Stone may shutdown on tranlog disk full at start of checkpoint
If the tranlog disk space becomes full while a pre-checkpoint record is being written, at the point after a checkpoint is initiated, but before the checkpoint actually starts, the stone will shut down.
While this bug is fixed in v3.7 and later, a similar bug, bug 50744, with a different code path is present and may still cause stone shutdown under tranlog disk full conditions.
Workaround
Avoiding running out of tranlog space is always strongly recommended.
Last updated: 10/26/23