4 Nodes Ceph

[yena]

Hello, i'm testing a 4 nodes Ceph Cluster:
Each node have two sata HD and two SSD for Journal.
-----------------------------------------------------------------------------------
ceph -w
cluster 1126f843-c89b-4a28-84cd-e89515b10ea2
health HEALTH_OK
monmap e4: 4 mons at {0=10.10.10.1:6789/0,1=10.10.10.2:6789/0,2=10.10.10.3:6789/0,3=10.10.10.4:6789/0}
election epoch 150, quorum 0,1,2,3 0,1,2,3
osdmap e359: 8 osds: 8 up, 8 in
flags sortbitwise,require_jewel_osds
pgmap v28611: 512 pgs, 1 pools, 15124 MB data, 3869 objects
45780 MB used, 29748 GB / 29793 GB avail
512 active+clean
client io 817 B/s wr, 0 op/s rd, 0 op/s wr
-----------------------------------------------------------------------------------

I have tested the fail of 2 Nodes and the cluster go away.
With a 3/2 Pool, how many OSD can i lost ? ( how can i calculate it ? )

 

[Ashley]

Min 1, Max 2, all depends on where the last 2 copies of data are when you take down 2 nodes.

If you wanted to be able to survive 2 nodes failing out of 4 you would need to have a 4/2 pool.

 

[yena]

Ok so now i can lost max 2 OSD.
I Will try the performance of 4/2 pool ..
Thanks!

 

[fabian]

if you have 4 monitors, your monitors will not be quorate if any two are down.. if you have three monitors, your cluster might be still quorate if a mon and the non-mon node fail. even monitor counts almost never make sense. either upgrade your cluster to 5 nodes (with 3 or 5 monitors) or keep 4 nodes, but only make 3 of them monitors.

 

[yena]

Yes! that make sense