Petr Jelinek

> Ran into the same issue. Is the README _trying_ to say that this is expected behaviour and has been done deliberately? > > > The state of sequences added...

It does not hang, it just sometimes takes like 5 minutes to run through.

We improved the tests stability in 2.1 (which is part of REL2_0_STABLE branch), please check if it helped.

That's expected in PG10, we use trigger for replication of truncates. Long term it's solved in PG11.

Triggers are not executed with `session_replication_role = 'replica'` that's documented in postgres manual. We have to use trigger on PG10 and older for supporting `TRUNCATE` replication, because logical decoding of...

From quick look at pg_repack, the way it does table rewrite is almost guaranteed to break logical decoding unless there is zero unconsumed changes for a given table as it...

The `--postgresql-conf` is not for publisher, it's for the new node.

@dwsteele there is check for `auditEventStack` not being `NULL` in one branch of `pgaudit_ExecutorCheckPerms_hook`, but then there is other branch where `log_select_dml` is called anyway. It feels like the `pgaudit_ExecutorCheckPerms_hook`...

I am saying that this is not Greenplum specific. Any extension that does checks permission in say `ProcessUtility_hook` (which you should be doing unless you want to have security holes...

@djw-m I wonder if this should go to the 5.5 PR , I edited it through the web interface so it ended up being own PR but...