrico Weigelt wrote:
> Today, it happened again. My notebook had several problems in
> the fs (broken dirs, readonly, ...) and cfs checkservers hung, 
> while the local router/server (which was started about a min 
> earlier) had no trouble.
> 
> Shutting down and restarting venus solved the problem.
> 
> The only noticable difference between these two boxes is that the
> notbook isn't always shut down cleanly. So maybe some things in 
> the cache may be broken. Obviously this got repaired on the second
> startup of venus.

Strange, I would think that anything that needs to be repaired will be
caught during salvage of the first startup. Also server information is
not (yet) cached at all, so cfs checkservers hanging wouldn't really
follow from problems with cached data.

The only thing I can think of is that the restarted venus will be
connecting from a different local port. So as far as the server is
concerned we see a new client connecting and the server will set up a
brand new callback connection.

Checkservers could hang in several places, either during the RPC2
connection handshake, which seems unlikely since the restarted client
doesn't have a problem. Or while the server is trying to ping back or if
that fails, while reestablishing the callback connection.

Jan