Removing a SQL cluster node

Could someone point me to some documentation on the best way to remove a
node from a SQL cluster.
It's a two node, two instance cluster, and I plan to rebuild one node, fail
to the rebuilt one, then rebuild the remaining node.
-Karl
From the BOL:
"In this scenario, failure is caused by hardware failure in Node 1 of a
two-node cluster. This hardware failure could be caused, for example, by the
failure of a small computer system interface (SCSI) card or the operating
system.
1.. After Node 1 fails, the Microsoft SQL Server 2000 failover cluster
fails over to Node 2.
2.. Run SQL Server Setup and remove Node 1. For more information, see How
to remove a failover clustered instance .
3.. Evict Node 1 from Microsoft Cluster Service (MSCS). To evict a node
from MSCS, from Node 2, right-click on the node to remove, and then click
Evict Node.
4.. Install new hardware to replace the failed hardware in Node 1.
5.. Install the operating system. For more information about which
operating system to install and specific instructions on how to do this, see
Before Installing Failover Clustering.
6.. Install MSCS and join the existing cluster. For more information, see
Before Installing Failover Clustering.
7.. Run the Setup program on Node 2 and add Node 1 back to the failover
cluster. For more information, see How to add nodes to an existing virtual
server (Setup). "
Tom
Thomas A. Moreau, BSc, PhD, MCSE, MCDBA
SQL Server MVP
Columnist, SQL Server Professional
Toronto, ON Canada
www.pinpub.com
..
"Karl" <karl@.nospam.postalias> wrote in message
news:uzFVttDfFHA.3936@.TK2MSFTNGP14.phx.gbl...
Could someone point me to some documentation on the best way to remove a
node from a SQL cluster.
It's a two node, two instance cluster, and I plan to rebuild one node, fail
to the rebuilt one, then rebuild the remaining node.
-Karl
|||In addition, you will need to re-run any Service Packs and HotFixes on the
new node to update the binaries. Read the README file included with the
service pack for detailed instructions on how to apply the SP to a newly
added cluster node.
Geoff N. Hiten
Microsoft SQL Server MVP
"Tom Moreau" <tom@.dont.spam.me.cips.ca> wrote in message
news:%23udWQyDfFHA.1412@.TK2MSFTNGP09.phx.gbl...
> From the BOL:
> "In this scenario, failure is caused by hardware failure in Node 1 of a
> two-node cluster. This hardware failure could be caused, for example, by
> the
> failure of a small computer system interface (SCSI) card or the operating
> system.
> 1.. After Node 1 fails, the Microsoft SQL Server 2000 failover cluster
> fails over to Node 2.
>
> 2.. Run SQL Server Setup and remove Node 1. For more information, see How
> to remove a failover clustered instance .
>
> 3.. Evict Node 1 from Microsoft Cluster Service (MSCS). To evict a node
> from MSCS, from Node 2, right-click on the node to remove, and then click
> Evict Node.
>
> 4.. Install new hardware to replace the failed hardware in Node 1.
>
> 5.. Install the operating system. For more information about which
> operating system to install and specific instructions on how to do this,
> see
> Before Installing Failover Clustering.
>
> 6.. Install MSCS and join the existing cluster. For more information, see
> Before Installing Failover Clustering.
>
> 7.. Run the Setup program on Node 2 and add Node 1 back to the failover
> cluster. For more information, see How to add nodes to an existing virtual
> server (Setup). "
> --
> Tom
> ----
> Thomas A. Moreau, BSc, PhD, MCSE, MCDBA
> SQL Server MVP
> Columnist, SQL Server Professional
> Toronto, ON Canada
> www.pinpub.com
> .
> "Karl" <karl@.nospam.postalias> wrote in message
> news:uzFVttDfFHA.3936@.TK2MSFTNGP14.phx.gbl...
> Could someone point me to some documentation on the best way to remove a
> node from a SQL cluster.
