On Tue, Mar 31, 2026 at 10:00:00AM +0300, Alexander Lakhin wrote: > So the backend is not completely stuck, but CommitTransactionCommand() > may take more than 5 seconds under some circumstances (maybe it's worth > investigating which exactly).
One could blame slow hardware, difficult to say, and I'm puzzled by these periodic bumps that don't seem to happen elsewhere. There is at least one alternative that I can think here to make the test more stable and make sure that worker_spi reaches its main loop before attempting the database commands. We could add a wait_for_event() at the end of launch_bgworker() based on WorkerSpiMain, and enlarge worker_spi_naptime to an insanely larger value, to make sure that we remain on the WaitLatch() until the worker is interrupted and that we don't attempt a new transaction. -- Michael
signature.asc
Description: PGP signature
