all WAL records that the DDL execution generates.
>
Yes, the DDL replication patch uses the parse tree and catalog
information to generate a deparsed form of DDL statement which is WAL
logged and used to replicate DDLs.
--
With Regards,
Amit Kapila.
sounds like we are checking whether datatype has any
operator class for btree or hash access methods but we are actually
checking if there is an equality operator (function) defined in the
default op class for those access methods. Am, I missing something?
--
With Regards,
Amit Kapila.
r seeing this, I am thinking about whether we add this restriction
on the Subscription page [1] or Restrictions page [2] as proposed. Do
you others have any preference?
[1] -
https://www.postgresql.org/docs/devel/logical-replication-subscription.html
[2] -
https://www.postgresql.org/docs/devel/logical-replication-restrictions.html
--
With Regards,
Amit Kapila.
On Tue, Jul 11, 2023 at 4:54 AM Peter Smith wrote:
>
> On Mon, Jul 10, 2023 at 2:21 PM Amit Kapila wrote:
> >
> > On Mon, Jul 10, 2023 at 7:55 AM Peter Smith wrote:
> > >
> > > On Sat, Jul 8, 2023 at 1:49 PM Amit Kapila
> > > wrote:
> >
03 basically improved performance from first two patches
>
> Agree, 0003 is definitely a good addition which was missing earlier.
>
+1.
[1] -
https://www.postgresql.org/message-id/CAA4eK1JOZHmy2o2F2wTCPKsjpwDiKZPOeTa_jt%3Dwm2JLbf-jsg%40mail.gmail.com
--
With Regards,
Amit Kapila.
brought up the amgettuple issue, which is fair. So,
> overall, as far as I can see, we can
> easily add hash indexes but all others are either very hard to add or not
> possible.
>
Agreed. So, let's update the patch with comments indicating the
challenges for supporting the other index types than btree and hash.
--
With Regards,
Amit Kapila.
r example, debezium.
>
Can you explain the difference and problem you are seeing? As per my
understanding, this is the behavior from the time logical replication
has been introduced.
[1] -
https://www.postgresql.org/message-id/TYAPR01MB5866C7B6086EB74918910F74F527A%40TYAPR01MB5866.jpnprd01.prod.outlook.com
--
With Regards,
Amit Kapila.
nnot be
replicated for the published tables that specifies REPLICA IDENTITY
FULL but the table includes an attribute whose datatype doesn't have
an equality operator defined for it on the subscriber."?
--
With Regards,
Amit Kapila.
? As per my understanding if we don't have PK or
replica identity then after the index scan, we do tuples_equal which
will fail for GIST or SP-GIST. Am, I missing something?
--
With Regards,
Amit Kapila.
On Mon, Jul 10, 2023 at 7:55 AM Peter Smith wrote:
>
> On Sat, Jul 8, 2023 at 1:49 PM Amit Kapila wrote:
> >
> > On Fri, Jul 7, 2023 at 1:36 PM Masahiko Sawada
> > wrote:
> > >
> > > I prefer the first suggestion. I've attached the updated patch.
On Sun, Jul 9, 2023 at 1:01 PM Bharath Rupireddy
wrote:
>
> On Fri, Jun 16, 2023 at 3:26 PM Amit Kapila wrote:
> >
> >
> > 1. Can you please try to explain the functionality of the overall
> > patch somewhere in the form of comments and or commit message?
&g
the individual tablesync workers can sync the
schema of partitions and their data. And when publish_via_root is
true, then the table sync worker is responsible to sync parent and
child tables along with data. Do you think such a mechanism can
address the partition table related cases?
--
With Regards,
Amit Kapila.
iction. It has existed before
> the
> commit. I (or my colleague) will post the patch to add the description, maybe
> after [1] is committed.
>
I don't think there is any dependency between the two. So, feel free
to post the patch separately.
--
With Regards,
Amit Kapila.
how about: "Candidate indexes must be btree,
non-partial, and the leftmost index column must be a non-expression
and reference to a published table column (i.e. cannot consist of only
expressions)."?
--
With Regards,
Amit Kapila.
