| From: | Andres Freund <andres(at)2ndquadrant(dot)com> |
|---|---|
| To: | Michael Paquier <michael(dot)paquier(at)gmail(dot)com> |
| Cc: | PostgreSQL mailing lists <pgsql-hackers(at)postgresql(dot)org>, Heikki Linnakangas <hlinnakangas(at)vmware(dot)com> |
| Subject: | Re: Assertion failure when streaming logical changes |
| Date: | 2015-02-10 13:43:22 |
| Message-ID: | 20150210134322.GF21017@alap3.anarazel.de |
| Views: | Raw Message | Whole Thread | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
On 2015-02-10 22:06:34 +0900, Michael Paquier wrote:
> On Tue, Feb 10, 2015 at 9:46 PM, Andres Freund <andres(at)2ndquadrant(dot)com>
> wrote:
>
> > Yea, it really looks like the above commit is to "blame". The new xmin
> > tracking infrastructure doesn't know about the historical snapshot...
> >
>
> I think that we need a better regression coverage here... For example, we
> could add some tap tests in test_decoding to test streaming of logical
> changes. This would help in the future to detect such problems via the
> buildfarm.
I agree. It's more or less a accident that the assert - which just
should be moved in the regd_count == 0 branch - didn't trigger for the
SQL interface. The snapshot acquired by the SELECT statement prevents it
there.
It's not entirely trivial to add tests for receivelogical though. You
need to stop it programatically after a while.
Greetings,
Andres Freund
--
Andres Freund http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Training & Services
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Pavel Stehule | 2015-02-10 13:46:23 | Re: For cursors, there is FETCH and MOVE, why no TELL? |
| Previous Message | Michael Paquier | 2015-02-10 13:32:57 | Re: Patch: add recovery_timeout option to control timeout of restore_command nonzero status code |