SQL Server Performance Forum – Threads Archive

Unable to connect to database after SQL fail over

Hi,
Has anyone had any experience/issues with a Load Balanced SharePoint Portal Server Farm and a SQL Cluster where the SQL Cluster is failed over and the SharePoint intranet can not contact the database?? Error:
Unable to connect to database. Check database connection information and make sure the database server is running. For tips on troubleshooting this error, search for article 823287 in the Microsoft Knowledge Base athttp://support.microsoft.com. Details are;
Cluster
2003 R2 X64 Cluster
SQL 2005 x64
MSA100 with HA
Physical Cluster servers
hncls1
hncls2
SQL is installed as Active/Active cluster
SQL installed to hnsqlv1 (virtual entity on hncls1)
SQL Instance SQLPhatcsSvr
SQL Data on shared disk V:
SQL Logs on shared disk W: SQL installed to hnsqlv2 (virtual entity on hncls2)
SQL Instance SQLIntranetSvr
SQL Data on shared disk X:
SQL Logs on shared disk Y: SharePoint
2003 R2 32bit
SPS 2003
2 front end web server using NLB, hnsps1 & 2
1 index server
accessed byhttp://mycompany 192.168.1.x (allocated to both FEWS in NLB)
SharePoint Portal Server is installed as front end servers to hnsps1 and 2 and set to the following
Configuration database server: HNSQLV2SQLIntranetSvr
Content database server: *HNSQLV2SQLINTRANETSVR
Component settings database server: *HNSQLV2SQLINTRANETSVR
Single sign-on credentials: HNSQLV2SQLIntranetSvr I have checked the KB Article mentioned above and all seems ok. Upon SQL Fail over (server restarted) if the Intranet is refreshed or a new link is clicked the above error is received. I have done some ODBC test to (below) while SQL is in a failed over state
Server: HNSQLV2SQLINTRANETSVR
Database: Workwise1_PROF
Test is successful which would indicate that SQL has failed over successfully.
I am finding the following errors in on the cluster servers
Event Type: Warning
Event Source: SQLBrowser
Event Category: None
Event ID: 3
Date: 29/03/2006
Time: 11:37:46 a.m.
User: N/A
Computer: HNCLS2
Description:
The configuration of the AdminConnectionTCP protocol in the SQL instance SQLINTRANETSVR is not valid. For more information, see Help and Support Center athttp://go.microsoft.com/fwlink/events.asp.
Event Type: Error
Event Source: SQLBrowser
Event Category: None
Event ID: 8
Date: 29/03/2006
Time: 11:37:46 a.m.
User: N/A
Computer: HNCLS2
Description:
The SQLBrowser service was unable to process a client request. For more information, see Help and Support Center athttp://go.microsoft.com/fwlink/events.asp.
Event Type: Failure Audit
Event Source: MSSQL$SQLINTRANETSVR
Event Category: (4)
Event ID: 18456
Date: 29/03/2006
Time: 11:37:48 a.m.
User: mycompanyzSPSserv
Computer: HNSQLV2
Description: Login failed for user ‘mycompanyzSPSserv’. [CLIENT: 192.168.1.19] For more information, see Help and Support Center athttp://go.microsoft.com/fwlink/events.asp. Data:
0000: 18 48 00 00 0e 00 00 00 .H……
0008: 17 00 00 00 48 00 4e 00 ….H.N.
0010: 53 00 51 00 4c 00 56 00 S.Q.L.V.
0018: 32 00 5c 00 53 00 51 00 2..S.Q.
0020: 4c 00 49 00 4e 00 54 00 L.I.N.T.
0028: 52 00 41 00 4e 00 45 00 R.A.N.E.
0030: 54 00 53 00 56 00 52 00 T.S.V.R.
0038: 00 00 07 00 00 00 6d 00 ……m.
0040: 61 00 73 00 74 00 65 00 a.s.t.e.
0048: 72 00 00 00 r…
Regards
Mike

Have you configured the 2 instances to use the same port?
Better check it before performing the failover, refer to SQL error logs. Satya SKJ
Contributing Editor & Forums Moderator
http://www.SQL-Server-Performance.Com
This posting is provided â€œAS ISâ€ with no rights for the sake of knowledge sharing.

No, Microsoft said we didn’t need to because we are using 2 different named instances
Looking at the Logs now. From the SQL Browser errors that were posted above, it looks like a clash of the 2 named instances in the SQL Port?? Maybe??

I have found that hnsqlv1 uses port 1392 and hnsqlv2 uses 1695 when i fail over hncls1 (physical server) which has the virtaul server hnsqlv1 to hncls2 it tries to use the same ports but also starts other port under hnssqlv1 TCP hncls2:1392 hncls2.mycompany.co.nz:0 LISTENING
TCP hncls2:1392 hnsqlv1.mycompany.co.nz:1856 ESTABLISHED
TCP hncls2:1392 hnsqlv1.mycompany.co.nz:1868 ESTABLISHED
TCP hncls2:1856 hnsqlv1.mycompany.co.nz:1392 ESTABLISHED
TCP hncls2:1862 hnsqlv1.mycompany.co.nz:1392 TIME_WAIT
TCP hncls2:1865 hnsqlv1.mycompany.co.nz:1392 TIME_WAIT
TCP hncls2:1867 hnsqlv1.mycompany.co.nz:1392 TIME_WAIT
TCP hncls2:1868 hnsqlv1.mycompany.co.nz:1392 ESTABLISHED
I have also found another post under SQL7.0 and 2000 clustering saying that because the Active/Active instances are installed to differnet named instances with different ip address you can set bot instance to use 1433
http://www.sql-server-performance.com/forum/topic.asp?TOPIC_ID=2922&SearchTerms=active/active,ports Doses this sound right?? SO…….If i set both instances to use 1433 upon fail over there should be no isses with reconenct to the SQL Browser[?]

I believe it is better to check from Sharepoint also, as it is involved here … which is new to me. Is that application is a cluster aware? Satya SKJ
Contributing Editor & Forums Moderator
http://www.SQL-Server-Performance.Com
This posting is provided â€œAS ISâ€ with no rights for the sake of knowledge sharing.

]]>