Thread: Monitoring Streaming Replication in 9.2
Newb question here. I have streaming replication working with 9.2 and I'm using Bucardo's check_postgres.pl to monitor replication. I see that it runs this query on the slave:
SELECT pg_last_xlog_receive_location() AS receive, pg_last_xlog_replay_location() AS replay
That returns hex, which is then converted to a number in the script. SELECT pg_last_xlog_receive_location() AS receive, pg_last_xlog_replay_location() AS replay
How do I determine a meaningful alert threshold for that value? Is there a reliable way to monitor replication lag in seconds? How do other people handle this?
If you want to find the lag in seconds, then you need to execute something like below.
SELECT
SELECT
pg_last_xact_replay_timestamp() - now();
On Fri, May 16, 2014 at 9:37 PM, J Adams <pacetownsley@gmail.com> wrote:
My question is this: what does that number represent? Is it just the log position? If so, how does the log position translate to queries? Does one log position = one query? (I did say this was a newb question.)Newb question here. I have streaming replication working with 9.2 and I'm using Bucardo's check_postgres.pl to monitor replication. I see that it runs this query on the slave:That returns hex, which is then converted to a number in the script.
SELECT pg_last_xlog_receive_location() AS receive, pg_last_xlog_replay_location() AS replay
How do I determine a meaningful alert threshold for that value? Is there a reliable way to monitor replication lag in seconds? How do other people handle this?