Re: Simplify standby state machine a bit in WaitForWALToBecomeAvailable() - Mailing list pgsql-hackers

From Michael Paquier
Subject Re: Simplify standby state machine a bit in WaitForWALToBecomeAvailable()
Date
Msg-id ZAWcs2oy7MkU5bPC@paquier.xyz
Whole thread Raw
In response to Re: Simplify standby state machine a bit in WaitForWALToBecomeAvailable()  (Bharath Rupireddy <bharath.rupireddyforpostgres@gmail.com>)
Responses Re: Simplify standby state machine a bit in WaitForWALToBecomeAvailable()  (Bharath Rupireddy <bharath.rupireddyforpostgres@gmail.com>)
List pgsql-hackers
On Sat, Mar 04, 2023 at 09:47:05AM +0530, Bharath Rupireddy wrote:
> Okay, here's a patch attached.

Thanks.

+ * When source == XLOG_FROM_ANY, this function first searches for the segment
+ * with a TLI in archive first, if not found, it searches in pg_wal. This way,
+ * if there is a WAL segment with same passed-in segno but different TLI
+ * present in both the archive and pg_wal, it prefers the one with higher TLI.
+ * The reason for this is that if for example we try to do archive recovery to
+ * timeline 2, which branched off timeline 1, but the WAL for timeline 2 is not
+ * archived yet, we would replay past the timeline switch point on timeline 1
+ * using the archived WAL segment, before even looking timeline 2's WAL
+ * segments in pg_wal.

This is pretty much what the commit has mentioned.  The first half
provides enough details, IMO.
--
Michael

Attachment

pgsql-hackers by date:

Previous
From: Michael Paquier
Date:
Subject: Re: add PROCESS_MAIN to VACUUM
Next
From: Peter Smith
Date:
Subject: Re: [PATCH] Use indexes on the subscriber when REPLICA IDENTITY is full on the publisher