On 05/15/2013 09:12 AM, Chen, Xiaoxi wrote: > Thanks， but i am not quite understand how to determine weather monitor > overloaded？
I've also found it difficult, and a process of elimination. Make sure your mon debugging is minimal. If it's not, lower it and see if it helps. Watch your mon host with vmstat or iostat, to see what it's disk write rate is during the periods you're having trouble with OSDs having heartbeat issues. If it's near what you think the limiting rate is for whatever device is holding the mon's store.db directory, that might be your bottleneck. > and if yes，will start several monitor help？ Sage or one of the other Ceph developers can speak better to this. I will note that with three monitors, if one is too busy, it'll miss renewing its lease. The result of that will be an election to get a new quorum. If you run multiple mons and they hold frequent elections where a mon drops out of quorum, then comes back in on the next election, I've found that to be a sign that my mons are too busy. -- Jim > > 发自我的 iPhone > > 在 2013-5-15，23:07，"Jim Schutt" <[email protected]> 写道： > >> On 05/14/2013 09:23 PM, Chen, Xiaoxi wrote: >>>> How responsive generally is the machine under load? Is there available >>>> CPU? >>> The machine works well, and the issued OSDs are likely the same, seems >>> because they have relative slower disk( disk type are the same but the >>> latency is a bit higher ,8ms -> 10ms). >>> >>> Top show no idle % but still have 30+% of io_wait, my colleague educate >>> me that io_wait can be treated as free. >>> >>> Another information is offload the heartbeat to 1Gb nic doesn't solve the >>> problem, what's more, when we doing random write test, we can still see >>> this flipping happen. So I would like to say it may related with CPU >>> scheduler ? The heartbeat thread (in busy OSD ) failed to get enough cpu >>> cycle. >>> >> >> FWIW, also take a close look at your monitor daemons, and >> whether they show any signs of being overloaded. >> >> I frequently see OSDs wrongly marked down when my >> mons cannot keep up with their workload. >> >> -- Jim >> >> _______________________________________________ ceph-users mailing list [email protected] http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
