Proper procedure to bring up secondary

[webguyz]

I opened a support ticket 2 hours ago and still waiting for a ticket number so I thought I would throw this out here.

I have 2 DSS's and am using autofailover. Yesterday the secondary was taken down (just pressed shutdown, did not stop any tasks) so that I could move the motherboard and disks to a new bigger server case. All this time the primary DSS was working great and everybody was connected and happy. I was getting errors that the replication was down and it was in degraded state byt still working. Everything was successful with rebuilding the secondary box and so I added it to the network and powered it up.

When I did this some Linux iscsi connected servers freaked out and stopped working. its like they did not know where to connect to. They would not work until I shut down the secondary DSS and did a iscsi restart on the linux servers. I had a windows server connect to DSS primary as well and it did not seem to have a problem with me bring up this secondary DSS.

So my dilema is this. What did I do wrong in bringing the secondary down initially. Should I have stopped services on the Primary or what? What is the correct procedure for downing secondary

Second question is how do I recover. How do I bring up the secondary DSS box without screwing up the Linux servers?

The secondary server has the same everything, usb dom, disk controller, all the network cards, all the same disks, just moved it to a bigger case to match the primary DSS so I could add more disks and have the setup be identical between the 2 DSS's in the autofailover.



Thanks!

[yataman]

If your primary node is in active mode and you only bring up secondary node you should only start iSCSI Failover service on secondary node in Failover Manager web module. After it you secondary node will be in passive mode. Next you should ensure that all statuses in Failover Monitor are correct (replication tasks are running and so on).

I suppose that you have correctly configured network and your Linux clients are connecting to VIP address (that is at this moment on primary/active node). Sitatution that you describe is a little bit strange, because after starting secondary node iSCSI Failover service is not started and no VIP is available at this moment on this node (it means that it should not interrput work of primary node).

Maybe you are doing some network IP conflicts? Please check your network configuration.

Cheers

[yataman]

There is one more option:

After starting iSCSI Failover on secondary node this node goes automatically to active mode (it means that VIP is available at this moment on both nodes). This sitution happens when your auxiliary connections between nodes are broken. Before starting iSCSI Failover service on secondary node please ping all IP addresses used in iSCSI Failover on Primary node in order to check connectivity between nodes (this step please do under console CTRL+ALT+T on secondary node).

Cheers

[webguyz]

Was working with support and we decided to try redoing the replication. The secondary DSS was acting very weird to the point I decided to just blow away the secondary and start with a fresh install. Its replicating correctly now and I'm averaging about 40 MB/sec.

What I should have done before shutting down the secondary was click the stop button under replication. I think that the secondary got very confused as a result.

Bad thing is if your doing autofailover your primary stop advertising its virtual IP because the volumes are not in sync. I am in the process of redoing my servers so don't have any production yet. If this was a production situation you would have to change your actual primary DSS ip to what the virtual one was temprorarily until you fixed the secondary.