Replacing failed node with dissimilar hardware

I need to replace a node with a server that is not the same as the failed one. When the cluster was built the hardware was identical down to the positioning of the NIC and HBA. What are the consequences of this? It is a SQL 2000 server SP3a running on Win
dows 2000 Advanced server.
Thanks for any input.
Paul L
Its not a best practice or something that I ever like to do, but it will
work. So no real consequences to speak of.
Cheers,
Rod
"Paul" <paul_lane@.supplyworks.com> wrote in message
news:AE7B3C3A-82B4-4F64-AF85-7F2165B9CD5D@.microsoft.com...
> I need to replace a node with a server that is not the same as the failed
one. When the cluster was built the hardware was identical down to the
positioning of the NIC and HBA. What are the consequences of this? It is a
SQL 2000 server SP3a running on Windows 2000 Advanced server.
> Thanks for any input.
> Paul L
|||Not something I would do in production. But we do it in the lab all the
time. I have a Win2003 3+1 cluster with three different kinds of hardware
from two vendors. It works fine.
Linchi Shea
linchi_shea@.NOSPAMml.com
"Paul" <paul_lane@.supplyworks.com> wrote in message
news:AE7B3C3A-82B4-4F64-AF85-7F2165B9CD5D@.microsoft.com...
> I need to replace a node with a server that is not the same as the failed
one. When the cluster was built the hardware was identical down to the
positioning of the NIC and HBA. What are the consequences of this? It is a
SQL 2000 server SP3a running on Windows 2000 Advanced server.
> Thanks for any input.
> Paul L
|||The major consequence would be if there was a failure and you needed
support from Microsoft. Only cluster configurations on the HCL are
supported.
Rand
This posting is provided "as is" with no warranties and confers no rights.
|||Thanks for all of your input, I appreciate it.

Replacing a cluster node: SQL silent install setup installshield error

Can anyone tell me how to get around this? Installing SQL- no issues, just when applying the sp

I have tried numerous things, including rebuilding the 2nd node again

Trying to install a service pack on SQL 2000 node, and Install shield keeps returning this error, .

[InstallShield Silent]
Version=v5.00.000
File=Log File
[ResponseResult]
ResultCode=-8

I have reset temp directories on both nodes,

I have made user permissions and profiles exist

I have copied install files to both local and remote locations

I know the response – Installshield cannot find the silent response file, but SQL setup creates this on its own.

Short of bringing this production system down and re-installing on both nodes ( I did test this on a non-prod box by the way, no issues), I am at a loss

Well

After 2 weeks of searching, finally!

It turns out, a couple of old log files and a registry key were the culprits..

I found this article this am, and it was the saviour

http://www.sqlteam.com/forums/topic.asp?TOPIC_ID=66737

Specific to my case were:

all mif and logs EVERYWHERE on the replaced node and:

\SOFTWARE\Microsoft\Microsoft SQL Server\80\Need_SP_Key - delete this key (backup 1st of course) on the new node

SP3a went on like a dream, and I can now finally move onto SP4 testing.

replace value of (sq:variable("@variable")) with ""

replace value of(/node/sql:variable("@.varname")/@.Status)[1] with
"pass"'

gives me The XQuery syntax '/function()' is not supported.

Is it not supported?

If it is not supported what is the best way to update a specific node out of many based on certain criteria?

like update Student Grade to A for Student node with name "John" and not any other, without having to use dynamic SQL.

Tim

|||

reproducing answer to this by Pohwan on another discussion alias. This is exactly what I wanted and it works if somebody else is in is same boat

*****************

Workaround,

/node/*[local-name()=sql:variable("@.varname")]/@.Status

instead of

/node/sql:variable("@.varname")/@.Status

--
Pohwan Han. Seoul.

**************

|||

Thanks, that worked wonders for me

replace value of (sq:variable("@variable")) with ""

replace value of(/node/sql:variable("@.varname")/@.Status)[1] with
"pass"'

gives me The XQuery syntax '/function()' is not supported.

Is it not supported?

If it is not supported what is the best way to update a specific node out of many based on certain criteria?

like update Student Grade to A for Student node with name "John" and not any other, without having to use dynamic SQL.

Tim

|||

reproducing answer to this by Pohwan on another discussion alias. This is exactly what I wanted and it works if somebody else is in is same boat

*****************

Workaround,

/node/*[local-name()=sql:variable("@.varname")]/@.Status

instead of

/node/sql:variable("@.varname")/@.Status

--
Pohwan Han. Seoul.

**************

|||

Thanks, that worked wonders for me

replace failed node without downtime? (sp1 question)

