1828225 Members
2486 Online
109975 Solutions
New Discussion

Re: Why shadow copy ?

 
Wim Van den Wyngaert
Honored Contributor

Re: Why shadow copy ?

Robert,

You're right about the title. The patch is no longer of importance since the cluster will be removed.

Wim
Wim
Petr Spisek
Regular Advisor

Re: Why shadow copy ?

It looks like your cluster to tries mount a non-master shadow member as first. Why do you don't use cluster shutdown? Do you have a qourum disk for this cluster?
Petr
Wim Van den Wyngaert
Honored Contributor

Re: Why shadow copy ?

Petr,

yes there is a q disk.
Cluster shutdown ? Why should I do that ?

Wim
Wim
Petr Spisek
Regular Advisor

Re: Why shadow copy ?

... because the cluster will be synchronized the shutdown and SHADOWSERVER process too (I hope). You will have to the equivalent shadow-members for next mount.
P.
Uwe Zessin
Honored Contributor

Re: Why shadow copy ?

It's my understanding that the so-called 'cluster shutdown' just syncronizes the final message in OPCCRASH.

I doubt that has anything to do with keeping shadow member contents consistant - do you get a merge operation when you fully shutdown/boot a single node? Normally not, is my experimence.
.
Volker Halle
Honored Contributor

Re: Why shadow copy ?

Uwe,

the 'cluster shutdown' actually synchronizes the departure of each node from the cluster, once all nodes have reached the same state during shutdown. The nodes will continue to do MSCP-serving etc. until that time.

In Wim's case, this doesn't matter, as the disks are behind HSZs and are accessible all the time. But it would matter, if shadowset members would be connected locally and you shutdown the nodes without an explicit 'cluster shutdown'. The node shutting down first may leave the cluster (with a locally connected shadowset member), while the other node has not yet dismounted the shadowset. This will lead to a shadowcopy on boot.

Volker.
Petr Spisek
Regular Advisor

Re: Why shadow copy ?

I think, that Wim has configured ShadowSets throught the MSCP. When shuts the single node, Full Copy operations will be started. And he wants to eliminate this, isn't it?
Wim, try to dismount all your shadowsets with /CLUSTER parameters (if you want shutdown the cluster), boot both nodes together and next mount sahdowsets must by clean. If not, problem could solve the ecos or new VMS version, which you can't to use :-)
Petr