Re: relcache reference leak with pglogical replication toinsert-only partitioned table?

Поиск
Список
Период
Сортировка
От craig.ringer
Тема Re: relcache reference leak with pglogical replication toinsert-only partitioned table?
Дата
Msg-id 43bcb555-c010-465f-bf2b-fddbb0475b0d@2ndquadrant.com
обсуждение исходный текст
Ответ на relcache reference leak with pglogical replication to insert-onlypartitioned table?  (Jeremy Finzel <finzelj@gmail.com>)
Список pgsql-hackers


On Sunday, 27 January 2019 11:20:03 UTC+8, Jeremy Finzel wrote:
I understand it's not fully supported to replicate to a differently partitioned setup on a subscriber with either pglogical or the native logical replication, however I also know that INSERT triggers can be fired in replication mode.  I have an insert-only OLTP table that I want partitioned only on the subscriber system.  I have this setup using the "old style" partitioning as it is a 9.6 system.

Provider is 9.6.6 pglogical 2.1.1
Subscriber is 9.6.10 pglogical 2.1.1

Everything appears good as far as the data.  It is partitioning correctly.  Queries on the data are planning correctly.  However, I am now getting these WARNING messages constantly.  How concerned should I be?  Is there a fix for this?  Any insight is much appreciated!

2019-01-27 03:12:34.150 GMT,,,135600,,5c4d1f44.211b0,6794,,2019-01-27 03:02:28 GMT,54/0,1057372660,WARNING,01000,"relcache reference leak: relation ""foo_pkey"" not closed",,,,,"apply COMMIT in commit before 14DB/34DB1B78, xid 1476598649 commited at 2019-01-26 21:12:34.071673-06 (action #10) from node replorigin 22",,,,"pglogical apply 16420:2094659706"

It won't corrupt anything but you can expect resource leaks.

I think there are changes to support this in pglogical3. I don't have any new information on when they might become public, but I do know work has been done to add a plugin mechanism etc as part of work toward opening pglogical3. 

В списке pgsql-hackers по дате отправления:

Предыдущее
От: Noah Misch
Дата:
Сообщение: Re: Unexpected "shared memory block is still in use"
Следующее
От: Julien Rouhaud
Дата:
Сообщение: Re: Wrong return code in vacuumdb when multiple jobs are used