On Wed, Jul 5, 2023 at 7:54 PM Ashutosh Bapat
wrote:
>
> On Wed, Jul 5, 2023 at 7:07 PM Euler Taveira wrote:
> >
> > On Wed, Jul 5, 2023, at 7:56 AM, Alvaro Herrera wrote:
> >
> > On 2023-Jul-05, Amit Kapila wrote:
> >
> > > I think after
no invalid message type will be reported.
>
Yeah, but it would still be displayed both in context and the actual message.
> I guess there's a
> point to that argument as well.
>
One point to note is that the user may also get confused if the actual
ERROR says message type as 'X' and the context says '???'. I feel in
this case duplicate information is better than different information.
--
With Regards,
Amit Kapila.
On Wed, Jul 5, 2023 at 7:20 PM Ashutosh Bapat
wrote:
>
> On Wed, Jul 5, 2023 at 2:29 PM Amit Kapila wrote:
> >
> > On Wed, Jul 5, 2023 at 2:28 PM Amit Kapila wrote:
> > >
> > > On Mon, Jul 3, 2023 at 4:49 PM vignesh C wrote:
> > > >
> > >
On Fri, Jun 30, 2023 at 8:26 AM Amit Kapila wrote:
>
> On Wed, Jun 28, 2023 at 7:26 AM Zhijie Hou (Fujitsu)
> wrote:
> >
>
> Thanks for the verification. Unless someone has any further comments
> or suggestions, I'll push this next week sometime.
>
Pushed but forg
mount of
> improvement, but not sure about the later patches yet.
>
I suggest then we should focus first on those 3, get them committed
and then look at the remaining.
> Amit Kapila , 3 Tem 2023 Pzt, 08:59 tarihinde
> şunu yazdı:
> > On thinking about this, I think the primary benefit we
On Wed, Jul 5, 2023 at 4:26 PM Alvaro Herrera wrote:
>
> On 2023-Jul-05, Amit Kapila wrote:
>
> > I think after returning "???" from logicalrep_message_type(), the
> > above is possible when we get the error: "invalid logical replication
> > message typ
On Wed, Jul 5, 2023 at 12:02 PM Masahiko Sawada wrote:
>
> On Wed, Jul 5, 2023 at 2:46 PM Amit Kapila wrote:
> >
> > On Wed, Jul 5, 2023 at 9:01 AM Peter Smith wrote:
> > >
> > > Hi. Here are some review comments for this patc
On Wed, Jul 5, 2023 at 2:28 PM Amit Kapila wrote:
>
> On Mon, Jul 3, 2023 at 4:49 PM vignesh C wrote:
> >
> > +1 for the first version patch, I also felt the first version is
> > easily understandable.
> >
>
> Okay, please find the slightly updated version (ch
a
day or two.
--
With Regards,
Amit Kapila.
t of these checks indicates that the leftmost column of the
index should be non-expression, second and third indicates what you
suggest in your wording. We can also think that what you wrote in a
way is a superset of "leftmost index column is a non-expression" and
"leftmost index column should be present in remote rel" but I feel it
would be better to explicit about the first part as it is easy to
understand for users at least in docs.
--
With Regards,
Amit Kapila.
n what about
the case when we forgot to handle some message in
logicalrep_message_type() but handled it in apply_dispatch()? Isn't it
better to return the 'action' from the function
logicalrep_message_type() for unknown type? That way the information
could be a bit better and we can easily catch the code bug as well.
--
With Regards,
Amit Kapila.
Yeah, I think it is good to mention that. Accordingly, the comments
atop build_replindex_scan_key(),
FindUsableIndexForReplicaIdentityFull(), IsIndexOnlyOnExpression()
should also be adjusted.
--
With Regards,
Amit Kapila.
On Mon, Jul 3, 2023 at 9:42 AM Amit Kapila wrote:
>
> On Wed, Jun 28, 2023 at 12:02 PM Hayato Kuroda (Fujitsu)
> wrote:
>
> > But in the later patch the tablesync worker tries to reuse the slot during
> > the
> > synchronization, so in this case the a
drop them but at this stage I
feel we need to investigate why those are not helping?
