Error on MSDTC after a SQL Server fail-over   

Hello

I have a SQL Server 2000 64 bits running on an two node cluster (NODE1 and 
NODE2) with two SQL Server instances (SQL1 and SQL1\2).
Instance SQL1 is running on NODE1 and instance SQL1\2 is running on NODE2.
MSDTC is configured on a exclusive group and it is running on NODE1.
The Cluster Group also is running on NODE1.
I use a lot of distributed transactions between SQL1\2 and SQL1 and it works 
just fine.

I had a problem last night with NODE1, and all the resources (Cluster, SQL1 
and MSDTC) failed to NODE2.
After the failover I started getting the error below when trying to execute 
a distributed transaction from SQL1\2 to SQL1.
Error: 8510, Severity: 20, State: 2
Enlist of MSDTC transaction failed: XACT_E_CONNECTION_DOWN.

The error happened during the failover and also after the finish of the 
failver.
On SQL1 error log I notice that there was no error messages when it 
initialize the MSDTC. Below is the message from the ErrorLog.
"Attempting to initialize Distributed Transaction Coordinator."

To solve the Error I brought up NODE1 back on-line (it failed because of a 
network problem). I move the MSDTC back to NODE1, move SQL1 back to NODE1 and 
also re-started SQL1\2 on NODE2 (brought offline, bring on-line). After that 
procedure I was able to use distributed transactions between SQL1 and SQL1\2.

I would like to know why that problem happened? 
I suposed that MSDTC was able to work on cluster and that the MSDTC should 
work after a failover.
I would like your help to understand the problem. If you need more 
information just let me know.

Regards.

-- 
Carlos Eduardo Selonke de Souza
http://carlos.geekbunker.org

date: Mon, 16 May 2005 05:59:01 -0700   author:   Carlos Selonke

RE: Error on MSDTC after a SQL Server fail-over   

Hi Carlos

MSDTC has to be running when SQL is starting. So it's a timing problem. To 
be certain, check the log in SQL, and look for messages about finding the 
Transaction "Server". It probably states that is was down. SQL will not retry 
it afterwards.
If the MSDTC resource was in the same group as the SQL Virtual Server, you 
could make SQL dependant of MSDTC, but unfortunately this is not the case.



-- 
Chau chau,

  Pascalos


"Carlos Selonke" wrote:

> Hello
> 
> I have a SQL Server 2000 64 bits running on an two node cluster (NODE1 and 
> NODE2) with two SQL Server instances (SQL1 and SQL1\2).
> Instance SQL1 is running on NODE1 and instance SQL1\2 is running on NODE2.
> MSDTC is configured on a exclusive group and it is running on NODE1.
> The Cluster Group also is running on NODE1.
> I use a lot of distributed transactions between SQL1\2 and SQL1 and it works 
> just fine.
> 
> I had a problem last night with NODE1, and all the resources (Cluster, SQL1 
> and MSDTC) failed to NODE2.
> After the failover I started getting the error below when trying to execute 
> a distributed transaction from SQL1\2 to SQL1.
> Error: 8510, Severity: 20, State: 2
> Enlist of MSDTC transaction failed: XACT_E_CONNECTION_DOWN.
> 
> The error happened during the failover and also after the finish of the 
> failver.
> On SQL1 error log I notice that there was no error messages when it 
> initialize the MSDTC. Below is the message from the ErrorLog.
> "Attempting to initialize Distributed Transaction Coordinator."
> 
> To solve the Error I brought up NODE1 back on-line (it failed because of a 
> network problem). I move the MSDTC back to NODE1, move SQL1 back to NODE1 and 
> also re-started SQL1\2 on NODE2 (brought offline, bring on-line). After that 
> procedure I was able to use distributed transactions between SQL1 and SQL1\2.
> 
> I would like to know why that problem happened? 
> I suposed that MSDTC was able to work on cluster and that the MSDTC should 
> work after a failover.
> I would like your help to understand the problem. If you need more 
> information just let me know.
> 
> Regards.
> 
> -- 
> Carlos Eduardo Selonke de Souza
> http://carlos.geekbunker.org

date: Wed, 18 May 2005 03:21:03 -0700   author:   Pascal Damman