Re: Multiple table synchronizations are processed serially

From: Petr Jelinek <petr(dot)jelinek(at)2ndquadrant(dot)com>
To: Peter Eisentraut <peter(dot)eisentraut(at)2ndquadrant(dot)com>, Masahiko Sawada <sawada(dot)mshk(at)gmail(dot)com>, PostgreSQL-development <pgsql-hackers(at)postgresql(dot)org>, Andres Freund <andres(at)anarazel(dot)de>
Subject: Re: Multiple table synchronizations are processed serially
Date: 2017-05-19 22:10:21
Message-ID: 7ba90ed5-b5e6-c778-2f47-991c3eb75529@2ndquadrant.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

On 19/05/17 21:47, Peter Eisentraut wrote:
> On 5/19/17 01:01, Masahiko Sawada wrote:
>> Seems all four table sync workers are launched at the almost same
>> time, but three workers of them get stuck in idle transaction state
>> when creating replication slot. That is these waiting workers cannot
>> proceed its work until first connected table sync worker finishes. ps
>> command shows the followings.
>
> Creating a replication slot waits for all transactions to finish. So if
> one of those transactions is a table copy of another subscription, it
> has to wait for that.
>

Well IMHO this concrete example is partially effect of the snapshort
builder fixes we did with Andres. Before those fixes the ondisk snapshot
may have been used in this situation. Makes me think if we want to mark
the exported snapshot as containing all transactions and using it if
that's the case.

--
Petr Jelinek http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Training & Services

In response to

Browse pgsql-hackers by date

  From Date Subject
Next Message Thomas Munro 2017-05-19 22:43:08 Re: transition table behavior with inheritance appears broken (was: Declarative partitioning - another take)
Previous Message Tom Lane 2017-05-19 21:31:50 Re: Removal of plaintext password type references