[Postgres-xc-developers] Pacemaker RA for datanode/coordinator

[Koichi S. <koi...@gm...>]

Hi,

I had a chance to meet people from Linux-HA Japan an discussed how to
provide Pacemaker RA (resource agent) for coordinator/datanode.   I
found present PostgreSQL RA (for 9.1) uses the following queries to
the hot standby node.

1. select pg_is_in_recovery();

works fine.

2. select application_name,upper(state),upper(sync_state) from
pg_stat_replication;

This does not work.  pg_stat_replication has zero row.

3. select pg_last_xlog_replay_location(),pg_last_xlog_receive_location();

works fine.

4. select now();

works fine.

Because this is datanode slave, we don't have GTM connection or
snapshot.   This might be a cause of the problem 2, or we might have
blocked to update pg_stat_replication system catalog.

Any suggestions?
----------
Koichi Suzuki