On Mon, Jan 11, 2021 at 11:22:05AM +0500, Andrey Borodin wrote:
> I'm marking patch as ready for committer.

Thanks.

> I can't tell should we backpatch insurance patch or not: it potentially fixes 
> unknown bugs, and potentially contains unknown bugs. I can't reason because 
> of such uncertainty. I've tried to look for any potential problem and as for 
> now I see none. Chances are <slru-truncate-t-insurance-v5.patch> is doing 
> code less error-prone.

What do you think of abandoning slru-truncate-t-insurance entirely?  As of
https://postgr.es/m/20200330052809.gb2324...@rfd.leadboat.com I liked the idea
behind it, despite its complicating the system for hackers and DBAs.  The
TruncateMultiXact() interaction rendered it less appealing.  In v14+, commit
cd5e822 mitigates the kind of bugs that slru-truncate-t-insurance mitigates,
further reducing the latter's value.  slru-truncate-t-insurance does mitigate
larger trespasses into unlink-eligible space, though.

> Fix <slru-truncate-modulo-v6.patch> certainly worth backpatching.

I'll push it on Saturday, probably.


Reply via email to