On 23 April 2017 at 17:17, Tom Lane <t...@sss.pgh.pa.us> wrote: > Simon Riggs <si...@2ndquadrant.com> writes: >>> Also, when I fix that, it gets further but still crashes at the same >>> Assert in SubTransSetParent. The proximate cause this time seems to be >>> that RecoverPreparedTransactions's calculation of overwriteOK is wrong: >>> it's computing that as "false", but in reality the subtrans link in >>> question has already been set. > >> Not sure about that, investigating. > > As a quick hack, I just hotwired RecoverPreparedTransactions to set > overwriteOK = true always, and with that and the SubTransSetParent > argument-order fix, HEAD passes the recovery tests. Maybe we can > be smarter than that, but this might be a good short-term fix to get > the buildfarm green again.
So my thoughts are now that this is indeed the long term fix. I can't think of why it would be necessary to call SubTransSetParent() twice with two different values. A subtransaction's top level xid never changes, so pg_subtrans should be either zero or the xid of the parent and never anything else. I can't see any reason now why overwriteOK should exist at all. I'm guessing that the whole "overwriteOK" idea was an incorrect response to xids appearing where they shouldn't have done because of the mistake you just corrected. So I will now remove the parameter from the call. (I'll address the discussion of the impact of that bug in separate post). -- Simon Riggs http://www.2ndQuadrant.com/ PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services
Description: Binary data
-- Sent via pgsql-hackers mailing list (email@example.com) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers