Re: PATCH: Configurable file mode mask

On 3/20/18 11:14 PM, Michael Paquier wrote:
> On Tue, Mar 20, 2018 at 05:44:22PM -0400, Stephen Frost wrote:
>> * David Steele (david@pgmasters.net) wrote:
>>> On 3/16/18 11:12 AM, Stephen Frost wrote:
>>> It seems to me that pg_basebackup and pg_receivexlog should have a -g
>>> option to control the mode of the files that they write to disk (not
>>> including the modes stored in the tar files).
>>>
>>> Or perhaps we should just update the perms in the tar files for now and
>>> leave the rest alone.
>>
>> Having options to pg_basebackup to control what's done makes sense to
>> me- but whatever those options do, I'd expect them to apply equally to
>> the tar files and to the files extracted with plain mode.  Having those
>> be different really strikes me as very odd.
>
> Agreed for the consistency part, permissions should be applied
> consistently for the folder and the tar format.
>
> Having the option for pg_receivewal definitely makes sense to me, as it
> is the one in charge of opening and writing the WAL segments.  For
> pg_basebackup, let's not forget that there is one tar file for each
> tablespace, and that each file is received separately using a COPY
> stream.  There is some logic already which parses the tar header part of
> an individual file in order to look for recovery.conf (see
> ReceiveTarFile() in pg_basebackup.c).  It would be possible to enforce
> grouping permissions when receiving each file, and this would be rather
> low-cost in performance I think.  Honestly, my vote would go for having
> the permissions set correctly by the source server as this brings
> consistency to the whole experience without complicating the interface
> of pg_basebackup, and this also makes the footprint of this patch on
> pg_basebackup way lighter.

These updates address Michael's latest review and implement group access
for pg_basebackup, pg_receivewal, and pg_recvlogical.  A new internal
GUC, data_directory_group_access, allows remote processes to determine
the correct mode using the existing SHOW protocol command.

I have dropped patch 01, which added the pg_resetwal tests.  The tests
Peter added recently are sufficient for this patch so I'll pursue adding
the other tests separately to avoid noise on this thread.

Thanks,
--
-David
david@pgmasters.net