| From: | Tomas Vondra <tomas(at)vondra(dot)me> |
|---|---|
| To: | Fujii Masao <masao(dot)fujii(at)gmail(dot)com> |
| Cc: | surya poondla <suryapoondla4(at)gmail(dot)com>, Chao Li <li(dot)evan(dot)chao(at)gmail(dot)com>, PostgreSQL Hackers <pgsql-hackers(at)lists(dot)postgresql(dot)org>, Thomas Munro <thomas(dot)munro(at)gmail(dot)com>, shihao zhong <zhong950419(at)gmail(dot)com> |
| Subject: | Re: A bug in recovery prefetcher, triggering asserts / FATAL: checkpoint request failed |
| Date: | 2026-11-19 23:22:19 |
| Message-ID: | a406cac9-11dd-4c77-8e2c-7ad9d7f7f4e0@vondra.me |
| Views: | Whole Thread | Raw Message | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
On 11/20/26 00:07, Fujii Masao wrote:
> On Fri, Nov 20, 2026 at 3:29 AM Tomas Vondra <tomas(at)vondra(dot)me> wrote:
>> I ended up committing it too.
>
> Thanks for working on this!
>
> The new test fails on some buildfarm members, such as indri on v17 [1].
> It sets maintenance_io_concurrency = 1, but v17 and earlier reject
> nonzero values on platforms without posix_fadvise(), causing the standby
> to fail to start.
>
> In v18, this restriction was removed for AIO support, so the setting is
> accepted, although recovery prefetching remains disabled on these platforms.
>
> Should we skip the test when recovery prefetching is unavailable?
>
D'oh! I completely forgot about this. Yes, we should skip the test for
these machines. I'll look into that tomorrow morning (unless you want to
take care of it).
What's a good way to check if a machine supports posix_fadvise? I don't
see any TAP test already doing that. I guess we could try assigning into
effective_io_concurrency, and skip the test if it fails.
regards
--
Tomas Vondra
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Masahiko Sawada | 2026-11-19 23:42:06 | Re: Add a hook for handling logical decoding messages on subscribers. |
| Previous Message | Sami Imseih | 2026-11-19 23:11:21 | Re: pg_restore fails on extracted tar archives with large objects |