Hi All,
Had an incident on a production box last night in which TempDB filled up on a ReadOnly Secondary node within Always on Group. It became full because of the Row Versioning which happens as part of allowing a second node to grant read access.
Both Primary and secondary are not running Snapshot Isolation or RCSI, however I do understand that RCSI and SI are still used on the 2nd node when ReadOnly is allowed.
My question is what should be clearing down the RowVersioning from tempDB on the secondary node? It's as if a clean up process or something could not run which just gradually increased the RowVersions until disk ran out of space. I tried to issue a manual checkpoint but to no avail. The only way to clear down the RowVersioning from TempDB was to restart the secondary node.
Any help is appreciated!
Thanks,
Chris
Had an incident on a production box last night in which TempDB filled up on a ReadOnly Secondary node within Always on Group. It became full because of the Row Versioning which happens as part of allowing a second node to grant read access.
Both Primary and secondary are not running Snapshot Isolation or RCSI, however I do understand that RCSI and SI are still used on the 2nd node when ReadOnly is allowed.
My question is what should be clearing down the RowVersioning from tempDB on the secondary node? It's as if a clean up process or something could not run which just gradually increased the RowVersions until disk ran out of space. I tried to issue a manual checkpoint but to no avail. The only way to clear down the RowVersioning from TempDB was to restart the secondary node.
Any help is appreciated!
Thanks,
Chris