mscs question 1

[anatheilper]

Hi,
I wanted to ask several questions regarding the MSCS abilities:
1. Registry replication - are there any known pitfalls regarding this ability, how does this work, and what happens with identical registry entries when moving from node to node? are the entries overriden?
2. service control through the cluster - how can i configure what to do in case that the service is stopped ( example, I want it to automatically restart\in case of a fail, i want the node to failover)

 

[Kuskaya]

Hi,

HKLM_Cluster will only replicated to the other nodes.In example if you reboot on of the node it will get the actual Cluster Rgistry Hive from the Q:\MSCS\chkXX.tmp.The .tmp is the CLuster Registry Hive.You can import it in you registry.

If you make in example some changes over the cluster adminiatrtor the changes first saved locally after that it it saved to the quorum disk.

The cluster service has a restart treshold.In example 3 times.If it cannot start after 3 attempts it will failover.

This is the reason why you are using a cluster.You should try to troubleshoot it based on the infos in the system event logs.

An automatization of the cluster service has no meaning.

Regards,
Cengiz Kuskaya

 

[anatheilper]

Hi,
I think that I didn't make myself clear enough regarding the second question:
I have written a service application which I want to add as a resource to a group in the cluster.
I want to have a similar behavior inside my cluster as I have in the service manager: I want to define that in case of recovery to restart the service ( so failure of this service won't cause a failover of the group to another node, but to try to restart the service again).
How is this possible in MSCS?

Thanks,
Anat.

 

[Kuskaya]

After creating a new resource for your service (Generic Service) Go to the resource propertiers and check the option "run in a seperate resource monitor".

If your own service makes trouble the cluster will not failover only you service will goes offline.

Regards,
Cengiz Kuskaya

 

[anatheilper]

Hi , i still have some things I dont understand:
I want to add 2 instances of the same service to 2 different groups. I get an error message that the object already exist.
Is this something that I'm doing wrong, r is it a limitation of MSCS, and if so, is there a documentation to it?
Thanks,
Anat

 

[Kuskaya]

Hi,

what type of instances are you trying to add to the 2 group and how many groups have you at all.

Some limitations are available.In example ip's.In wich type of groups are you trying to add wich type of instances.

Are there registered any system events in the eventnwr.msc from the source clusssvc ?

Regards,
Cengiz

 

[xmsre]

Number 1 sounded to me like you wanted to use registry checkpoints.

http://support.microsoft.com/kb/174070/en-us

A little outdated, but we still have registry checkpoints in W2K

http://support.microsoft.com/kb/816913/en-us

and 2K3

http://support.microsoft.com/kb/897666/en-us

Many cluster aware applications depend on them. Check the cluster SDK for more info.

 

[anatheilper]

I've created a service that needs to be run at all of the nodes of the cluster, meaning that if I have 2 nodes, I'll define 2 resource groups that each of them will contain this service as a resource ( and the ip of the machine) and I want it to be run in active/active mode, meaning that when moving from node to node, the different instances of the application on the different resource groups will run on the same machine.

 

[Kuskaya]

Hi anatheilper,

you cannot use the machine ip in your new group.As soon as you add the physical node to the domain a computer object is created with this ip and network name.

If you try to use the ip again you will get the message object already exist.

If you explain us what you are trying to do with this new group maybe we can try to find a different way to do this.

Regards,
Cengiz Kuskaya