Re: Auxiliary Processes and MyAuxProc - Mailing list pgsql-hackers

From Mike Palmiotto
Subject Re: Auxiliary Processes and MyAuxProc
Date
Msg-id CAMN686FcMWT6AK6ZfyaS3Jzx4TP5UOYnUraeatqfzUxhihxV7w@mail.gmail.com
Whole thread Raw
In response to Re: Auxiliary Processes and MyAuxProc  (Mike Palmiotto <mike.palmiotto@crunchydata.com>)
Responses Re: Auxiliary Processes and MyAuxProc
List pgsql-hackers
On Thu, Sep 26, 2019 at 6:03 PM Mike Palmiotto
<mike.palmiotto@crunchydata.com> wrote:
>
> On Thu, Sep 26, 2019 at 10:56 AM Alvaro Herrera
> <alvherre@2ndquadrant.com> wrote:
> <snip>
> >
> > Well, I think it would be easier to manage as split patches, yeah.
> > I think it'd be infrastructure that needs to be carefully reviewed,
> > while the other ones are mostly boilerplate.  If I were the committer
> > for it, I would push that initial patch first immediately followed by
> > conversion of some process that's heavily exercised in buildfarm, wait
> > until lack of trouble is evident, followed by a trickle of pushes to
> > adapt the other processes.
>
> Thanks for the feedback! I've rebased and tested on my F30 box with
> and without EXEC_BACKEND. Just working on splitting out the patches
> now and will post the new patchset as soon as that's done (hopefully
> sometime tomorrow).

Attached is the reworked and rebased patch set. I put the hook on top
and a separate commit for each process type. Note that avworker and
avlauncher were intentionally left together. Let me know if you think
those should be split out as well.

Tested again with and without EXEC_BACKEND. Note that you'll need to
set randomize_va_space to 0 to disable ASLR in order to avoid
occasional failure in the EXEC_BACKEND case.

Please let me know if anything else jumps out to you.

Thanks,
-- 
Mike Palmiotto
https://crunchydata.com

Attachment

pgsql-hackers by date:

Previous
From: Alvaro Herrera
Date:
Subject: Re: Commit fest 2019-09
Next
From: Peter Eisentraut
Date:
Subject: Revert back to standard AC_STRUCT_TIMEZONE Autoconf macro