Trying to Move Group - SQL Instance Won't Start

#**101**

On Apr 4, 11:28*am, "Geoff N. Hiten" <SQLCrafts... (AT) gmail (DOT) com> wrote:

Quote:

OK. *The SQL installer thinks there are SQL binaries/etc on Nodes C and D
for ISNo5 and INS06. *This is most likely to some "hand-hacking" of allowed
nodes. *Then let the SQL installer go ahead and remove SQL from Nodes C and
D. *It doesn't work there anyway. *Once you have SQL convinced that Nodes C
and D are not allowed nodes, you can then re-install SQL to those nodes,
again using the maintenance mode of the installer.

There may be some issues patching the binaries if you are on a post-sp
hotfix, but I am working on developing a workaround for that issue. *I will
be testing it in a maintenance window on Tuesday evening and will write upa
blog post shortly afterword if it works as expected.

--
Geoff N. Hiten
Senior SQL Infrastructure Consultant
Microsoft SQL Server MVP

edsm... (AT) pascocountyfl (DOT) net> wrote in message

news:65b11331-b552-4fae-b949-edd9845cd362 (AT) m3g2000hsc (DOT) googlegroups.com...
On Apr 3, 10:20 am, "Geoff N. Hiten" <SQLCrafts... (AT) gmail (DOT) com> wrote:

There is another solution.

You can do a maintenance install and add the new nodes. You will have to
undo the manual allowed nodes settings and then follow the instructions in
"maintaining a failover cluster" for recovery from failure scenario 1. I
have added nodes to a cluster and made this work. Just be sure and
re-apply
service packs and hotfixes to each instance. Be ready for a lot of
reboots.

SQL 2000 on Windows Server 2003 allows up to four nodes in a cluster so
you
should be good to go.

--
Geoff N. Hiten
Senior SQL Infrastructure Consultant
Microsoft SQL Server MVP

edsm... (AT) pascocountyfl (DOT) net> wrote in message

news:c76f2b1f-37fd-4e29-9820-f8daac4d5359 (AT) d1g2000hsg (DOT) googlegroups.com...
On Mar 31, 9:58 am, edsm... (AT) pascocountyfl (DOT) net wrote:

On Mar 28, 9:45 am, "Geoff N. Hiten" <SQLCrafts... (AT) gmail (DOT) com> wrote:

It sounds like when you added the nodes, you manually added the new
nodes to
the "allowed nodes" list for that instance and did not run the
installer
to
put the binaries down on the new nodes. This is what I would expect to
see
after adding nodes without doing the install properly.

Did you follow the instructions in "Maintaining a Failover Cluster" in
BOL
on recovering from a cluster node failure?

--
Geoff N. Hiten
Senior SQL Infrastructure Consultant
Microsoft SQL Server MVP

edsm... (AT) pascocountyfl (DOT) net> wrote in message

news:9c148020-46b5-4251-8e5e-c021ba6bc552 (AT) e67g2000hsa (DOT) googlegroups.com...
On Mar 27, 3:48 pm, "Tom Moreau" <t... (AT) dont (DOT) spam.me.cips.ca> wrote:

Are you running theses instances with domain account that does not
have
the
appropriate privileges on the other nodes?

--
Tom

----------------------------------------------------
Thomas A. Moreau, BSc, PhD, MCSE, MCDBA, MCITP, MCTS
SQL Server MVP
Toronto, ON
Canadahttps://mvp.support.microsoft.com/profile/Tom.Moreau

edsm... (AT) pascocountyfl (DOT) net> wrote in message

news:61340dcb-4f73-48bd-8a8f-581c1a0a3e88 (AT) 8g2000hsu (DOT) googlegroups.com...
Server Environment:
N + 1 (Four Nodes), IBM X-Series, 32-bit, 4 processors, 16 GB RAM
SQL 2000, SP4 with hotfixes, AWE and PAE configured

