"RG" <RG@discussions.microsoft.com> wrote in message
news:CB56387E-6EDD-46BD-BD9F-8654BD4E3A0C@microsoft.com...
>I have a three node cluster (Windows 2003 ENT with SQL 2000 SP4). We are
> going to be applying Cumulative Hotfix for SQL Server Service Pack 4 Build
> 2187 (KB916287). I'm having a disagreement with our DBA. He says that
> the
> entire cluster and all named instances need to come down on all three
> nodes
> to install this patch. I don't think they have to. I'm under the
> assumption
> that we can just move all named instances off the node we want to patch,
> apply the patch, reboot the machine, and then move the named instances
> back
> to this node. Then we can do the other nodes.
>
> Additionally, our DBA is under the impression that all we need to do is
> install the patch on one node and that it will install on the other nodes.
> I
> said that is not the case. You have to apply the patch on every node in
> the
> cluster because the install of SQL isn't clustered. It's just the
> databases/logs/quorom that is clustered not SQL itself.
>
> Does anybody have any input on this?

"RG" <RG@discussions.microsoft.com> wrote in message
news:F151C0B9-A416-417B-8A0A-DD44E71CC06A@microsoft.com...
> Unfortunately I can't see a readme for this patch.
>
> "Geoff N. Hiten" wrote:
>
>> Your DBA and you are both right and both wrong.
>>
>> SQL hotfixes are cluster-aware, just like a SQL service pack. The
>> installer
>> will update binaries on all nodes and will update the database on the
>> host
>> node. You must start the hotfix installer from the node that currently
>> owns
>> the instance you wish to update. SQL uses local binaries on each node,
>> but
>> the installer knows this and will handle things appropriately across the
>> cluster. The only time you run a local binary only update is when
>> rebuilding a failed node. The Service Pack and Hotfix README files
>> explain
>> how and when to do this.
>>
>> That leads us where you are both wrong. SQL hotfixes are applied to each
>> instance independently. During the installation process,the installer
>> takes
>> the instance offline. You may have to reboot the cluster host nodes
>> after
>> installation, but the installer will alert you if that is required. Best
>> practices suggest a rolling reboot of all nodes prior to any SP, hotfix,
>> or
>> OS patch to ensure a clean system.
>>
>> If you have multiple instances, you will need to apply the hotfix to each
>> instance independently. Note that Microsoft does support running
>> different
>> instances at different patch and even version levels on the same cluster,
>> so
>> you don't have to update all the instances at once.
>>
>> --
>> Geoff N. Hiten
>> Senior Database Administrator
>> Microsoft SQL Server MVP
>>
>>
>>
>>
>> "RG" <RG@discussions.microsoft.com> wrote in message
>> news:CB56387E-6EDD-46BD-BD9F-8654BD4E3A0C@microsoft.com...
>> >I have a three node cluster (Windows 2003 ENT with SQL 2000 SP4). We
>> >are
>> > going to be applying Cumulative Hotfix for SQL Server Service Pack 4
>> > Build
>> > 2187 (KB916287). I'm having a disagreement with our DBA. He says that
>> > the
>> > entire cluster and all named instances need to come down on all three
>> > nodes
>> > to install this patch. I don't think they have to. I'm under the
>> > assumption
>> > that we can just move all named instances off the node we want to
>> > patch,
>> > apply the patch, reboot the machine, and then move the named instances
>> > back
>> > to this node. Then we can do the other nodes.
>> >
>> > Additionally, our DBA is under the impression that all we need to do is
>> > install the patch on one node and that it will install on the other
>> > nodes.
>> > I
>> > said that is not the case. You have to apply the patch on every node
>> > in
>> > the
>> > cluster because the install of SQL isn't clustered. It's just the
>> > databases/logs/quorom that is clustered not SQL itself.
>> >
>> > Does anybody have any input on this?
>>
>>
>>