Is there a manual procedure whereby I could get a replacement node up to SP1
binaries without having to interrupt service by failing the cluster over
just to apply SP1 to the newly rebuilt node?
Hopefully I'm making sense - the node that went down was a passive node and
it's been rebuilt - but in order to get the binaries up to SP1 the only way
I know how is to fail the cluster to the new node just to get the SP1
binaries on it.
Thanks in advance!
Karl
Hi,
My understanding of your issue is: After adding a new node in SQL 2005
cluster instance, the SQL version on new node is RTM and old node is SP1
In such scenario, we have following two ways to apply SQL Sp1 for this SQL
2005 cluster.
1. Apply SP1 again on the active node (OLD node) and the new node
will be upgraded to SP3 as well.
2. Add registry Need_SP_Key on the passive node and install SQL
service pack 1on Passive node directly, In this way, the SP1 will only be
installed on Passive node.
Hope this helps.
Best regards,
Vincent Xu
Microsoft Online Partner Support
================================================== ====
Get Secure! - www.microsoft.com/security
================================================== ====
When responding to posts, please "Reply to Group" via your newsreader so
that others
may learn and benefit from this issue.
================================================== ====
This posting is provided "AS IS" with no warranties,and confers no rights.
================================================== ====
--[vbcol=seagreen]
SP1[vbcol=seagreen]
and[vbcol=seagreen]
way[vbcol=seagreen]
|||Yes, you understand what I'm trying to do. I have a couple of questions,
please.
If I try #1 on the old (already SP1) node - will it stop the SQL Service or
will it leave the active node online while it patches the passive node
remotely?
For #2 when I search for Need_SP_Key I only get SQL 2000 information and the
2005 BOL has nothing I can find about it. Sounds like the best option
though.
-Karl
"Vincent Xu [MSFT]" <v-xuwen@.online.microsoft.com> wrote in message
news:1EObEyf4GHA.2336@.TK2MSFTNGXA01.phx.gbl...
Hi,
My understanding of your issue is: After adding a new node in SQL 2005
cluster instance, the SQL version on new node is RTM and old node is SP1
In such scenario, we have following two ways to apply SQL Sp1 for this SQL
2005 cluster.
1. Apply SP1 again on the active node (OLD node) and the new node
will be upgraded to SP3 as well.
2. Add registry Need_SP_Key on the passive node and install SQL
service pack 1on Passive node directly, In this way, the SP1 will only be
installed on Passive node.
Hope this helps.
Best regards,
Vincent Xu
Microsoft Online Partner Support
================================================== ====
Get Secure! - www.microsoft.com/security
================================================== ====
When responding to posts, please "Reply to Group" via your newsreader so
that others
may learn and benefit from this issue.
================================================== ====
This posting is provided "AS IS" with no warranties,and confers no rights.
================================================== ====
--[vbcol=seagreen]
SP1[vbcol=seagreen]
and[vbcol=seagreen]
way[vbcol=seagreen]
|||Hi,
Found more information:
You can even install SP1 on node 2 directly:
- Verified that SP1 was successfully installed on Node1, we did that by
running
Select @.@.Version & physically checking the File Version property of
sqlservr.exe on
Node1 which is 2005.90.2047.0
- We made Node1 as a active node.
- Started installation on Node2.
Note: This time setup wont allow us to check the checkbox for instance in
"Feature
Selection" window, as Node1 (which is a active node) is already on SP1, so
we will
continue so we continued setup without it.
- Let it run for all products till setup got completed.
- Verified that SP1 was successfully installed on Node2, we did that by
running
Select @.@.Version & physically checking the File Version property of
sqlservr.exe on
Node2 which is 2005.90.2047.0
- Verified by failing over also.
Hope this helps.
Best regards,
Vincent Xu
Microsoft Online Partner Support
================================================== ====
Get Secure! - www.microsoft.com/security
================================================== ====
When responding to posts, please "Reply to Group" via your newsreader so
that others
may learn and benefit from this issue.
================================================== ====
This posting is provided "AS IS" with no warranties,and confers no rights.
================================================== ====
--[vbcol=seagreen]
<1EObEyf4GHA.2336@.TK2MSFTNGXA01.phx.gbl>[vbcol=seagreen]
or[vbcol=seagreen]
the[vbcol=seagreen]

Replace Cluster Disks

Hello,

I have a two node SQL 2000 cluster running on windows 2003 enterprise server. We need to replace the SAN disks. Can we not disable SQL service & Cluster service, copy the contents from existing disks to target disks, swap the drive letter & start the services?

What is the best practice to do this? Appreciate your help.

Thanks
Rajan

That 'should' work just fine. But STOP the SQL services, including SQL Agent and SQL Browser that depend upon the drive.|||

It was not that straight forward. Cluster does not use drive letters for the volume....it uses the disk signatures.

I had to use clusterrecovery.exe utility from resource kit. It is a great tool.

Regards
Rajan

|||

This technique will work just fine. I've used it many times.

In the cluster administrator stop the SQL Server and the SQL Agent.

Attach the new luns to the servers, and make them cluster resources.

Copy ALL the files and folders from the old drive to the new drive.

Switch out the drive letters so that everything looks eactly like before (just with the new drives).

Update the cluster resources lables for the drives so that they match the correct new drive letters.

Change the dependencies on the SQL Server cluster resource so that it now depends on the correct drive letters.

Startup the SQL Server and SQL Agent services in the cluster administrator.

All should be working well after that.

Replace Cluster Disks

Hello,

I have a two node SQL 2000 cluster running on windows 2003 enterprise server. We need to replace the SAN disks. Can we not disable SQL service & Cluster service, copy the contents from existing disks to target disks, swap the drive letter & start the services?

What is the best practice to do this? Appreciate your help.

Thanks
Rajan

That 'should' work just fine. But STOP the SQL services, including SQL Agent and SQL Browser that depend upon the drive.|||

It was not that straight forward. Cluster does not use drive letters for the volume....it uses the disk signatures.

I had to use clusterrecovery.exe utility from resource kit. It is a great tool.

Regards
Rajan

|||

This technique will work just fine. I've used it many times.

In the cluster administrator stop the SQL Server and the SQL Agent.

Attach the new luns to the servers, and make them cluster resources.

Copy ALL the files and folders from the old drive to the new drive.

Switch out the drive letters so that everything looks eactly like before (just with the new drives).

Update the cluster resources lables for the drives so that they match the correct new drive letters.

Change the dependencies on the SQL Server cluster resource so that it now depends on the correct drive letters.

Startup the SQL Server and SQL Agent services in the cluster administrator.

All should be working well after that.