On Mon, Jun 16, 2008 at 12:03:49PM -0700, Garrett D'Amore wrote: > If optimizing i.services is not a reasonable task, how about just > emitting a message, filing a P4 bug to improve the performance of > i.services, and worrying about the 1.5 minutes later?
That's what I want to do. I just thought I'd at least mention this and ask for advice rather than put something back and surprise any of you (Garret, Jim, ...). In particular, I'm willing to accept that 1.5 minute upgrade perf degradation. I don't think we've got guidelines w.r.t. upgrade performance regressions. But perhaps I'd missed something, and perhaps that 1.5 minutes really, really matters. So the bottom-line question then is: Is it OK for a project to regress upgrade performance by 1.5 minutes? John Levon objected (on IRC) (mildly?) to any scheme involving postrun-type services as I was considering something like that. Lubomir Sedlacik (salo), also on IRC, claimed that i.services breaks zone installation, though salo gave no details on that. Nico -- _______________________________________________ networking-discuss mailing list [email protected]