--
With Regards,
Amit Kapila.
lots on primary, so launcher disconnects from primary and waits a time
> period (up to 3min).
> Even if the apply worker creates the slot on publisher, but the launcher on
> standby
> cannot notice that. The synchronization may start 3 min later.
>
I feel this should be based on some GUC l
eceived for this transaction during which we would have outputted
> the begin message, this condition is to skip outputting begin message
> if the begin message was already outputted.
>
I feel the use of last_write has reduced the readability of this part
of the code. It may be that we can add comments to make it clear but I
feel your previous version was much easier to understand.
--
With Regards,
Amit Kapila.
---
BEGIN 739
message: transactional: 1 prefix: test, sz: 4 content:msg1
COMMIT 739
message: transactional: 0 prefix: test, sz: 4 content:msg2
(4 rows)
Thoughts?
--
With Regards,
Amit Kapila.
On Wed, Jun 28, 2023 at 7:26 AM Zhijie Hou (Fujitsu)
wrote:
>
> On Monday, June 26, 2023 12:18 PM Amit Kapila wrote:
> >
> > Fair enough. If we could have been convinced of this then we can keep the
> > required change only for HEAD. But anyway let's remove the work rel
ync slot workers corresponding to
slots that need sync on standby and apply workers for active
subscriptions on primary (which will be a subscriber in this context).
If this is correct, then do you expect to launch a separate kind of
standby launcher for sync slots?
--
With Regards,
Amit Kapila.
On Wed, Jun 28, 2023 at 12:19 PM Drouvot, Bertrand
wrote:
>
> On 6/26/23 12:34 PM, Amit Kapila wrote:
> > On Mon, Jun 26, 2023 at 11:15 AM Drouvot, Bertrand
> > wrote:
> >>
> >> On 6/20/23 12:22 PM, Amit Kapila wrote:
> >>> On Mon, Jun 19
_flush in
memory but not in the disk? BTW, have we ensured that discarding the
additional records are already sent to the subscriber, if so, why for
those records confirmed_flush LSN is not progressed?
--
With Regards,
Amit Kapila.
0003) which
> allows reusing connection.
>
Why did you change the application name during the connection?
--
With Regards,
Amit Kapila.
eating in patch 0003. Is it possible to avoid disconnect/connect
each time the patch needs to reuse the same tablesync worker? Once, we
do that and save the cost of drop_slot and associated network round
trip, you may see the benefit of 0003 and 0004 patches.
--
With Regards,
Amit Kapila.
On Mon, Jun 26, 2023 at 11:15 AM Drouvot, Bertrand
wrote:
>
> On 6/20/23 12:22 PM, Amit Kapila wrote:
> > On Mon, Jun 19, 2023 at 9:56 PM Drouvot, Bertrand
> > wrote:
>
> >> In such a case (slot valid on the primary but invalidated on the standby)
> &
bug. Why is pg_decode_message()
> exception?
>
I can't see a reason why we shouldn't have a similar check for
transactional messages. So, agreed this is a bug.
--
With Regards,
Amit Kapila.
On Fri, Jun 23, 2023 at 10:07 PM Andres Freund wrote:
>
> On 2023-06-23 14:04:15 +0530, Amit Kapila wrote:
> > OTOH, if the above theory is wrong or people are not convinced, I am
> > okay with removing all the changes in commits 72e78d831a and
> > 3ba59ccc89.
>
>
On Thu, Jun 22, 2023 at 9:16 AM Amit Kapila wrote:
>
> On Wed, Jun 21, 2023 at 10:57 AM Andres Freund wrote:
> >
> > As far as I can tell 72e78d831a as-is is just bogus. Unfortunately that
> > likely
> > also means 3ba59ccc89 is not right.
> >
>
> In
of the statement. We may need to
think about that as well.
Having said that, we are still trying to write a patch to see how it
looks, which may help us to jointly evaluate if we can do anything
better.
--
With Regards,
Amit Kapila.
