On Sun, Aug 9, 2026 at 11:38 PM chee.wooson <[email protected]> wrote: > Attached is v5. It fixes the test only; the server-side code is unchanged > from v4. The point is now PID-filtered to the VACUUM backend (exactly one > waiter) and is detached before waking.
I pushed your fix just now. I did not include your test case. It was useful for proving that ProcArrayInstallImportedXmin was buggy, but didn't seem worth keeping forever. I was worried about test stability issues. Thanks! -- Peter Geoghegan
