MSDTC fail on ActivePasive cluster | SQL Server Performance Forums

SQL Server Performance Forum – Threads Archive

MSDTC fail on ActivePasive cluster

Hello, we have win2k3 sp1 and SQL2000 SP4.
The problem is that I cann’t to force MSDTC to come online. The short review of what I do and what I get:
1. following KB, after enabling network DTC access I find MSDTC resource installed in the cluster group with Q disk and cluster network name dependencies. There is suggestion that it would be better if we put MSDTC resource in separate group with MSDTC dedicated disk, MSDTC network name, MSDTC IP address. So I stop this new created MSDTC resource in the cluster group, delete it, and delete Q:msdtclog folder.
2. Foloowing 301600KB I create separate MSDTC group with all 4 resources, but MSDTC resource fails to come online. I check this new physical disk and find out that there is no msdtclog folder created. What I’m doing wrong In the event log I find the next error: EventID 4389
MS DTC setup failed to populate the shared cluster registry. The DTC resource will not be able to come online.Error Specifics: d:srvrtmcomcomplusdtcsharedmtxclumtxclusetuphelper.cpp:120, Pid: 2068, CmdLine: "C:WINDOWScluster
esrcmon.exe" -e 944 -m 984 -p 3540 Does anyone has any ideas? See the section "Configuring the Microsoft Distributed Transaction Coordinator"
PLease check other posts related to MSDTC, specifically search for Argyle posts for a solution. Satya SKJ
This posting is provided “AS IS” with no rights for the sake of knowledge sharing.
Make sure your cluster account is a member of the administrators group and can write to the registry keys. For example you’re NT administrators might be pushing out a domain security policy that prevents you from completing the install. There is a solution in this post on how to reinstall MSDTC and solve your problem:…read/90a4a22c91f589a7/20ad175bde6a0ddb?lnk=st Copied below:
Delete the DTC resource
Delete MSDTC shared folder from the quorum disk
On Node 1:
* Stop Cluster service
* Remove Enable network DTC access
* Run msdtc -uninstall
* Verify success in event log
* Delete registry keys:
* Reboot
* Reinstall Enable network DTC access (at this point, the DTC resource
should be created and come online)
Node 2
* Stop Cluster service
* Evict node from cluster
* Remove Enable network DTC access
* Run msdtc -uninstall
* Verify success in event log
* Delete registry keys
* Reboot
* Reinstall Enable network DTC access
* Verify service started, registry keys created
* Rejoin cluster
Then fail the cluster group over to the second node to verify that the DTC
resource comes online successfully on the second node. After that, reboot
both servers and verify that the resource would fail over properly to both
nodes with the rest of the group.

Scott Schnoll
This posting is provided "AS IS" with no warranties, and confers no
rights. Please do not send email directly to this alias. This alias is for
purposes only.
Thank you for your posts!
Argyle, I wanted to check couple points: 1. What cluster service startup type should be when taking all these actions above: automatic/manual 2. should I evict only node 2 from the cluster? 3. After doing that, how can I rejoin the cluster and should I take some special steps to do that. Or there is only one or few simple coommands to bring this node back to the cluster Thank you again Vidas
1 Manual, until after the last failover test. 2 Yes 3 In the Open Connection to Cluster dialog box that appears, in Action, select Add nodes to cluster. Then in Cluster or server name, either type the name of an existing cluster, select a name from the drop-down list box, or click Browse to search for an available cluster.
Manual? So if I correctly understood, after Node1 restart the cluster service on that node wouldn’t start. If I will stop the cluster service on the Node2 and after that I will evict it from the cluster, all cluster resources groups(cluster, SQL, MSDTC) will come offline. The Quorum too. As I remember in my practice such situation when the Q goes offline the cluster administrator application hangs up and I cann’t do almost anything..
So how it goes in this situation or how should it be? Thanks,

zvidas, I always set my services to manual when I’m working on a fix, so that I can order the steps I am carrying out properly. Setting it to manual doesn’t mean don’t start it, it means start it manually when you are ready.
mulhall, Yes, that’s clear for me. I just want to be sure, that after stopping the cluster service on both nodes I’ll be able to start the cluster again without any problems. If the Quorum disk status is offline, then how the cluster service will start? Both nodes will try to search it and won’t find. Am I right?