On Tue, Apr 23, 2013 at 12:59 PM, Michael K. Johnson <[email protected]> wrote: > On Tue, Apr 23, 2013 at 06:39:21AM -0400, Michael K. Johnson wrote: >> While adding "condgraceful" to httpd and using condrestart for >> memcached is probably a good idea, it is important to solve the > > Actually, it's not at all clear to me that condrestart for memcached > is a good idea. > > I can't see any reason to *restart* memcached at all. We are not > changing its config. > > If it is not running, then we should not start it. Even with a > conary cache running, if memcached goes away the cache is just > slower, it does not quit running. If it is running, the only thing > that restarting it will do is throw away cached entries. > > With the DIRECT fallthrough implemented, I do not think we should > touch memcached at all. >
FYI, I just committed conary-proxy:source 0.8-2[1] (and 0.9-1 version bump) which implements mkj's suggestions. It needs to be baked together with mkj's httpd condgraceful.patch as it relies on the condgraceful function being available in its PostInstall and PostUpgrade scripts. [1]: http://lists.foresightlinux.org/pipermail/foresight-commits/2013-April/060840.html _______________________________________________ Foresight-devel mailing list [email protected] https://lists.foresightlinux.org/mailman/listinfo/foresight-devel
