Hi, I did some testing again and I see some clusters are started but the backend are not available. They are seen as UP but no traffic goes to them or the frontend doesn't accept it.
This is my main concern. I think it's just some finetuning on timings, or in the config or in ZEN itself. Cheers, Matt 2015-10-04 17:31 GMT+02:00 Matt . <yamakasi....@gmail.com>: > Something I forgot to mention. > > it happened in the past, older version, a lot when you puppet the > master into maintenance mode. This now works much better but I thought > sometimes on the slave a cluster keeps running when turning back and > the same for the master node where some clusters didn't came up well. > > As this is much more stabke now I'm also thinking that you can' t have > them on high mem used HyperVisors, Zen is quite sensible for it, is > this known ? No ballooning is set on these Zen VM's. > > For the rest perfect en becoming better and better! > > Cheers, > > matt > > 2015-10-04 17:24 GMT+02:00 Matt . <yamakasi....@gmail.com>: >> Hi, >> >> OK thanks, my ratio was kinda low, it was set on 5. I was not sure >> there but as I have experience with HaProxy which handles that I was >> not sure here. >> >> Timesync is good! >> >> Thanks! >> >> Matt >> >> 2015-10-04 5:48 GMT+02:00 Emilio Campos <emilio.campos.mar...@gmail.com>: >>> When cluster fails once configured is because there is ntp sync issues or >>> network latency issues. >>> >>> Check the connection to your configured ntp server and increase the cluster >>> dead ratio value to 10 secs. >>> >>> Regards >>> >>> El 02/10/2015 21:57, "Matt ." <yamakasi....@gmail.com> escribió: >>>> >>>> Hi, >>>> >>>> I see again clusters which stop without reason, some are running but >>>> the frontend servers cannot connect to the cluster. >>>> >>>> I need to check the logs but I wonder if this has something todo with >>>> performance as a VM. The machine has no ballooning on memory and >>>> doesn' t have much load. the Memory is currently set on 2GB. and also >>>> the HyperVisors don't use that much CPU, MEM is not on 100% usage >>>> there. >>>> >>>> I think I seen this before when consolidating VM's on less >>>> HyperVisors, the loadbalancers seem to fail more often, and failover >>>> is not started because the loadbalancer itself is still running. >>>> Memory usage on the HyperVisors is mostly around 80% as we have >>>> hotstandby ones. >>>> >>>> Any clue here what we can optimize ? >>>> >>>> At least the latest version is much more stable. >>>> >>>> Thanks! >>>> >>>> >>>> ------------------------------------------------------------------------------ >>>> _______________________________________________ >>>> Zenloadbalancer-support mailing list >>>> Zenloadbalancer-support@lists.sourceforge.net >>>> https://lists.sourceforge.net/lists/listinfo/zenloadbalancer-support >>> >>> >>> ------------------------------------------------------------------------------ >>> >>> _______________________________________________ >>> Zenloadbalancer-support mailing list >>> Zenloadbalancer-support@lists.sourceforge.net >>> https://lists.sourceforge.net/lists/listinfo/zenloadbalancer-support >>> ------------------------------------------------------------------------------ _______________________________________________ Zenloadbalancer-support mailing list Zenloadbalancer-support@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/zenloadbalancer-support
