| From: | Nathan Bossart <nathandbossart(at)gmail(dot)com> |
|---|---|
| To: | Jeff Davis <pgsql(at)j-davis(dot)com> |
| Cc: | Michael Paquier <michael(at)paquier(dot)xyz>, Ted Yu <yuzhihong(at)gmail(dot)com>, Pavel Luzanov <p(dot)luzanov(at)postgrespro(dot)ru>, Justin Pryzby <pryzby(at)telsasoft(dot)com>, pgsql-hackers(at)postgresql(dot)org |
| Subject: | Re: allow granting CLUSTER, REFRESH MATERIALIZED VIEW, and REINDEX |
| Date: | 2023-06-20 17:49:27 |
| Message-ID: | 20230620174927.GA494037@nathanxps13 |
| Views: | Raw Message | Whole Thread | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
On Tue, Jun 20, 2023 at 10:40:32AM -0700, Nathan Bossart wrote:
> On Tue, Jun 20, 2023 at 10:04:37AM -0700, Jeff Davis wrote:
>> I think v4-0001 broke the handling of toast tables? It looks like you
>> removed the check for !skip_privs but need to add it to the flags in
>> vacuum_is_permitted_for_relation().
>
> Good catch. I'm not sure why some of the calls to
> vacuum_is_permitted_for_relation() are masking the options. AFAICT we can
> simply remove the masks. I've done so in the attached patch.
Oh, I think I see why. This appears to be used to control which WARNING
message is emitted. If you lose permissions before you get to analyzing in
a VACUUM (ANALYZE) command, you'll get a "permission denied to vacuum"
message instead of a "permission denied to analyze" message. IMO a better
way to do that would be to control only those two bits (VACOPT_VACUUM and
VACOPT_ANALYZE) in calls to vacuum_is_permitted_for_relation(), and to
leave the rest untouched.
Patch incoming...
--
Nathan Bossart
Amazon Web Services: https://aws.amazon.com
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Nathan Bossart | 2023-06-20 17:56:34 | Re: allow granting CLUSTER, REFRESH MATERIALIZED VIEW, and REINDEX |
| Previous Message | Andres Freund | 2023-06-20 17:46:10 | Re: remap the .text segment into huge pages at run time |