Multisite cluster / Status No Quorum after primary site is down

[DSchaps]

Hello,
i hope someone can help me or can give me the information i need it.

First, ist only a test environment. :-)

I have installed a mutlisite cluster with 3 nodes on each site, so i have 5 managers. On the primary site 3 Managers and in the other side 2 Managers. After "reset" the primary site for testing i have only 2 active managers. So i cannot access the volumens, thats OK because of the two managers.

BUT

How is it possible to take the volumes (secondary site) online?
I cannot start or great other managers in the management group. I cannot found any information in the cli reference guide.

I know the best practive for multisite cluster, but we dont have more then two sites. And i think it must be possible to recover the quorum majority.

Hope someone can help me

Regards
Daniel

[Gauche]

In the case you describe you can run the "recoverquorum" command. But that should only be done if you really have lost, forever, the primary side because it effectively kicks the other managers out of the group.

The multi-site user manual has a section on quorum that describes the different ways to solve this.

I'll summarize though. You should either...
1-run 4 manangers, two in each site, and have a "virtual manager" added but not running. This way if either site fails you simply right click the virtual manager in the surviving site and start it. This design only allows for manual failover, but works if you really only have two sites.
2-run 4 managers at each site and have a "fail-over manager" running somewhere nuetral. This gives you completely automated failover in either direction.... so is obviously the preferred method.

The way you have it running right now is okay if you really want the site with three managers to be the only site that can survive automatically... and will only failover to the 2nd site if the first is crushed by a meteor or something similar.
Not the usual setup for multi-site configration though.

[teledata]

I'll add an idea to the above... If you don't HAVE a neutral 3rd location:

Run 2 managers at each site

Run a failover manager at the primary site.

Have a copy of the failover manager VM NOT running, but hosted on a VM at the secondary site...

If the primary site goes down, start up the copy of the Failover Manager VM at the secondary site. Quorum should then return at the secondary site.

[teledata]

Ack - ignore that last reply.. I misread the 2nd option....

the 4 mgr in each location = the win! ;)

Blast this forum... how do I edit a post?

[DSchaps]

Thanks. It works fine. I forgott to add the virtual Manager first.
After the failover you cannot add this manager.
In this case, i would like to try the recoverquorum command. I found some information in the manuals, that is possilbe the recover the quorum with the cli. But in the cli (Version 8.1.0.102) i cannot found this command line tool.

[Olvi_1]

Situation: 2-site cluster, FOM running at primary site.

Noticed that I can't add Virtual Manager when using FOM. So I can't have Virtual Manager as 'spare' in Management Group waiting for primary site to go down.

I'm was also thinking adding copy of FOM to secondary site. Is this supported / documented configuration?

Have anybody founded systax for recoverquorum command? There is no mention about the command in CLIQ user guide.

-Miika

[jim cook_5]

I'd really like more info on the recoverquorum command, but haven't found it anywhere either.

thanks,
Jim

[Uwe Zessin]

It is hidden in the scripting examples.

Please be careful - do not experiment with it in a production environment!!

[jim cook_5]

I may not have a choice, our 2 node cluster lost quorum after an extended power outage, the manager on the second node is stuck "offline".

Thanks for the warning.

[Uwe Zessin]

You've been running two nodes with a manager function, but no FailOver Manager (FOM) or Virtual Manager?

Then you're right :-(