On Wed, Jun 21, 2023 at 8:12 AM Masahiko Sawada wrote:
>
> On Thu, Jun 15, 2023 at 7:50 PM Amit Kapila wrote:
> >
> > On Tue, Jun 13, 2023 at 2:06 PM Masahiko Sawada
> > wrote:
> > >
> > > On Sun, Jun 11, 2023 at 5:31 AM Andres Freun
On Tue, Jun 20, 2023 at 9:02 AM Peter Smith wrote:
>
> FYI - I have created and tested back-patches for Amit's v5 patch,
> going all the way to REL_10_STABLE.
>
Pushed. I haven't used PG10 patch as REL_10_STABLE is out of support now.
--
With Regards,
Amit Kapila.
s relation level lock is
> supposed to take care of?
>
Yeah, I also can't see why that shouldn't be sufficient for VACUUM.
Assuming your observation is correct, what do you suggest doing in
this regard?
--
With Regards,
Amit Kapila.
for an index).
--
With Regards,
Amit Kapila.
On Tue, Jun 20, 2023 at 10:39 PM Jacob Champion wrote:
>
> On Fri, Jun 16, 2023 at 9:24 PM Amit Kapila wrote:
>
> > The other idea that came across my mind was to provide some schema
> > mapping kind of feature on subscribers where we could route the tuples
> > from
h + test infra, and they all passed.
> >
> >We still have a week till beta2 wrap, so I'd say push. If the buildfarm
> >gets unhappy you can revert.
>
> Hah. Somehow I confused myself into thinking you're wrapping later today.
> Calendar math vs Andres: 6753:3
>
Can we close the open item corresponding to this after your commit 0d369ac650?
--
With Regards,
Amit Kapila.
On Mon, Jun 19, 2023 at 9:56 PM Drouvot, Bertrand
wrote:
>
> On 6/19/23 12:03 PM, Amit Kapila wrote:
> > On Mon, Jun 19, 2023 at 11:34 AM Drouvot, Bertrand
> > wrote:
> >>
> >> Also I think we need to handle the case of invalidated replication
> >>
On Mon, Jun 19, 2023 at 5:13 PM Amit Kapila wrote:
>
> On Sun, Jun 18, 2023 at 12:18 AM Peter Geoghegan wrote:
> >
> > On Sat, Jun 17, 2023 at 11:29 AM Jaime Casanova
> > wrote:
> > > I have been testing 16beta1, last commit
> > > a14e75eb0b6a738
else, and
> > today I found a core with the attached backtrace.
>
> The assertion that fails is the IsPageLockHeld assertion from commit
> 72e78d831a.
>
I'll look into this and share my analysis.
--
With Regards,
Amit Kapila.
primary/publisher should have been dropped/invalidated by that time.
BTW, does the patch handles drop of logical slots on standby when the
same slot is dropped on publisher/primary?
--
With Regards,
Amit Kapila.
ariable name telems to telems_present to make it bit easy
to follow.
7. It would be better if we can further split the patch to move Alter
case into a separate patch. That will help us to focus on reviewing
Create/Drop in detail.
--
With Regards,
Amit Kapila.
On Mon, Jun 19, 2023 at 6:50 AM Masahiko Sawada wrote:
>
> On Sat, Jun 17, 2023 at 6:45 PM Amit Kapila wrote:
> >
> > On Tue, May 16, 2023 at 8:00 PM Masahiko Sawada
> > wrote:
> > >
> > > On Thu, May 11, 2023 at 5:12 PM Masahiko Sawada
> >
backpatched to 15.
> Since in pg15, since we don't create the subscription stats at CREATE
> SUBSCRIPTION time but do when the first error is reported,
>
AFAICS, the call to pgstat_create_subscription() is present in
CreateSubscription() in 15 as well, so, I don't get your point.
--
With Regards,
Amit Kapila.
On Fri, Jun 16, 2023 at 7:15 PM Peter Eisentraut wrote:
>
> On 15.06.23 04:49, Amit Kapila wrote:
> >
> > Now, along with this change, there is a change in errhint as well
> > which I am not sure about whether to backpatch or not. I think we have
> > the followi
On Sat, Jun 17, 2023 at 1:51 AM Jacob Champion wrote:
>
> On Fri, Jun 16, 2023 at 6:26 AM Amit Kapila wrote:
>
> > > But if it's because you've implemented a partitioning scheme of your
> > > own (the docs still list reasons you might want to [2], even today),
>
t by you where
people want to use inheritance as a mechanism for partitioning but I
feel those would still be in the minority. Personally, I am not very
excited about this idea.
