[pgpool-hackers: 3816] Re: Proposal: language cleanup in Pgpool-II

Wed Sep 16 17:28:02 JST 2020

Hi Umar,

>> Replacing to red and green is a little bit confusing to me. What
>> about denylist/allowlist? It's suggested by Linux.
>>
>> Yes, I read few Linux related articles and few other sources
> too. denylist/allowlist is used for decision where we either allow or deny.
> In our case, based on mutability factor we decide whether we should send
> query to primary or standby. So in-fact we are not blocking queries.

That makes sense.

> IMO we can use
> 1. deny/allow ( e.g allow_function_list, deny_function_list,
> allow_query_pattern, allow_memqcache_table_list, deny_memqcache_table_list)
> ?
> 2. or descriptive term like mutable/immutable & safe/unsafe based on pgpool
> usage ( e.g. mutable_function_list, immutable_function_list,
> safe_query_pattern, safe_memqcache_table_list, unsafe_memqcache_table_list
> ) ?

I like 2. What about:

black/white_function_list -> write_function_list, read_only_function_list
black_query_pattern -> primay_routing_query_pattern
black/white_memqcache_table_list -> cache_unsafe/cache_safe_table_list

>> v1-002-Watchdog-master-leader
>> >    Replace master to 'leader' for 'master' watchdog nodes
>>
>> Sounds reasonable choice to me.
>>
>> > v1-003-ALWAYS_MASTER-to-ALWAYS_PRIMARY
>> >   Replace backend_flagx option 'ALWAYS_MASTER' with 'ALWAYS_PRIMARY'
>>
>> Sounds good to me.
>>
>> > v1-004-relcache_query_target--master-to-primary
>> >    Replace relcache_query_target option 'master' to 'primary'
>>
>> Sounds good to me.
>>
>> > v1-005-master-slave-to-primarystandby
>> >    Replace Master/Slave with Primary/Secondary
>>
>> I think we could replace "Master/Slave mode" to "Native replication
>> mode".
>>
>> Using "Primary" as the replacemnet for "Master" sounds is confusing as
>> there's "Primary" in the streaming replication mode.
>>
>> I will use Native Replication  mode in next patch.
> 
>> > v1-006-masterslave-to-mainreplica
>> >    Replace Master Node with Main node ( node id with 0 or youngest live )
>> >    Couldn't translate with primary/leader as they were already used in
>> > different context
>>
>> Sounds good to me.
>>
>> > v1-007-follow_master-and-recovery-and-misc
>> >    Replace follow_master with follow_primary with parameters replaced
>> with
>> > new/old master with new/old main.
>> >    Replace some remaining occurrences of master with primary/main/leader
>>
>> Sounds good to me.
>>
>> > In the above patches I didn't change older releases notes ( except for
>> some
>> > links so they don't break and point to new terminology), history, TODO
>> and
>> > Japanese Doc string.
>>
>> Sounds good to me. (I will take care of Japanese Doc string).
>>
>> > Postmaster ( It's used around 150 times ) and very few other occurrences
>> of
>> > 'master' are still present.
>>
>> No objection from me.
>>
>> > There are alot of inter-dependent terminologies and so I might have
>> missed
>> > something. Looking at them as a whole makes some sense, changing one term
>> > might lead to inconsistency in some other areas.
>> > I tried to come up with the above terminologies, just to start
>> > conversation. Feel free to share feedback and comments.
>>
>> Thanks.
>>
>> > Regards
>> > Umar Hayat
>> >
>> >
>> > On Fri, Sep 4, 2020 at 8:03 AM Umar Hayat <m.umarkiani at gmail.com> wrote:
>> >
>> >> Hi Tatsuou Ishii,
>> >> Yes I'm planning for 4.2 and work is in progress, I will send patches in
>> >> the upcoming week.
>> >>
>> >> Regards
>> >> Umar Hayat
>> >>
>> >> On Thu, Sep 3, 2020 at 5:59 AM Tatsuo Ishii <ishii at sraoss.co.jp> wrote:
