I have to admit that I equivocated about this when I went back to address the max_timeout * 0-mirrors case. It's probably sufficient to retry 2 times per host if we only have mirror and origin (assuming the default is 4, in this particular case); however, you've managed to convince me that we probably should write this as num_mirrors + 1 instead.
Thanks, -j On Thu, Oct 30, 2008 at 02:43:31PM -0700, Brock Pytlik wrote: > filelist.py: > Should line 233 be: retry_count = max_timeout * (num_mirrors + 1) > If we normally retry 5 times without any mirrors, do we still want to > retry 5 times when we add a mirror, or do we want to retry 10? > > Other than that LGTM > > Brock > > > [EMAIL PROTECTED] wrote: > > Folks, > > This is a code reveiw that addresses some problems with the initial > > mirror selection heuristic. It was weighting the primary server equally > > with the mirrors. In situations where the mirror was much faster than > > the master, we'd still end up choosing the master and going much slower. > > > > The webrev can be found at the location below. > > > > http://cr.opensolaris.org/~johansen/webrev-3803/ > > > > With this fix in place, we'll only go to the master server for data if > > we can't find the file in any of the mirrrors' repositories. > > > > This is also a small code review. > > > > Thanks, > > > > -j > > _______________________________________________ > > pkg-discuss mailing list > > [email protected] > > http://mail.opensolaris.org/mailman/listinfo/pkg-discuss > > > > _______________________________________________ > pkg-discuss mailing list > [email protected] > http://mail.opensolaris.org/mailman/listinfo/pkg-discuss _______________________________________________ pkg-discuss mailing list [email protected] http://mail.opensolaris.org/mailman/listinfo/pkg-discuss
