Just double checking autofailover behaviour

[Gr-R]

If it is rebooted without the secondary node present, it will not restart the failover service on its own. You will need the secondary online to re-activate failover.

[webguyz]

So I will lose the virtual IP when the primary server comes up and does not see the secondary? Don't care about failover, just don't want to lose the virtual IP.

In the processing of moving everything off of my main DSS's onto a temporary DSS so that I can rebuild my main DSS (primary/secondary) with new motherboards and other hardware.

New users out there make sure you pick a motherboard that has at least 3 or 4 pci-e x8 slots otherwise you will have to do what I'm doing and rebuild everything when you run out of x8 slots. Most disk controllers and 10G cards take a minimum of x8.

[Gr-R]

So I will lose the virtual IP when the primary server comes up and does not see the secondary? Don't care about failover, just don't want to lose the virtual IP.

You will lose the VIP

[webguyz]

I guess I could quickly change the real nic IP addresses from what they were to those of the Virtual IP's and bring the server back up in an emergency.

[Toni Bizjak]

I guess I could quickly change the real nic IP addresses from what they were to those of the Virtual IP's and bring the server back up in an emergency.

Maybe, you can't, because address is already reserved for VIP (failover/ifaces.conf).
Anyway, you'll have problems later, to start Failover again.


Can you use multipath on iSCSI client?
You can temporary use both addresses, VIP and RIP (for source Logical Volume).

[webguyz]

How do you get into the OS of DSS, or is that not possible?

[Toni Bizjak]

How do you get into the OS of DSS, or is that not possible?

I can't!

But you can look into log file (Status/HW/Download...) and use some common logic - not 100%, but it helps to sompose an image.

[Toni Bizjak]

I guess I could quickly change the real nic IP addresses from what they were to those of the Virtual IP's and bring the server back up in an emergency.

I've just tested a little different scenario.

After power off both serveres, I've switched on Secondary first.

In this case, VIP is not active until we:
- change Volume replication modes to SOURCE and
- START Failover service.

After that, VIP is active again.

So try to reorganise your upgrade plan so, that active server is allways Secondary.