June 2, 2009 at 4:35 pm
We have 2 node clustered SQL 2005 server with SP2.
SQL1 runs on node A and SQL2 runs on node B.
Last week the node A server suddenly shut down and had to be rebooted.Prior to the crash there are several login failure messages for around 3-4 hrs.
SSPI handshake failed with error code 0x80090311 while establishing a connection with integrated security; the connection has been closed.
Error: 17806, Severity: 20, State: 2
The client applications on 3 different servers were all intermittely able to login and then not.They have Windows Authentication enabled and I have Mixed mode enabled on the SQLs.
My question is whether a Windows server crash simply due to login failures in SQL?
SPN is registered on the server nodes,but clients use NTLM authetication.This has been since 1 yr but the login errors have only just happened.
My guess is there could be problems with the network which suddenly arise and logins fail in between due to it.
What do you think?
June 2, 2009 at 8:29 pm
Hmm when node goes down and not brought up it takes a while to build it...but yes the users will not be able to get it if it did not roll to node b...and during the slight section if it did roll users are disconnected for that split second...
If it went down it should be able to login in with the users once it comes back up...have you done a check with all the databases - database integrity checks to ensure it all good..
June 3, 2009 at 9:25 am
I am sorry if I was not clear earlier.Logins do not fail after the node goes down,rather they begin to fail around 2-3 hrs prior to server going down.
June 3, 2009 at 9:40 am
I dont suppose you have checked the blog below out?
I too have been bitten with DNS and Clustering!
June 3, 2009 at 9:41 am
Last week the node A server suddenly shut down and had to be rebooted.
How did it "shut down"? Server was unresponsive from the network or you could not gain access to the console of the server?
Prior to the crash there are several login failure messages for around 3-4 hrs.
Were there any messages in the Windows Application or System event logs prior to the login failure messages that stand out to you?
This has been since 1 yr but the login errors have only just happened.
Has the application had any updates applied to it that may have altered the login process?
I have come across this error before with an application that used a JDS open source package to perform a SQL login (a form of JAVA). The issue I had though was the program did not account for using NTLMv2 authentication. The simple fix would be to add in a parameter for the connection string generated, however the code to generate the connection string is hardcoded in the EXE of the program.
Does the application use log files for error reporting or write to the Windows Event Log? Does this application show any logs prior to the problem starting?
Shawn Melton
Twitter: @wsmelton
Blog: wsmelton.github.com
Github: wsmelton
Viewing 5 posts - 1 through 4 (of 4 total)
You must be logged in to reply to this topic. Login to reply