> It's a two node, two instance cluster, and I plan to rebuild one node,
> fail
> to the rebuilt one, then rebuild the remaining node.
> -Karl
>
|||I'm unclear about step #2. Following the BOL topic "How to remove a failover
clustered instance" appears to lead me down the path of removing the
instance entirely from the virtual server. The virtual server/instance is
working fine on node 2. I'm afraid to proceed because the prompts appear to
indicate that I'd be uninstalling the entire instance from the virtual
server.
Is it possible that if a node fails, there is no need to run SQL setup? Just
to go remaining node's MSCS and evict the unresponsive node (regardless if
its still online or not), reinstall and rejoin MSCS?
"Tom Moreau" <tom@.dont.spam.me.cips.ca> wrote in message
news:%23udWQyDfFHA.1412@.TK2MSFTNGP09.phx.gbl...
> From the BOL:
> "In this scenario, failure is caused by hardware failure in Node 1 of a
> two-node cluster. This hardware failure could be caused, for example, by
> the
> failure of a small computer system interface (SCSI) card or the operating
> system.
> 1.. After Node 1 fails, the Microsoft SQL Server 2000 failover cluster
> fails over to Node 2.
>
> 2.. Run SQL Server Setup and remove Node 1. For more information, see How
> to remove a failover clustered instance .
>
> 3.. Evict Node 1 from Microsoft Cluster Service (MSCS). To evict a node
> from MSCS, from Node 2, right-click on the node to remove, and then click
> Evict Node.
>
> 4.. Install new hardware to replace the failed hardware in Node 1.
>
> 5.. Install the operating system. For more information about which
> operating system to install and specific instructions on how to do this,
> see
> Before Installing Failover Clustering.
>
> 6.. Install MSCS and join the existing cluster. For more information, see
> Before Installing Failover Clustering.
>
> 7.. Run the Setup program on Node 2 and add Node 1 back to the failover
> cluster. For more information, see How to add nodes to an existing virtual
> server (Setup). "
> --
> Tom
> ----
> Thomas A. Moreau, BSc, PhD, MCSE, MCDBA
> SQL Server MVP
> Columnist, SQL Server Professional
> Toronto, ON Canada
> www.pinpub.com
> .
> "Karl" <karl@.nospam.postalias> wrote in message
> news:uzFVttDfFHA.3936@.TK2MSFTNGP14.phx.gbl...
> Could someone point me to some documentation on the best way to remove a
> node from a SQL cluster.
> It's a two node, two instance cluster, and I plan to rebuild one node,
> fail
> to the rebuilt one, then rebuild the remaining node.
> -Karl
>
|||Hi Karl,
In step#2 we should refer to the "How to remove a node from an existing
failover cluster" topic:
http://msdn.microsoft.com/library/en...ering_7hyq.asp
It appears that "see How to remove a failover clustered instance" in the
"How to recover from failover cluster failure in Scenario 1" topic is
incorrect, I've sent a feedback to the appropriate team. The remain part is
correct and we need to follow every step.
Please let me know if you have any other questions.
Sincerely,
William Wang
Microsoft Online Partner Support
When responding to posts, please "Reply to Group" via your newsreader so
that others may learn and benefit from your issue.
This posting is provided "AS IS" with no warranties, and confers no rights.
|||It does appear ambigious in the docs. When you remove SQL, you will remove
it from the dead node only. That will clean up the SQL clustering
information on the remaining node(s). Once the new or newly
rebuild/repaired node is rejoined to the cluster, you still have to install
SQL on it. This means installing local binaries and configuring the cluster
to recognize it as a possible host for the SQL resource group. That is what
the SQL installation steps do. It is not enough to just add the node to the
cluster.
Again, don't forget to apply any service packs or hotfixes after adding the
node back.
