Horton

Re: A new function to wait for the backend exit after termination

Bharath Rupireddy <bharath.rupireddyforpostgres@gmail.com>

View thread View on postgr.es
From: Bharath Rupireddy <bharath.rupireddyforpostgres@gmail.com>
To: Magnus Hagander <magnus@hagander.net>
Cc: Fujii Masao <masao.fujii@oss.nttdata.com>, "Hou, Zhijie" <houzj.fnst@cn.fujitsu.com>, PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>, Muhammad Usama <m.usama@gmail.com>
Date: 2021-03-17T01:31:39Z
Lists: pgsql-hackers

Attachments

On Tue, Mar 16, 2021 at 9:48 PM Magnus Hagander <magnus@hagander.net> wrote:
> Does it really make sense that pg_wait_for_backend_termination()
> defaults to waiting *100 milliseconds*, and then logs a warning? That
> seems extremely short if I'm explicitly asking it to wait.

I increased the default wait timeout to 5seconds.

> Wait events should be in alphabetical order in pgstat_get_wait_ipc()
> as well, not just in the header (which was adjusted per Fujii's
> comment)

Done.

>
> +                       (errmsg("could not wait for the termination of
> the backend with PID %d within %lld milliseconds",
>
> That's not true though? The wait succeeded, it just timed out? Isn't
> itm ore like "backend with PID %d did not terminate within %lld
> milliseconds"?

Looks better. Done.

Attaching v10 patch for further review.

With Regards,
Bharath Rupireddy.
EnterpriseDB: http://www.enterprisedb.com

Commits

  1. Copy-edit text for the pg_terminate_backend() "timeout" parameter.

  2. Remove pg_wait_for_backend_termination().

  3. Add functions to wait for backend termination

  4. Fix typo in comment.