On 2005-07-25T14:41:35, Christophe Varoqui <christophe.varoqui@xxxxxxx> wrote: > Maybe I miss some important point already raised in the past, but ... > How about adding more params to queue_if_no_path ? eg, > > - timer value > - pending io queue length > > That way kernel is kept free from policies, If the timer value is handled inside the kernel, that would help. The response it triggers could be the same like the user-space-to-kernel heartbeat I described. The pending queue length is probably not a good measure, I think, but it would be cheap to add. > and the framework survives a daemon crash. Sincerely, Lars Marowsky-Brée <lmb@xxxxxxx> -- High Availability & Clustering SUSE Labs, Research and Development SUSE LINUX Products GmbH - A Novell Business -- Charles Darwin "Ignorance more frequently begets confidence than does knowledge"