On Tue, Nov 02, 2004 at 11:22:04PM +0100, christophe varoqui wrote: > > [Even ignoring the suspend/resume queueing issues that aren't > > resolved yet.] > I would guess they need resolving anyway Yes, but no guarantees it can be done efficiently yet. (i.e. expect a performance hit whenever a table is reloaded) > But in a low memory situation I feel your scheme won't bring much more > garanties : it relies on userspace too after all. After paths have genuinely failed. The other point is that I'm trying to view the loaded table as a static long-term representation of the configuration. It should normally only need changing if the admin decides to change things. Once it has loaded the table, the roles of userspace are: - to deal with short-term state changes by testing paths pro-actively and reinstating/failing them and/or enabling/disabling PGs according to requested policy, - to maintain statistics and report problems to the sysadmin. Alasdair -- agk@xxxxxxxxxx