--
With Regards,
Amit Kapila.
er
standby's promotion, the subscriber can start syncing from the wrong
position.
--
With Regards,
Amit Kapila.
On Thu, Jun 15, 2023 at 11:18 PM Alvaro Herrera wrote:
>
> On 2023-Jun-13, Amit Kapila wrote:
>
> > I'll push this tomorrow unless there are any suggestions or comments.
>
> Note the proposed commit message is wrong about which commit is to blame
> for the origina
e any such
reports and can we see how much is the benefit with large
transactions? Because we do have the handling of concurrent aborts
during sys table scans and that might help sometimes for large
transactions.
--
With Regards,
Amit Kapila.
makes error strings in the code look a bit shorter. I
think it is better to park the patch for the next CF to avoid
forgetting about it.
--
With Regards,
Amit Kapila.
On Wed, Jun 14, 2023 at 6:52 PM Peter Eisentraut wrote:
>
> On 14.06.23 05:09, Amit Kapila wrote:
> > The latest version looks good to me as well. I think we should
> > backpatch this change as this is a user-facing message change in docs
> > and code. What do y
On Wed, Jun 14, 2023 at 9:25 AM Peter Smith wrote:
>
> On Wed, Jun 14, 2023 at 1:10 PM Amit Kapila wrote:
> >
> > On Wed, May 17, 2023 at 11:57 AM Peter Smith wrote:
> > >
> > > On Wed, May 17, 2023 at 2:53 PM Robert Sjöblom
> > > wrote:
&g
this change as this is a user-facing message change in docs
and code. What do you guys think?
--
With Regards,
Amit Kapila.
On Tue, Jun 13, 2023 at 2:25 PM Amit Kapila wrote:
>
> On Fri, May 12, 2023 at 3:28 PM Zhijie Hou (Fujitsu)
> wrote:
> >
> >
> > I can reproduce this via gdb following similar steps in [1].
> >
> > I think we need to move this call into a transaction as wel
avoid having an additional ObjTree layer unless you or others
feel we need it. I think that will reduce the overall patch size and
help us to focus on one of the approaches. Surely, as suggested by
you, we should also evaluate if we can generate this code for the
various command types.
--
With Regards,
Amit Kapila.
On Tue, Jun 13, 2023 at 2:25 PM Amit Kapila wrote:
>
> On Fri, May 12, 2023 at 3:28 PM Zhijie Hou (Fujitsu)
> wrote:
> >
> >
> > I can reproduce this via gdb following similar steps in [1].
> >
> > I think we need to move this call into a transaction as wel
ssue following the steps mentioned by you
and the proposed patch to fix the issue looks good to me.
--
With Regards,
Amit Kapila.
On Thu, Jun 8, 2023 at 5:32 PM shveta malik wrote:
>
> On Thu, Jun 8, 2023 at 10:36 AM Amit Kapila wrote:
>
> Please find new set of patches addressing below:
> a) Issue mentioned by Wang-san in [1],
> b) Comments from Peter given in [2]
> c) Comments from Amit given
unning->nextXid)". I think the main problem is that we started
decoding immediately from the point where we restored a snapshot as at
that point we could have some partial running xacts.
--
With Regards,
Amit Kapila.
if we want to optimize the default mode such
that we only WAL log the DDL with user-specified options then
appending options for default GUCs would make the string to be WAL
logged unnecessarily long. I am thinking that in default mode we can
allow subscribers to perform operations with their default respective
GUCs.
--
With Regards,
Amit Kapila.
