Our usual case is 1-2 vms doing io bound tasks while others are waiting. So page cache seems ok for our needs - let it give more memory for more loaded vm.
Shared cache and cache flood can produce problems and hard to test. Test with 3 vms tells little about 15 vms. Maybe divide images between...
Upgrade is planned, will try to speed it up...
We catch this bug sometimes (22464) but usually all osds is in place and green, with adequate latency
I know, I know... :( But we have what we have and _very_ limited budget.
Well, small size of librbd cache can explain performance boost with...
Hello!
We are running Proxmox + Ceph for vm image storage on 5 nodes. Ceph run on dedicated network with little unmanaged switch we want to replace.
Is there a safe way to live replace one ceph-network switch with another? I think we need about 15-30 seconds to reconnect the cables. Will ceph/vm...
This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
By continuing to use this site, you are consenting to our use of cookies.