Adrian, the query for the 0000 removal returns f.
Alvaro, Yes the pg_upgrade was done with 9.3.4. Following the bug #11264
you referenced:
select datname, datfrozenxid, datminmxid from pg_database;
datname | datfrozenxid | datminmxid
--------------+--------------+------------
acustream | 1799 | 1
> A better way not involving mxid_age() would be to use pg_controldata to
> extract the current value of the mxid counter, then subtract the current
> relminmxid from that value.
It's not clear which lines from pg_controldata to use for updating
pg_database.datminmxid.
I also assume I would do the pg_database update on a idle database.
--
View this message in context:
http://postgresql.1045698.n5.nabble.com/could-not-access-status-of-transaction-pg-multixact-issue-tp5822248p5822268.html
Sent from the PostgreSQL - sql mailing list archive at Nabble.com.