BUG: Correcting Mismatched Binaries on SQL Clustered Virtual Server
http://support.microsoft.com/default...b;en-us;295589
I agree that the docs are incorrect in that they do not describe how to
remove a node from a SQL cluster without uninstalling the entire virtual
server . Since William already reported it, I will leave it alone.
Geoff N. Hiten
Senior Database Administrator
Microsoft SQL Server MVP
"Karl" <karl@.nospam.postalias> wrote in message
news:eXQQcFAiFHA.1244@.TK2MSFTNGP14.phx.gbl...
> I'm unclear about step #2. Following the BOL topic "How to remove a
> failover clustered instance" appears to lead me down the path of removing
> the instance entirely from the virtual server. The virtual server/instance
> is working fine on node 2. I'm afraid to proceed because the prompts
> appear to indicate that I'd be uninstalling the entire instance from the
> virtual server.
> Is it possible that if a node fails, there is no need to run SQL setup?
> Just to go remaining node's MSCS and evict the unresponsive node
> (regardless if its still online or not), reinstall and rejoin MSCS?
> "Tom Moreau" <tom@.dont.spam.me.cips.ca> wrote in message
> news:%23udWQyDfFHA.1412@.TK2MSFTNGP09.phx.gbl...
>
|||Thanks to Geoff and William for pointing me to the right procedures. I've
rebuilt the node from scratch and have added it back to the cluster. I have
a question related to patching and mismatched binaries - and avoiding
reboots.
I want to apply SQL2000-KB815495-8.00.0818-ENU.exe hotfix to the new node
but I don't want to reboot the good/production node which is currently
hosting the instance.
I attempt to run the hotfix on the passive node but it says it has to run on
the active node. Attempting the install on the active node reports that it
will take the resource offline.
Is it possible to copy binaries into the passive node so as not to incur any
SQL downtime? So far the entire rebuild of this passive node has had no
effect on the active production instance. I'd hate to ruin the record this
late in the game!
-Karl
"Geoff N. Hiten" <sqlcraftsman@.gmail.com> wrote in message
news:OV61RWHiFHA.1464@.TK2MSFTNGP14.phx.gbl...
It does appear ambigious in the docs. When you remove SQL, you will remove
it from the dead node only. That will clean up the SQL clustering
information on the remaining node(s). Once the new or newly
rebuild/repaired node is rejoined to the cluster, you still have to install
SQL on it. This means installing local binaries and configuring the cluster
to recognize it as a possible host for the SQL resource group. That is what
the SQL installation steps do. It is not enough to just add the node to the
cluster.
Again, don't forget to apply any service packs or hotfixes after adding the
node back.
BUG: Correcting Mismatched Binaries on SQL Clustered Virtual Server
http://support.microsoft.com/default...b;en-us;295589
I agree that the docs are incorrect in that they do not describe how to
remove a node from a SQL cluster without uninstalling the entire virtual
server . Since William already reported it, I will leave it alone.
Geoff N. Hiten
Senior Database Administrator
Microsoft SQL Server MVP
"Karl" <karl@.nospam.postalias> wrote in message
news:eXQQcFAiFHA.1244@.TK2MSFTNGP14.phx.gbl...
> I'm unclear about step #2. Following the BOL topic "How to remove a
> failover clustered instance" appears to lead me down the path of removing
> the instance entirely from the virtual server. The virtual server/instance
> is working fine on node 2. I'm afraid to proceed because the prompts
> appear to indicate that I'd be uninstalling the entire instance from the
> virtual server.
> Is it possible that if a node fails, there is no need to run SQL setup?
> Just to go remaining node's MSCS and evict the unresponsive node
> (regardless if its still online or not), reinstall and rejoin MSCS?
> "Tom Moreau" <tom@.dont.spam.me.cips.ca> wrote in message
> news:%23udWQyDfFHA.1412@.TK2MSFTNGP09.phx.gbl...
