[Bucardo-general] Increasingly delayed target update

Enrique eps.lab at protonmail.com
Tue Jan 15 15:58:05 UTC 2019

Hello, list.

I'm beginning to use Bucardo and I made a quick and dirty test:

* Databases test1 (locaslhost) and test2 (different host) created with pgbench.
* Sync "sync01" Relgroup "alpha"  DB group "sync01" test1:source test2:target
* pgbench -c 5 -j 5 -t $VERYVERYLARGENUMBER test1

I noticed a very fast update rate in test2, but then the more tests I ran, the
slower it was to show updates in test2 (most time spent 'counting all deltas').
I recorded the "time to run" from bucardo status sync01 each 10 seconds, and
here is a graph of what I obtained:


That's for the first hour or so, then it kept climbing until there was no space
left on the drive.

I have these questions:
* Is there anything I should specifically take into account beforehand so a kind
of purge is done? 'bucardo purge all' used to return 0 rows deleted. The count
in the delta tables was very large, as expected.
* How can I prevent the delta tables to grow indefinitely?
* Is there anything I'm missing completely? :D
* I didn't find much updated information in the website docs. How can I help?

Thank you so much beforehand :)

Best regards.


More information about the Bucardo-general mailing list