| From: | Anthonin Bonnefoy <anthonin(dot)bonnefoy(at)datadoghq(dot)com> |
|---|---|
| To: | Mircea Cadariu <cadariu(dot)mircea(at)gmail(dot)com> |
| Cc: | PostgreSQL Hackers <pgsql-hackers(at)postgresql(dot)org>, Japin Li <japinli(at)hotmail(dot)com> |
| Subject: | Re: Propagate XLogFindNextRecord error to callers |
| Date: | 2026-02-11 08:43:51 |
| Message-ID: | CAO6_XqrUzBT70kE20uHH=FWyfGJaTDMwqv++yNppdusywsG8-w@mail.gmail.com |
| Views: | Whole Thread | Raw Message | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
On Mon, Feb 9, 2026 at 1:15 PM Mircea Cadariu <cadariu(dot)mircea(at)gmail(dot)com> wrote:
> In XLogNextRecord, right before the above code, we do *errormsg = NULL. Should this be done also in XLogFindNextRecord in the patch?
>
> If so, what about even extracting a helper method which will be called from both places?
XLogReadRecord may already have consumed errormsg_deferred and set
errormsg. We can't set it to NULL, or that would erase a valid error
message.
A simplification would have been to remove processing the deferred
error processing from XLogNextRecord and leave it to
XLogFindNextRecord, but there are some calls to XLogNextRecord (like
in xlogprefetcher.c), so being able to get the error message when
calling XLogNextRecord is necessary.
> A nit for the commit message: Propage -> Propagate
Fixed
| Attachment | Content-Type | Size |
|---|---|---|
| v4-0001-Propagate-errormsg-to-XLogFindNextRecord-caller.patch | application/octet-stream | 8.2 KB |
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Zhijie Hou (Fujitsu) | 2026-02-11 08:44:33 | RE: Improve pg_sync_replication_slots() to wait for primary to advance |
| Previous Message | Etsuro Fujita | 2026-02-11 08:14:36 | Re: Instability in postgres_fdw regression tests |