We have 2 active datacenters, DC1 and DC2. For each datacenters we have a separate EqualLogic cluster CL1 for DC1 and CL2 for DC2 both with a storagepool named ISCSI. In datacenter DC2, we have an additional EqualLogic and this one is configured with a separate Storage pool called SYNCREP and is configured in the cluster CL1 for DC1. So far so good...
In cluster CL1/ISCSI, we created a volume VOL which is syncrepped to DC2 in cluster CL1/SYNCREP.
Now we need to take DC1 offline for maintenance, and the volume VOL needs to stay online during this maintenance. Is it possible to make the volume active in cluster CL1/SYNCREP (switch to syncalternate) without downtime, and will CL1 still be active if the other members in cluster CL1 are down?
The server which connects to this volume is on vmware and has the Host Integration Toolkit installed and points to the cluster ip for CL1.
Thanks in advance,
It sounds like you have two groups in different locations, and are doing asynchronous replication between them. SYNC rep only works within the same EQL group never between groups.
If you had two pools in the same group you could then do SYNC rep and failover to that pool, but it would still be within the same Datacenter.
You can't switch to another Datacenter when you are using async replication live. VMware Site Recovery Manager helps to make the transition easier however.
Thanks for your reply, we are doing a syncrep in a single group over 2 pools. We stretched the group over our 2 datacenters and the volumes are in sync. We have a 10G darkfiber between the datacenters which is fast enough for the SyncRep. I like to know if I can switchover to the SyncAlternate (in DC2) without losing the connection and, if DC1 is offline, the virtual machine will hold the connection to the SyncAlternate volume.
Thanks in advance,
Assuming the link has low latency, then yes you can switch to, and run from the alternate pool. However, the connection does not 'stay'. The iSCSI connection is only to the primary side. If you switch to the alternate pool, the array will force the servers to logout and then log right back into the group. When it does the connection will be moved to the alternate pool, which is now the active side.
What's the distance between the members?
The connections are 14km and 54km over a very low latency link.
clss001> ping 172.24.5.31
PING 172.24.5.31 (172.24.5.31): 56 data bytes
64 bytes from 172.24.5.31: icmp_seq=0 ttl=255 time=0.000 ms
64 bytes from 172.24.5.31: icmp_seq=1 ttl=255 time=0.000 ms
64 bytes from 172.24.5.31: icmp_seq=2 ttl=255 time=0.000 ms
64 bytes from 172.24.5.31: icmp_seq=3 ttl=255 time=0.000 ms
64 bytes from 172.24.5.31: icmp_seq=4 ttl=255 time=0.000 ms
So the connection will shortly disconnect and automatically reconnect? Host Integration toolkit is installed..
Just an FYI: when using ping to test for latency it's good to add some data along with it for a more accurate test.
re: Connection. Yes. It will happen very quickly just a few seconds.
re: HIT kit. That's good to have but does not impact anything here. This is all on the array side and works with any iSCSI standards compatible initiator.