Thread

Commits

Same data as JSON: GET /api/v1/messages/:b64id/commits the thread's linked commits as JSON, with link sources. API reference →

Print new OldestXID value in pg_resetwal when it's being changed
- 11cc0f452253 14.21 landed
- 7c494072bcb2 15.16 landed
- 890cc81b6ee2 16.12 landed
- f2e0ca0af97b 17.8 landed
- 19594271c1a2 18.2 landed
- d5b4f3a6d4e0 19 (unreleased) landed
pg_resetxlog: add option to set oldest xid & use by pg_upgrade
- 74cf7d46a91d 15.0 cited

pg_resetwal prints new OldestXID in wrong circumstances

Heikki Linnakangas <hlinnaka@iki.fi> — 2025-11-19T09:52:31Z

When the --oldest-transaction-id option is used, pg_resetwal doesn't 
print the new value like it does for other similar options. For example, 
if you use --next-oid and --oldest-transaction-id, only the new NextOID 
value is printed:

> $ pg_resetwal --dry-run -D data --next-oid 10  --oldest-transaction-id 100
> Current pg_control values:
> 
> ...
> 
> Values to be changed:
> 
> First log segment after reset:        000000010000000000000005
> NextOID:                              10

Printing the new OldestXID value is incorrectly tied to whether the 
--next-transaction-id option is given, so this prints it, even though 
OldestXID is not being modified:

> $ pg_resetwal --dry-run -D data --next-oid 10  --next-transaction-id 100
> Current pg_control values:
> 
> ...
> 
> 
> Values to be changed:
> 
> First log segment after reset:        000000010000000000000005
> NextOID:                              10
> NextXID:                              100
> OldestXID:                            744
> OldestXID's DB:                       1

This seems to have been an oversight when the --oldest-transaction-id 
option was added (commit 74cf7d46a91d). Before that, OldestXID was reset 
when the --next-transaction-id option was given.

Fix attached. Barring objections, I will commit and backpatch this.

- Heikki

Re: pg_resetwal prints new OldestXID in wrong circumstances

Bertrand Drouvot <bertranddrouvot.pg@gmail.com> — 2025-11-19T10:05:12Z

Hi,

On Wed, Nov 19, 2025 at 11:52:31AM +0200, Heikki Linnakangas wrote:
> Printing the new OldestXID value is incorrectly tied to whether the
> --next-transaction-id option is given, so this prints it, even though
> OldestXID is not being modified:

Nice catch!

> This seems to have been an oversight when the --oldest-transaction-id option
> was added (commit 74cf7d46a91d).

yeah.

> Before that, OldestXID was reset when the
> --next-transaction-id option was given.
> 
> Fix attached. Barring objections, I will commit and backpatch this.

LGTM, thanks!

Regards,

-- 
Bertrand Drouvot
PostgreSQL Contributors Team
RDS Open Source Databases
Amazon Web Services: https://aws.amazon.com

Re: pg_resetwal prints new OldestXID in wrong circumstances

Heikki Linnakangas <hlinnaka@iki.fi> — 2025-11-19T16:08:22Z

On 19/11/2025 12:05, Bertrand Drouvot wrote:
> On Wed, Nov 19, 2025 at 11:52:31AM +0200, Heikki Linnakangas wrote:
>> Printing the new OldestXID value is incorrectly tied to whether the
>> --next-transaction-id option is given, so this prints it, even though
>> OldestXID is not being modified:
> 
> Nice catch!
> 
>> This seems to have been an oversight when the --oldest-transaction-id option
>> was added (commit 74cf7d46a91d).
> 
> yeah.
> 
>> Before that, OldestXID was reset when the
>> --next-transaction-id option was given.
>>
>> Fix attached. Barring objections, I will commit and backpatch this.
> 
> LGTM, thanks!

Committed, thanks!

- Heikki