On 2026-Oct-09, Hayato Kuroda (Fujitsu) wrote: > Also, while thinking on it, I started to think whether we have to clean up > Tuplecids even if the transactions will be skipped. In this case transactions > are > Not replied thus the issue won't happen, right?
Yeah, I think that's correct. I was wondering myself if the 2PC case is correct -- we call ReorderBufferCleanupAbortedTupleCids() only when two_phase is false. Do we call DecodeAbort() again for those? I am not familiar with the 2PC support here. The comment for DecodeAbort() could be better for sure. -- Álvaro Herrera 48°01'N 7°57'E — https://www.EnterpriseDB.com/
