From: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
---|---|
To: | Amit Kapila <amit(dot)kapila16(at)gmail(dot)com> |
Cc: | Robert Haas <robertmhaas(at)gmail(dot)com>, Pavan Deolasee <pavan(dot)deolasee(at)gmail(dot)com>, pgsql-hackers <pgsql-hackers(at)postgresql(dot)org> |
Subject: | Re: Parallel worker error |
Date: | 2017-08-30 12:04:27 |
Message-ID: | 24985.1504094667@sss.pgh.pa.us |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
Amit Kapila <amit(dot)kapila16(at)gmail(dot)com> writes:
> I am able to reproduce this without involving session authorization
> guc as well. One needs to drop the newly created role from another
> session, then also we can see the same error.
Hm. I suspect the basic shape of what's happening here is "an existing
session can continue to run with OuterUserId corresponding to a dropped
role, but we fail when trying to duplicate that state into a parallel
worker". I wonder whether there aren't similar gotchas for other GUCs
whose interpretation depends on catalog lookups, eg search_path.
We might need to redesign the GUC-propagation mechanism so it sends
the various internal representations of GUC values, not the user-visible
strings. (I'm thinking of the blobs that guc.c can use to restore a
previous state at transaction abort ... don't recall what the code
calls them ATM.)
regards, tom lane
From | Date | Subject | |
---|---|---|---|
Next Message | Antonin Houska | 2017-08-30 12:08:34 | Re: WIP: Separate log file for extension |
Previous Message | Amit Khandekar | 2017-08-30 12:02:52 | Re: Parallel Append implementation |