>
|||The old installer sometimes let you do a "binary-only upgrade" to a node
that is not currently hosting the target SQL instance. That has caused
issues in the past and was removed at build 977 when the new hotfix
installer debuted. A hotfix typically takes about 5-10 minutes to run and
yes, the instance does have to go offline.
Geoff N. Hiten
Senior Database Administrator
Microsoft SQL Server MVP
"Karl" <karl@.nospam.postalias> wrote in message
news:udt6OmYrFHA.3788@.TK2MSFTNGP12.phx.gbl...
> Thanks to Geoff and William for pointing me to the right procedures. I've
> rebuilt the node from scratch and have added it back to the cluster. I
> have
> a question related to patching and mismatched binaries - and avoiding
> reboots.
> I want to apply SQL2000-KB815495-8.00.0818-ENU.exe hotfix to the new node
> but I don't want to reboot the good/production node which is currently
> hosting the instance.
> I attempt to run the hotfix on the passive node but it says it has to run
> on
> the active node. Attempting the install on the active node reports that it
> will take the resource offline.
> Is it possible to copy binaries into the passive node so as not to incur
> any
> SQL downtime? So far the entire rebuild of this passive node has had no
> effect on the active production instance. I'd hate to ruin the record this
> late in the game!
> -Karl
>
> "Geoff N. Hiten" <sqlcraftsman@.gmail.com> wrote in message
> news:OV61RWHiFHA.1464@.TK2MSFTNGP14.phx.gbl...
> It does appear ambigious in the docs. When you remove SQL, you will
> remove
> it from the dead node only. That will clean up the SQL clustering
> information on the remaining node(s). Once the new or newly
> rebuild/repaired node is rejoined to the cluster, you still have to
> install
> SQL on it. This means installing local binaries and configuring the
> cluster
> to recognize it as a possible host for the SQL resource group. That is
> what
> the SQL installation steps do. It is not enough to just add the node to
> the
> cluster.
> Again, don't forget to apply any service packs or hotfixes after adding
> the
> node back.
> BUG: Correcting Mismatched Binaries on SQL Clustered Virtual Server
> http://support.microsoft.com/default...b;en-us;295589
> I agree that the docs are incorrect in that they do not describe how to
> remove a node from a SQL cluster without uninstalling the entire virtual
> server . Since William already reported it, I will leave it alone.
> Geoff N. Hiten
> Senior Database Administrator
> Microsoft SQL Server MVP
> "Karl" <karl@.nospam.postalias> wrote in message
> news:eXQQcFAiFHA.1244@.TK2MSFTNGP14.phx.gbl...
>
>
|||To clarify, when adding a node to an existing virtual server, the SQL
Instance will come OffLine for about 5-10 minutes while the binaries are
written? Thanks for the input as we may be replacing a node as well.
John
Geoff N. Hiten wrote:[vbcol=seagreen]
>The old installer sometimes let you do a "binary-only upgrade" to a node
>that is not currently hosting the target SQL instance. That has caused
>issues in the past and was removed at build 977 when the new hotfix
>installer debuted. A hotfix typically takes about 5-10 minutes to run and
>yes, the instance does have to go offline.
>[quoted text clipped - 106 lines]
|||Yes. The hotfix will be applied to all nodes and any update scripts will
get re-run. The instance may stop and start several times in single-user
mode to apply updates. I have found that if you have a very large memory
system (16GB or more), it speeds things up considerably to lower the memory
settings before applying the hotfix and increase them after it has
completed.
Geoff N. Hiten
Senior Database Administrator
Microsoft SQL Server MVP
"John via droptable.com" <forum@.droptable.com> wrote in message
news:53AB45D140ED3@.droptable.com...[vbcol=seagreen]
> To clarify, when adding a node to an existing virtual server, the SQL
> Instance will come OffLine for about 5-10 minutes while the binaries are
> written? Thanks for the input as we may be replacing a node as well.
> John
> Geoff N. Hiten wrote: