| From: | Jakub Wartak <jakub(dot)wartak(at)enterprisedb(dot)com> |
|---|---|
| To: | Bryan Green <dbryan(dot)green(at)gmail(dot)com>, Michael Paquier <michael(at)paquier(dot)xyz> |
| Cc: | pgsql-hackers(at)postgresql(dot)org |
| Subject: | Re: [PATCH] Add Windows support for backtrace_functions (MSVC only) |
| Date: | 2025-10-27 08:06:53 |
| Message-ID: | CAKZiRmxO1fnp=q_KsP==s0g+uzXEcLe+ZGZWTERps3pkXVA-VA@mail.gmail.com |
| Views: | Whole Thread | Raw Message | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
On Wed, Oct 22, 2025 at 3:09 AM Bryan Green <dbryan(dot)green(at)gmail(dot)com> wrote:
Hi Bryan,
[..]
> Yes. We should call cleanup at the backend shutdown, as initialize is
> called once. I have put together a new patch (for better patch naming)
> and added the cleanup code.
I've played a little time with this and this looks good to me ,
including 5-min pgbench runs with backtrace_functions set (it behaves
stable even for pgbench -C which stresses it much). Cfbot is also
green.
One thing i've I think I've noticed (but I've double-checked that's
not related to backtrace_functions set) - so it's that apparently
backends on windows leak(?) a tiny bit of memory - it's like 5
backends leak like 5 * <1kB / second @ ~3k TPS total as seen by
Resource Manager, yet i have no time to investigate that), anyway it's
does not seem to be connected to $topic.
Maybe one outstanding question is the answer to Michael's earlier question.
-J.
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Jim Jones | 2025-10-27 08:08:17 | Re: display hot standby state in psql prompt |
| Previous Message | Ajin Cherian | 2025-10-27 07:47:28 | Re: Improve pg_sync_replication_slots() to wait for primary to advance |