asked
Problem applying SQL2008R2SP3 on Single node cluster
We have a single-node SQL Server 2008 R2 instance running SQL Server 2008 SP1 x64. I am trying to patch this up to SQL Server 2008 R2 SP3, but the service pack is failing to install. The installation process begins, but ends very quickly when installing feature `SQL_Engine_Core_Inst_sql_engine_core_inst_Cpu64` Looking in the log files, under the MSSQLSERVER directory, it seems as though the installation is failing to stop the SQL Server instance because the SQL Server instance *has already stopped*... Which seems... wrong. I go back to the windows failover cluster manager, and find that all the clustered bits (storage, services) have gone offline. Any suggestions / advice for how to get round / through this problem gratefully received. I've done some yabingleing and failed to find a solution. [Patch log for MSSQLSERVER component][1] attached. Zipped, 'cos it's over 512KB. Sorry about that. [1]: /storage/temp/3845-mssqlserverpatchlogtxt.zip
Just wondering if it's because there's a file server / shared disk resource that's not available... Hmm. Will investigate. Hang on.
OK, so longer answer. An actual answer. That can be accepted, and everything. Go to Failover Cluster Manager, and look at the clustered service and all its dependencies. In this instance, one of those was FAILED. And it wouldn't come back online. The SQL Server service pack installer is cluster-aware, and you can watch it shut down and bring back parts of the cluster during the installation process. We could see the whole lot go down (server, disk drives, storage other resources) just fine at the start of the installation. When the File Server resources & Disk Drives were coming back online very early on in the service pack installation, however, two of them loaded up fine, and the third one didn't. And the Service Pack Installer recognised this as a failure (incorrectly? who knows - after all, something didn't come back, but seeing as it wasn't actually required for the service... but then why was it configured as a dependency?) Anyway. This resource didn't fire up, so the service pack installer interpreted this as a failure, and aborted the installation. To fix: - Temporarily create another cluster service - after making sure there are no dependencies on the broken component, move it into that new service. - Install the service pack - Move the broken component back into the original cluster service. Not the most fun way of spending time, but it seems to have worked.
17 People are following this question.
