On Mon, Nov 12, 2018 at 11:55 PM Jeremy Finzel <finzelj@gmail.com> wrote: > > I believe I found a race condition in WaitForBackgroundWorkerStartup in the case where it encounters an ERROR during startup. I found that depending on the speed of the system, it will unreliably return either status BGWH_STOPPED or BGWH_STARTED. But I can reliably reproduce getting BGWH_STOPPED by tweaking the worker_spi.c test module. >
Yeah, I think it is possible that you get different values in such cases because we consider worker status as started after we have launched the worker. Now, if we get the error in the worker initialization, then the user can get any of those values. I think this is what is happening in your case where you are saying "ERROR during startup". Am I missing something?
Perhaps. What I am saying is that some machines show ERROR during startup, and some machines don't get an error at all, return successfully, then immediately error and die in the background, but the client is not shown this. The behavior isn't predictable. However, I can get a predictable ERROR to happen always if I put a short pause before WaitForBackgroundWorkerStartup.
I'm unclear on what counts as "worker initialization". The error is happening in the worker_spi_main function, not in the worker_spi_launch function. So does an immediate error in worker_spi_main count as part of the worker initialization?