Hi, On Thursday 09 February 2012 19:39:02 José Román Bilbao wrote: > We have this scenario: > - A datacenter > - 1 Server running kvm VMs: > * 1 Openfiler to distribute hard disk to other VMs (stores VMs and > VMs's data) and other uses > * Other VMs... > - DRBD configured to replicate Openfiler's volumes as primary > - B datacenter > - 1 Server running kvm as backup for center A failure events > - DRBD configured to serve as backup of Openfiler so VMs can be restarted > in case of failure with updated data. It also works as primary. > > Both centers are connected through wireless connections (500 Mb/s) which is > good for our requirements. Nevertheless, there is a ping of 300 ms because > of multiple routers in whithin... > > We have been experiencing multiple split-brain situations and we don't know > why... perhaps the link is down for a while but I don't understand the > source of the problem as although they are primary-primary, one of the > servers should never write to disk as all machines are kept on datacenter > A, is this assumption right?, is this split-brain just "conceptual" telling > that network was lost but no real uncoherences have appeared?. Under such > asumption.. would it be ok to apply autorecover?.
If the second machine is completely passive and failover of the vms happens manually, why is the drbd in dual-primary (and thus C-sync)? Would be much easier to have single-primary which resolves many of the split- brain headaches, especially when the slave is only promoted by hand. And using B or even A sync will reduce your disks latency by 600ms (two times the 300ms). Good luck, Arnold
signature.asc
Description: This is a digitally signed message part.
_______________________________________________ drbd-user mailing list [email protected] http://lists.linbit.com/mailman/listinfo/drbd-user
