From: | Ron Johnson <ronljohnsonjr(at)gmail(dot)com> |
---|---|
To: | Samed YILDIRIM <samed(at)reddoc(dot)net> |
Cc: | kaido vaikla <kaido(dot)vaikla(at)gmail(dot)com>, Pgsql-admin <pgsql-admin(at)lists(dot)postgresql(dot)org> |
Subject: | Re: Best way to stop Streaming Replication? |
Date: | 2023-12-23 17:34:22 |
Message-ID: | CANzqJaB7sWG9PKJ3-=PXMwLRJEg9SNKmLcE_VxkBe-BxF=ijuQ@mail.gmail.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-admin |
On Fri, Dec 22, 2023 at 4:38 PM Samed YILDIRIM <samed(at)reddoc(dot)net> wrote:
> Hi Ron,
>
> What is your main goal?
>
The main goal is to *stop* replication, not pause it. (I chose the word
carefully, but your questions are understandable.)
> What is the purpose of stopping replication?
> What will you do with replication when you have stopped it?
>
Are you trying to stop replaying transaction logs from the master and keep
> the replica running?
>
Want the replica still up, but does not matter if it's in replica mode.
> Will you resume log replay at some point?
>
No.
> Do you want to break replication completely while keeping standby
> receiving read-only queries?
>
How do you do that?
>
> Have you checked the pg_wal_replay_pause() function?
> Ref:
> https://www.postgresql.org/docs/15/functions-admin.html#FUNCTIONS-RECOVERY-CONTROL-TABLE
>
Yes, that's for pausing, not stopping.
> Best regards.
> Samed YILDIRIM
>
>
> On Fri, 22 Dec 2023 at 23:12, kaido vaikla <kaido(dot)vaikla(at)gmail(dot)com> wrote:
>
>> Use a some tools like patroni. I'm patroni user since 2016 and happy with
>> it.
>> br
>> Kaido
>>
>> On Fri, 22 Dec 2023, 21:49 Ron Johnson, <ronljohnsonjr(at)gmail(dot)com> wrote:
>>
>>> PG 14.10 and 15.5, if it matters.
>>>
>>> Running pg_terminate_backend() on the walsender pid (and then dropping
>>> the replication slot) seems the canonical solution, based on StackExchange,
>>> but that leaves primary_conninfo in postgresql.auto.conf and standby.signal
>>> in $PGDATA.
>>>
>>> Promoting the secondary (and then dropping the replication slot) handles
>>> all those issues, but leaves it in R/W mode, which might cause split brain
>>> issues with cluster managers.
>>>
>>> Is there a better way than the two I've described
>>>
>>
From | Date | Subject | |
---|---|---|---|
Next Message | Rajesh Kumar | 2023-12-23 18:08:34 | Switchover |
Previous Message | Scott Ribe | 2023-12-23 13:36:58 | Re: Need inputs on postgresql HA with consul cluster |