The SQL cluster was originally built with 2 nodes (node-B is the
passive failover node). Months later we added nodes C & D. Nodes A,
C and D each have two SQL instances assigned and all failover to
node
B. Everything worked great for months.
Now we have plans to evict node-D from the cluster for other usage..
So
I decreased max server memory on all instances, changed "Possible
Owners" and did a "Move Group" to move SQL instance 3 to node-A.
This
worked fine.
However, when I tried to move instances 5 & 6 from node-D to node-C,
the SQL Server resouce fail to start. (Instance 4 is currently on
node-
C all by itself).
I then deselected the "Affect the Group" option for all resources
for
SQL instance 6. Now the "Move Group" actually moved the resources to
node-C, but the SQL Server resource still fails to start. I can move
it back to node-D just fine.

Here's the error from the Cluster.log:

ERR SQL Server <SQL Server (INS06)>: [sqsrvres] OnlineThread:
RegOpenKeyExW failed (status 2)
ERR SQL Server <SQL Server (INS06)>: [sqsrvres] OnlineThread: Error
2
bringing resource online.
INFO [RM] RmpSetResourceStatus, Posting state 4 notification for
resource <SQL Server (INS06)

We've been trying to troubleshoot for weeks now... we can move SQL
instances 5 & 6 to node-B just fine.
Please help--Thanks

The same domain account is used for all SQL services and for all
instances across all nodes.

Ed- Hide quoted text -

- Show quoted text -

Thanks Geoff, I'll be discussing with the server team.- Hide quoted
text -

- Show quoted text -

When we added nodes C & D to the cluster, the "possible owners" (In
Cluster Administrator) for these nodes were set to only include
themselves (Node C or D) and the failover node (Node B in our case).
Consequently, the SQL binaries were only installed on these two nodes.
Now I want to move the two virtual SQL instances (INS05 and INS06)
from node D to node C but as Geoff pointed out, the binaries for the
node-D instances don't exist on node-C. (Node-C has instances INS04
and INS05 installed)
I see no way to legitimately installing instance INS05 and INS06 to
node-C without first uninstalling these virtual SQL instances from the
nodes their currently on (node-D and failover node-B).
Am I correct, or is their another solution?- Hide quoted text -

- Show quoted text -

Geoff, I have read all of the references and have tried the maint
install several times. But based on your instructions, I am just not
getting it. Here's where I'm getting hung up:
- Again, I have four node cluster which includes one standby node.
When I refer to a "node" I mean physical server. I have two virtual
SQL servers/instances on each node.
- I'm trying to move the two virtual SQL servers/instances (INS05 and
INS06) from node-D to node-C but it won't work because the SQL
directories/binaries for INS05 and INS06 never got installed on node-C
because we never intended to fail those instances over to that node.
(However, node-C does have SQL binaries for INS03 & INS04)
- The SQL maint install will not let my install INS05 or INS06 on Node-
C. I only have the option to remove the virtual SQL server.
- Even if I were to evict the node, I can't see how I would be able to
install these two SQL instances while the node has been removed from
the cluster.
- The BOL "...failure scenario 1" mainly addresses the reinstalling of
the Windows OS and MSCS and then rejoining to the cluster.
- This has got to be a fairly common need in that when you lose a node
(hardware failure, etc.) you pretty much would need to do what I'm
trying to do now. But really don't see instructions anywhere that
specifically addresses the how to install either or a new node or an
existing node when the SQL Virtual Server/instance already exists on
other nodes already.
- Again, I can not see anyway this would work without first
uninstalling these instances from the existing nodes (active and
standby nodes).- Hide quoted text -

- Show quoted text -

Yes, It seems like removing these instances is the direction we'll
have to take. And I'll be keeping an eye out for the patch workaround
you mention. Geoff, as always, thanks a bunch for your valuable and
timely contributions.

Thread Tools
Show Printable Version Email this Page
Display Modes
Linear Mode Switch to Hybrid Mode Switch to Threaded Mode

Trying to Move Group - SQL Instance Won't Start

microsoft.public.sqlserver.clustering microsoft.public.sqlserver.clustering

Re: Trying to Move Group - SQL Instance Won't Start - 04-04-2008 , 02:54 PM