On Fri, Dec 18, 2015 at 9:55 AM, Bajin, Joseph <jba...@verisign.com> wrote: > That was me.. > > We are using McRouter which then speaks to memcached to talk to all the > memcache servers that you have up and running. It keeps track of what is up > and down so it knows where to send traffic. You can get pretty complicated > with it and we have started to do more complex operations such as warming up > cache’s, going across regions to look in that cache for things like tokens, > and a few others. > > To start off, you can simple use this configuration [1] which will get you > the latest token and update/delete/create to the fastest location. > > I’ve started to write a small blog post about it, but I haven’t finished out > the pictures yet.
That's great, looking forward to the post. :) I didn't have much time to experiment, but I did find in my limited testing that it was a bit slower than I thought it would be to have mcrouter in place. But as you say it's very configurable and perhaps I was using a poor config. If you have any performance information to share in the post that'd be great too. Thanks, Curtis. > > Let me know if you have questions though. I’m happy to share more > information. > > —Joe > > [1] https://gist.github.com/RaginBajin/0339436c17e814e16e99 > > From: Pedro Sousa <pgso...@gmail.com> > Date: Friday, December 18, 2015 at 7:36 AM > To: "Ajay Kalambur (akalambu)" <akala...@cisco.com> > Cc: "openstack-operators@lists.openstack.org" > <openstack-operators@lists.openstack.org> > Subject: Re: [Openstack-operators] Keystone token HA > > Hi Ajay, > > someone in this mailing list mentioned mcrouter + memcached to achieve that, > I'm also looking to test it soon on my lab. > > Regards, > Pedro Sousa > > On Fri, Dec 18, 2015 at 6:48 AM, Ajay Kalambur (akalambu) > <akala...@cisco.com> wrote: >> >> Hi >> If we deploy Keystone using memcached as token backend we see that >> bringing down 1 of 3 memcache servers results in some tokens getting >> invalidated. Does memcached not support replication of tokens >> So if we wanted HA w.r.t keystone tokens should we use SQL backend for >> tokens? >> >> Ajay >> >> >> _______________________________________________ >> OpenStack-operators mailing list >> OpenStack-operators@lists.openstack.org >> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-operators >> > > > _______________________________________________ > OpenStack-operators mailing list > OpenStack-operators@lists.openstack.org > http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-operators > -- Blog: serverascode.com _______________________________________________ OpenStack-operators mailing list OpenStack-operators@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-operators