Re: race condition when writing pg_control

On 2020/05/27 16:10, Michael Paquier wrote:
> On Tue, May 26, 2020 at 07:30:54PM +0000, Bossart, Nathan wrote:
>> While an assertion in UpdateControlFile() would not have helped us
>> catch the problem I initially reported, it does seem worthwhile to add
>> it.  I have attached a patch that adds this assertion and also
>> attempts to fix XLogReportParameters().  Since there is only one place
>> where we feel it is safe to call UpdateControlFile() without a lock, I
>> just changed it to take the lock.  I don't think this adds any sort of
>> significant contention risk, and IMO it is a bit cleaner than the
>> boolean flag.
>
> Let's see what Fujii-san and Thomas think about that.  I'd rather
> avoid taking a lock here because we don't need it and because it makes
> things IMO confusing with the beginning of StartupXLOG() where a lot
> of the fields are read, even if we go without this extra assertion.

I have no strong opinion about this, but I tend to agree with Michael here.