stop/start CCR CMS

[Chris]

I have a 2007 sp1 CCR cluster. If I stop the active node by using either
cmdlet or EMC

Q1: will it automatically failover to the passive node? I think it will.

Q2: after maintenance and starting original active node (was stopped early)
will the cluster failback? Or can I started it since there will be two
active nodes.

Thanks.

[Scott Schnoll [MSFT]]

Hi Chris:

A1: No, it will not. Failover implies an unscheduled outage. A move, or
what we also call a handoff, is a scheduled outage. So if you need to
perform maintenance, instead of stopping the clustered mailbox server, you
move the clustered mailbox server to the passive node, thereby making it the
new active node.

A2: When you're done with the maintenance, you move the clustered mailbox
server back to the node, and then if necessary perform the same maintenance
on the other node.

Hope this helps.
--
Regards,

Scott Schnoll
Microsoft Corporation
This posting is provided "AS IS" with no warranties, and confers no
rights. Please do not send email directly to this alias. This alias is for
newsgroup purposes only.

"Chris" <Ch...@discussions.microsoft.com> wrote in message
news:1B086D7F-49ED-492B...@microsoft.com...

[Chris]

Scott,
Thanks for your reply. I have a few more questions.

1. When or why should I stop cluster node? I'm not quite understand its
purpose. I think it's new to CCR.

2. If I moved an active node to passive node, then stopped cluster on active
node. I guess this will make the node "offline" or not available for the
cluster? If I start the node again does it resume its previous state, i.e.
passive node?

3. I think CCR is using the majority node concept. With just two nodes if
one is stopped is 50% a majority? Is witness also considered as a "node"?

thanks.

[Scott Schnoll [MSFT]]

Stopping and starting a clustered mailbox server (or shutting down or
rebooting a node hosting a clustered mailbox server) is not new to CCR, or
unique to Exchange 2007. It's just part of possible maintenance you might
need to perform. See
http://technet.microsoft.com/en-us/library/aa997676(EXCHG.80).aspx.

Say you have two nodes: NODEA and NODEB. NODEA is currently hosting the
clustered mailbox server (CMS). If you need to perform maintenance on
NODEA, you (1) move the CMS to NODEB, and if necessary, also move the
Cluster Group from NODEA to NODEB. At this point, NODEA becomes your
passive, and NODEB becomes your active. When you restart NODEA, it will
remain the passive node (provided that you have not modified the default
cluster policy settings, such as preferred owner or failback settings).

CCR uses a variant of an MNS quorum called a Majority Node Set with File
Share Witness. In this model, you have two nodes, plus a third quorum voter
which is implemented as a file share on a separate system. The witness is
not considered a node because the term node refers to a member of the
cluster. The server hosting the file share for the file share witness is
not part of the cluster, but rather, a separate server entirely. In fact,
this was one of reasons we came up with the file share witness; we wanted
CCR to be two cluster nodes only but still have a way to maintain quorum
without using a shared disk. The file share witness feature was created so
that you don't need a third node in your cluster. See
http://technet.microsoft.com/en-us/library/bb124521(EXCHG.80).aspx for more
information.

--
Regards,

Scott Schnoll
Microsoft Corporation
This posting is provided "AS IS" with no warranties, and confers no
rights. Please do not send email directly to this alias. This alias is for
newsgroup purposes only.

"Chris" <Ch...@discussions.microsoft.com> wrote in message

news:793DA15B-883F-4092...@microsoft.com...

[kevmcfaul]

Just following on regarding CCR, I would appreciate if someone can answer the following as its not clear from MS articles how this work.

I follow that there is a difference between scheduled or unscheduled failover/outage but what I am not clear on is failback, and re-enabling CCR back to the original passive node.

My understanding is that once you have failed over and made the passive node copy into the active one then only a single active copy of the database exists. CCR replication needs to be stopped and the database re-seeded to what was the active node before failback to the original node. Or is it different when the outage is planned ?

I can't find a definitive answer for this and keep seeing references to failing over and then back after a scheduled outage without explaining how this impacts on the CCR Replication.

The implication seems to be that in a planned outage the operation works similarly to a conventional 2 node cluster on san shared storage in that it can be failed over and then back again and somehow the databse is still current and can replicate again as though nothing had happened. However that doesn't tie in with my understanding of Log shipping if the node is no longer the active database.

Apologies for being so long winded and thanks in advance for any responses.

Regards

Kev

Scott Schnoll [MSFT] wrote:

Stopping and starting a clustered mailbox server (or shutting down or

06-Mar-08

--
Regards,

Previous Posts In This Thread:

On Wednesday, March 05, 2008 6:58 PM
Chri wrote:

stop/start CCR CMS

I have a 2007 sp1 CCR cluster. If I stop the active node by using either
cmdlet or EMC

Q1: will it automatically failover to the passive node? I think it will.

Q2: after maintenance and starting original active node (was stopped early)
will the cluster failback? Or can I started it since there will be two
active nodes.

Thanks.

On Wednesday, March 05, 2008 7:42 PM
Scott Schnoll [MSFT] wrote:

Re: stop/start CCR CMS
Hi Chris:

A1: No, it will not. Failover implies an unscheduled outage. A move, or
what we also call a handoff, is a scheduled outage. So if you need to
perform maintenance, instead of stopping the clustered mailbox server, you
move the clustered mailbox server to the passive node, thereby making it the
new active node.

A2: When you're done with the maintenance, you move the clustered mailbox
server back to the node, and then if necessary perform the same maintenance
on the other node.

Hope this helps.
--
Regards,

Scott Schnoll
Microsoft Corporation
This posting is provided "AS IS" with no warranties, and confers no
rights. Please do not send email directly to this alias. This alias is for
newsgroup purposes only.

"Chris" <Ch...@discussions.microsoft.com> wrote in message
news:1B086D7F-49ED-492B...@microsoft.com...

On Thursday, March 06, 2008 10:50 AM
Chri wrote:

Scott, Thanks for your reply. I have a few more questions.1.
Scott,
Thanks for your reply. I have a few more questions.

1. When or why should I stop cluster node? I'm not quite understand its
purpose. I think it's new to CCR.

2. If I moved an active node to passive node, then stopped cluster on active
node. I guess this will make the node "offline" or not available for the
cluster? If I start the node again does it resume its previous state, i.e.
passive node?

3. I think CCR is using the majority node concept. With just two nodes if
one is stopped is 50% a majority? Is witness also considered as a "node"?

thanks.

"Scott Schnoll [MSFT]" wrote:

On Thursday, March 06, 2008 11:50 AM
Scott Schnoll [MSFT] wrote:

--
Regards,

Submitted via EggHeadCafe - Software Developer Portal of Choice
Internet Explorer 8 Beta 2 on Virtual PC 2007
http://www.eggheadcafe.com/tutorials/aspnet/1350e66d-eef3-4bcc-ace0-a8278cba76ab/internet-explorer-8-beta.aspx

[Ed Crowley [MVP]]

In general, you should be able fail over manually between nodes and back
without having to reseed. One reason is that Move-ClusteredMailboxServer
won't let you fail over unless the logs are current. An unplanned outage
may or may not require reseeding depending on the circumstances.
--
Ed Crowley MVP
"There are seldom good technological solutions to behavioral problems."
.

"Kev McFaul" wrote in message news:2009124124...@hotmail.com...