Re: pg_rewind succeed but postgresql showing error when trying to make standby with common ancestor

Hello.

At Thu, 15 Dec 2022 15:03:28 +0530, niraj nandane <niraj(dot)nandane(at)gmail(dot)com> wrote in
> Have 3 node setup: node1(50.2), node2(50.3) and node3(50.4). node1(50.2) is
> primary and node2(50.3) and node3(50.4) are standby. I promoted node3 first
> and then node2 to make them independent. When i try to make the node3 slave
> of node2, using pg_rewind, it says below:

pg_rewind is intended to be used for the case where new primary has
been promoted in the past of new standby (called "diverging") that has
not promoted. Thus this is not the case.

In this case, if you want to make node3 a standby of node2, just
disconnect (shutdown) node3 first, then promote node2, at last start
node3 as a standby of node2. No need for pg_rewind since node3 is
surely in the past of node2. (But a large replication delay might
badly affect.) pg_rewind doesn't work also in this case since node3
is likely be at before node2's promotion LSN.

> After this, when i start postgresql on the node3 in standby mode, i get
> below:

...(edited)
> Dec 15 07:57:33 LOG: started streaming WAL from primary at 0/2F000000 on timeline 2
> Dec 15 07:57:33 FATAL: could not receive data from WAL stream: ERROR: requested WAL segment 00000002000000000000002F has already been removed
(repeats)

I'm not sure how long the interval between the promotions of node2 and
node3, but since node2 is likely to be advanced than node3, it's not
surprising if node2 has been removed a segment that node3 is at.

> There is no WAL archiving set. This is PostgreSQL 14.4. Does this means
> PostgreSQL pg_rewind can't be used when forming cluster between two
> independent primary having common ancestor?

Yes, as explained above.

> Note: If i just promote node2 and run pg_rewind on node3 to join with
> node2, then it works correctly.

Yeah. But I think pg_rewind is not needed at all in this case.

regards.

--
Kyotaro Horiguchi
NTT Open Source Software Center