>> >>
>> >>> Hi Umar,
>> >>>
>> >>> Now that we are getting closer to release Pgpool-II 4.2, I would like
>> >>> to know if you wish to bring your work into 4.2 or not. If you wish,
>> >>> can you please tell us the time line for the patch?
>> >>>
>> >>> > On Fri, Jul 24, 2020 at 4:33 PM Tatsuo Ishii <ishii at sraoss.co.jp>
>> >>> wrote:
>> >>> >
>> >>> >> > Hi Hackers,
>> >>> >> > Recently PG community did a language clean up ( to make it more
>> >>> >> inclusive )
>> >>> >> > for Server, including docs and code (see here
>> >>> >> > <
>> >>> >>
>> >>>
>> https://www.postgresql.org/message-id/20200615182235.x7lch5n6kcjq4aue@alap3.anarazel.de
>> >>> >> >).
>> >>> >> > Most changes were related to renaming "slave" and "master"
>> >>> terminologies.
>> >>> >> > Do we have any plans to make such changes to make pgpool
>> consistent
>> >>> with
>> >>> >> > Server as well as in general Pgpool specific things ?
>> >>> >> > If we do have a plan, it would need multiple patches as it would
>> >>> affect
>> >>> >> > docs, code and configuration and samples.
>> >>> >> > Let me know suggestions and thoughts and I can work on this.
>> >>> >>
>> >>> >> It seems the discussion is still on going? I mean it has not been
>> >>> >> committed yet?
>> >>> >>
>> >>> >>
>> >>> >>
>> >>>
>> https://www.postgresql.org/message-id/flat/20200615182235.x7lch5n6kcjq4aue%40alap3.anarazel.de
>> >>> >>
>> >>> >> Discussion is going on for two pending point which are not part of
>> >>> patches
>> >>> > ( postmaster & master branch) and one WIP patch(multi-master), rest
>> of
>> >>> the
>> >>> > 7 patches for that thread are committed.
>> >>> >
>> >>>
>> https://git.postgresql.org/gitweb/?p=postgresql.git;a=commit;h=229f8c219f8fffacc253eca6023eab10a16eb009
>> >>> >
>> >>>
>> https://git.postgresql.org/gitweb/?p=postgresql.git;a=commit;h=5e7bbb528638c0f6d585bab107ec7a19e3a39deb
>> >>> >
>> >>>
>> https://git.postgresql.org/gitweb/?p=postgresql.git;a=commit;h=e07633646a22734e85d7fc58a66855f747128e6b
>> >>> >
>> >>>
>> https://git.postgresql.org/gitweb/?p=postgresql.git;a=commit;h=9e101cf60612f4be4f855d7393531900c2986a55
>> >>> >
>> >>>
>> https://git.postgresql.org/gitweb/?p=postgresql.git;a=commit;h=09dfd430118f1fadf52a782db5ee161b1eb16337
>> >>> >
>> >>>
>> https://git.postgresql.org/gitweb/?p=postgresql.git;a=commit;h=7c89f8a5b810d10dae300ec58ea7d70024e9123e
>> >>> >
>> >>>
>> https://git.postgresql.org/gitweb/?p=postgresql.git;a=commit;h=a9a4a7ad565b136cbee735d4bb505d98d06da522
>> >>> >
>> >>> >
>> >>> >> The topic has not been discussed yet in pgpool-hackers, but I
>> >>> >> personally think we need to do something soon or later anyway. My
>> >>> >> concern is that that might require user-visible changes like:
>> >>> >>
>> >>> >> master
>> >>> >> slave
>> >>> >> white_function_list
>> >>> >> black_function_list
>> >>> >> black_query_pattern_list
>> >>> >> white_memqcache_table_list
>> >>> >> black_memqcache_table_list
>> >>> >> follow_master_command
>> >>> >> relcache_query_target = master
>> >>> >> [maybe more...]
>> >>> >>
>> >>> >> Is it possible for you to come with a proposal for this so that
>> pgpool
>> >>> >> hackers could start discussion?
>> >>> >>
>> >>> >> Best regards,
>> >>> >> --
>> >>> >> Tatsuo Ishii
>> >>> >> SRA OSS, Inc. Japan
>> >>> >> English: http://www.sraoss.co.jp/index_en.php
>> >>> >> Japanese:http://www.sraoss.co.jp
>> >>> >
>> >>> >
>> >>> > Sure, I will create proposal for different sections and patches to
>> >>> discuss
>> >>> > it further.
>> >>> >
>> >>> > Thanks
>> >>> > Umar Hayat
>> >>>
>> >>
>>