>> So if you are doing maintenance on a mon host in a 5 mon cluster you will still have 3 in the quorum. > > Exactly. I was in exactly this situation, doing maintenance on 1 and screwing up number 2. Service outage Been there. I had a cluster that nominally had 5 mons. Two suffered hardware issues, so it ran on 3 for a distressingly extended period, a function of sparing policies, a dilatory CSO, and a bizarre chassis selection that was not used for any other purpose. The cluster ran fine, though there was a client side bug (Kefu might remember this). The moral of the story is not that 5 doesn’t protect you, but rather that with 3 it could have been much, much worse. Especially with overseas unstaffed DCs. Mons are lightweight and inexpensive compared to compute or OSD nodes, the burgeoning constellation of ceph-mgr plugins notwithstanding. Double failures happen. They happen to OSD nodes, which is one reason why 2R is a bad idea. They happen to mons too. Back …. I think it was the 2015 OpenStack Summit in Vancouver, there was a Ceph operators BoF of sorts where the question was raise if anyone found going to 7 to be advantageous. The consensus seemed to be that any RAS benefits were down the tail of diminishing returns, but that one would be climbing the traffic curve of inter-mon communication. imho, ymmv, aad > . I will update to 5 as soon as I can. > > Secondly: I actually do not believe a MON update has any meaning. It will be behind the current term the moment the down MON rejoins quorum. If you loose all MONs, you would try to bring the cluster up on an outdated backup. Did you ever try this out? I doubt it works. I would expect this MON to wait for an up-to-date MON to show up as a sync source. I suspect it will neither form nor join quorum. If you force it into quorum you will likely loose data if not everything. > > You don't need a backup of the MON store. It can be rebuild from the OSDs in the cluster as a last resort, the procedure has been added to the ceph documentation. Otherwise, just make sure you always have a quorum up. If you really need to refresh a MON from scratch, shut it down, wipe the store and bring it up again. > > Best regards, > ================= > Frank Schilder > AIT Risø Campus > Bygning 109, rum S14 > > ________________________________________ > From: Freddy Andersen <freddy@xxxxxxxxxxxxxx> > Sent: 12 February 2021 16:47:08 > To: huxiaoyu@xxxxxxxxxxxx; Marc; Michal Strnad; ceph-users > Subject: Re: Backups of monitor > > I would say everyone recommends at least 3 monitors and since they need to be 1,3,5 or 7 I always read that as 5 is the best number (if you have 5 servers in your cluster). The other reason is high availability since the MONs use Paxos for the quorum and I like to have 3 in the quorum you need 5 to be able to do maintenance. (2 out of 3, 3 out of 5… ) So if you are doing maintenance on a mon host in a 5 mon cluster you will still have 3 in the quorum. > > From: huxiaoyu@xxxxxxxxxxxx <huxiaoyu@xxxxxxxxxxxx> > Date: Friday, February 12, 2021 at 8:42 AM > To: Freddy Andersen <freddy@xxxxxxxxxxxxxx>, Marc <Marc@xxxxxxxxxxxxxxxxx>, Michal Strnad <michal.strnad@xxxxxxxxx>, ceph-users <ceph-users@xxxxxxx> > Subject: Re: Backups of monitor > Why 5 instead of 3 MONs are required? > > > > huxiaoyu@xxxxxxxxxxxx > > From: Freddy Andersen > Date: 2021-02-12 16:05 > To: huxiaoyu@xxxxxxxxxxxx; Marc; Michal Strnad; ceph-users > Subject: Re: Re: Backups of monitor > I would say production should have 5 MON servers > > From: huxiaoyu@xxxxxxxxxxxx <huxiaoyu@xxxxxxxxxxxx> > Date: Friday, February 12, 2021 at 7:59 AM > To: Marc <Marc@xxxxxxxxxxxxxxxxx>, Michal Strnad <michal.strnad@xxxxxxxxx>, ceph-users <ceph-users@xxxxxxx> > Subject: Re: Backups of monitor > Normally any production Ceph cluster will have at least 3 MONs, does it reall need a backup of MON? > > samuel > > > > huxiaoyu@xxxxxxxxxxxx > > From: Marc > Date: 2021-02-12 14:36 > To: Michal Strnad; ceph-users@xxxxxxx > Subject: Re: Backups of monitor > So why not create an extra start it only when you want to make a backup, wait until it is up to date, stop it and then stop it to back it up? > > >> -----Original Message----- >> From: Michal Strnad <michal.strnad@xxxxxxxxx> >> Sent: 11 February 2021 21:15 >> To: ceph-users@xxxxxxx >> Subject: Backups of monitor >> >> Hi all, >> >> We are looking for a proper solution for backups of monitor (all maps >> that they hold). On the internet we found advice that we have to stop >> one of monitor, back it up (dump) and start daemon again. But this is >> not right approach due to risk of loosing quorum and need of >> synchronization after monitor is back online. >> >> Our goal is to have at least some (recent) metadata of objects in >> cluster for the last resort when all monitors are in very bad >> shape/state and we could start any of them. Maybe there is another >> approach but we are not aware of it. >> >> We are running the latest nautilus and three monitors on every cluster. >> >> Ad. We don't want to use more monitors than thee. >> >> >> Thank you >> Cheers >> Michal >> -- >> Michal Strnad >> > > _______________________________________________ > ceph-users mailing list -- ceph-users@xxxxxxx > To unsubscribe send an email to ceph-users-leave@xxxxxxx > > _______________________________________________ > ceph-users mailing list -- ceph-users@xxxxxxx > To unsubscribe send an email to ceph-users-leave@xxxxxxx > _______________________________________________ > ceph-users mailing list -- ceph-users@xxxxxxx > To unsubscribe send an email to ceph-users-leave@xxxxxxx > _______________________________________________ > ceph-users mailing list -- ceph-users@xxxxxxx > To unsubscribe send an email to ceph-users-leave@xxxxxxx > _______________________________________________ > ceph-users mailing list -- ceph-users@xxxxxxx > To unsubscribe send an email to ceph-users-leave@xxxxxxx _______________________________________________ ceph-users mailing list -- ceph-users@xxxxxxx To unsubscribe send an email to ceph-users-leave@xxxxxxx
 |