Replication 3 to 2 - ceph - pgs degraded

[sahostking]

I think I may have found something.

I had an issue with diskspace and hence changed from replication x3 to x2 knowing the possible risks.

However it was to be temporary while I add more OSDs.

But when I add more OSDs to new servers now I am noticing very high IO wait and servers "freeze".

I am thinking its because it pauses the PGS it is copying over to new disks? This is the issue you have with replicate x2.

When I had replication x3 with 5 nodes never had this issue. Adding more seems to be the issue now changing after replication x2.

I'll be changing back to x3 rather but just wanted to know if anyone experienced this or if my assumption is correct?

 

[sahostking]

yip confirmed. Changed back to x3 replication now , no more VMs freezing due to high io wait when taking out OSDs and rebalancing happening. Everything still runs smoothly at a cost of a little performance penalty.