Re: Measuring replay lag

On 23 March 2017 at 06:42, Simon Riggs <simon(at)2ndquadrant(dot)com> wrote:
> On 23 March 2017 at 01:02, Thomas Munro <thomas(dot)munro(at)enterprisedb(dot)com> wrote:
>
>> Thanks! Please find attached v7, which includes a note we can point
>> at when someone asks why it doesn't show 00:00:00, as requested.
>
> Thanks.
>
> Now I look harder the handling for logical lag seems like it would be
> problematic in many cases. It's up to the plugin whether it sends
> anything at all, so we should make a LagTrackerWrite() call only if
> the plugin sends something. Otherwise the lag tracker will just slow
> down logical replication.
>
> What I think we should do is add an LSN onto LogicalDecodingContext to
> represent the last LSN sent by the plugin, if any.
>
> If that advances after the call to LogicalDecodingProcessRecord() then
> we know we just sent a message and we can track that with
> LagTrackerWrite().
>
> So we make it the plugin's responsibility to maintain this LSN
> correctly, if at all. (It may decide not to)
>
> In English that means the plugin will update the LSN after each
> Commit, and since we reply only on commit this will provide a series
> of measurements we can use. It will still give a saw-tooth, but its
> better than flooding the LagTracker with false measurements.
>
> I think it seems easier to add that as a minor cleanup/open item after
> this commit.

Second thoughts... I'll just make LagTrackerWrite externally
available, so a plugin can send anything it wants to the tracker.
Which means I'm explicitly removing the "logical replication support"
from this patch.

--
Simon Riggs http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services