Connection to SCCM SQL Database Server lost
hello all,I hope someone has a solution for me for the following problem.I'm investigating a strange connectivity problem with my SQL server database from the SCCM SiteServer. It has been running correctly for a while (haven't seen that but I've been told ...) and stopped yesterday. I personally have no idea what might have been changed.What happens is that the SCCM Server cannot connect to the database server anymore. The DB Server is W2K8 x64 with SQL 2005 x64 instances installed. Each instance runs under it's own service account, and eachaccount registers it's SPN correctly with the instance portnumber in the AD.If I run ODBCAD32 on the SCCMserver and setup a system DSN it can't connect to the server and throws error:Connection FailedSQL State: '01000'SQL Server Error: 11001Connection Failed:SQLState: '08001'SQL Server Error: 6[Microsoft][ODBC SQL Server Driver][TCP/IP Sockets]Specified SQL server not found.Running the same ODBCAD32 config from another server in the domain I can connect without a problem to the same SQL Instance!I see that on that server I see a MSSQLSvc Kerberos Connection token when I run KLIST, but on the SCCM Server I don't get this Kerberos Ticket from the SQL Service.While the SCCM Services are running on the box, and looking under the hood with Network Monitor 3.3. it shows me that the SCCM Services constantly try to connect to the server:"SMB2: C TREE CONNECT (0x3), Path=\\SQLSERVERNAME\NI01, Mid = 66442"and the SQL server responds to that with "SMB2: R , Mid = 66442 - NT Status: System - Error, Code = (204) STATUS_BAD_NETWORK_NAME"I also see:Dfs: Get DFS Referral Request, FileName: \SQLSERVERNAME\NI01, MaxReferralLevel: 4with response:SMB2: R , Mid = 70153 - NT Status: System - Error, Code = (412) STATUS_FS_DRIVER_REQUIREDI don't really get that DFS referal to the SQL Server.... Don't know if it's relevant.Basically I have no real idea anymore where to start looking.The SCCM server is a single server running the MP, DP, SP, FBSP, PXE... rolseIf all the SCCM Services are stopped and I just run the ODBCAD32 tool to connect to the SQL Server I only see outgoing traffic from the SCCM Server to the SQL server but nothing comes back.If all services run I can't connect to the database with the SCCM Admin Console, but that seems quit logical regarding the curcomstances :)I hope anyone can give me a clue.Best regards,Eric Vegter
December 10th, 2009 4:17am

Have you verified name resolution in both directions?STATUS_BAD_NETWORK_NAME is a major indicator of that of something weird going on with name resolution.Has someone recently installed a DFS namespace that is causing a name collison perhaps?Jason | http://myitforum.com/cs2/blogs/jsandys | http://blogs.catapultsystems.com/jsandys/default.aspx | Twitter @JasonSandys
Free Windows Admin Tool Kit Click here and download it now
December 10th, 2009 5:12am

Hi Jason,thanks for your reply. There doesn't seem to be a nameresolution problem as far as I can see. The servers can ping correctly and respond with their FQDN and proper active IP. The servers are multihomed, and the 'production' adapter is the topmost in the list.I can connect with ODBCAD32 using the IP address and connecting directly to the Database port. Of cource that excludes the necessity of name-resolution al together :)The servers are not part of any DFS structure so I don't really know why this traffic appears in the snif. Maybe it is just some fallback check that happens when you connect via smb. I'll have to do a similar sniff from an operational SCCM Server in a test-env.I uninstalled a few components from the SCCM Server to make the investigation range a bit narrower (and also hoping to solve this issue by removing components :) ) but that didn't help either.Any more ideas?Rgds,Eric
December 10th, 2009 1:17pm

The problem must be located on the siteserver itself, because "from another server in the domain I can connect without a problem to the same SQL Instance"What happens if you would start a cmd in system context (on the siteserver) and fire some magic osql command, for example something like osql.exe -SYourSqlServer -dYourConfigMgrDB -q "select * from v_R_System" -EMaybe that brings up some error message that will help in further troubleshooting.
Free Windows Admin Tool Kit Click here and download it now
December 10th, 2009 1:32pm

Hi... I am getting the same problem - right down to the Network Trace error on my server. Do you remember what the resolution to this problem was?
January 18th, 2011 12:31pm

What error do you get when you execute the OSQL command that Torsten posted?http://www.enhansoft.com/
Free Windows Admin Tool Kit Click here and download it now
January 18th, 2011 1:03pm

I'm not familiar with OSQL.exe.. our SQL cluster is 2008... I found the OSQL.exe and some related rll files. I copied these to the Secondary site server (having an MP problem) and then tested. Ran cmd using PStools, psexec cmd -s (and verified I was in the system context) then ran the command C:\Temp\Binn>osql.exe -S server\instance -d sms_xxx -q "select * from v_R_System" -E and got this: [ODBC Driver Manager] Data source name not found and no default driver
January 18th, 2011 1:26pm

I am more familiar with UDL files... so, this is a the test I did: 1) create a test.udl file and plugged in my SQL server name and instance and selected by database... saved the file. 2) Ran PSEXEC -s - i cmd 3) Ran whoami to verify that I was running under system account... 4) ran c:\temp\test.udl - the UDL gui was displayed (checked Task manager to verify that this was running as system) 5) ran TEST Connection -- result = Test connect succeeded.
Free Windows Admin Tool Kit Click here and download it now
January 18th, 2011 1:49pm

This topic is archived. No further replies will be accepted.

Other recent topics Other recent topics