On Thu, Jun 8, 2023 at 6:32 AM Masahiko Sawada wrote:
>
> On Mon, Jun 5, 2023 at 3:15 AM Amit Kapila wrote:
> >
> > On Fri, May 26, 2023 at 6:18 PM Masahiko Sawada
> > wrote:
> > >
> > > On Thu, May 25, 2023 at 5:41 PM Amit Kapila
> > >
On Tue, Jun 6, 2023 at 4:26 PM Amit Kapila wrote:
>
>
> Few assorted comments:
> ===
>
Few comments on 0008* patch:
==
1. After 0008*, deparse_CreateStmt(), forms dpcontext before forming
identity whereas it is required only after it. It
On Tue, Jun 6, 2023 at 7:45 AM Zhijie Hou (Fujitsu)
wrote:
>
> On Monday, June 5, 2023 8:03 AM Amit Kapila wrote:
> >
> >
> > I am able to reproduce the problem with the script and steps mentioned by
> > you. The patch provided by you fixes the problem and l
alysis/thoughts. BTW, do you mean that it skips
the assignment (a) because the assignment record is before we reach a
consistent point, or (b) because we start reading WAL after the
assignment, or (c) something else? If you intend to say (a) then can
you please point me to the code you are referring to for the same?
--
With Regards,
Amit Kapila.
/test/modules/test_ddl_deparse_regress/t/001_compare_dumped_results.pl.orig
Will this file require for the 0008 patch? Or is this just a leftover?
--
With Regards,
Amit Kapila.
refactoring for v16.
However, sometimes, we do decide to backpatch refactoring in tests to
avoid backpatch effort. I am not completely sure if that is the case
here.
--
With Regards,
Amit Kapila.
On Wed, May 17, 2023 at 8:34 AM Amit Kapila wrote:
>
> On Wed, May 17, 2023 at 7:18 AM Zhijie Hou (Fujitsu)
> wrote:
> >
> >
> > The attached tap test shows how the failure happened.
> >
>
> I haven't yet tried to reproduce it but will try later sometime
On Fri, May 26, 2023 at 6:18 PM Masahiko Sawada wrote:
>
> On Thu, May 25, 2023 at 5:41 PM Amit Kapila wrote:
>
> I've attached the updated patch. Please review it.
>
Few comments:
1.
+ /* get the owner for ACL and RLS checks */
+ run_as_owner = MySubscription->runaso
ot;ALL", see [2])
>
Right, this should be changed.
--
With Regards,
Amit Kapila.
On Thu, May 25, 2023 at 12:33 PM Masahiko Sawada wrote:
>
> On Tue, May 23, 2023 at 8:21 PM Amit Kapila wrote:
> >
> > On Mon, May 22, 2023 at 6:06 PM Masahiko Sawada
> > wrote:
> > >
> > > Thank you for updating the patch! Here are review comments:
&g
the permission to the target table. The
> test passes only if run_as_owner = true works.
>
Why in the test do you need to give additional permissions to
regress_admin2 when the actual operation has to be performed by the
table owner?
+# Because the initial data sync is working as the t
feature complex
> and I'm not sure users can use it properly.
>
> Third idea is that since the use case of synchronizing the whole
> database can be achievable even by pg_dump(all), we support
> synchronizing only tables (+ indexes) in the initial sync feature,
> which can not b
further
discussion, it won't be easy to conclude that.
Tom/Nathan, do you have any further suggestions here?
[1] -
https://www.postgresql.org/message-id/capphfdsld6e--epngqxenqlp6dlwunzrpmcnyb3wj87wr7u...@mail.gmail.com
--
With Regards,
Amit Kapila.
(it's very rare and only happened once in 4
> months) which I think is due to the low frequent reload in apply worker.
>
> The attached tap test shows how the failure happened.
>
I haven't yet tried to reproduce it but will try later sometime.
Thanks for your analysis.
[1] - https://www.postgresql.org/message-id/2138662.1623460441%40sss.pgh.pa.us
--
With Regards,
Amit Kapila.
eration on a particular table, it will switch roles to the table
owner and perform the operation with the table owner's privileges." to
be bit more specific about initial sync process as well?
[1] - https://www.postgresql.org/docs/devel/logical-replication-security.html
--
With Regards,
Amit Kapila.
On Fri, May 12, 2023 at 5:25 PM Ajin Cherian wrote:
>
> On Fri, May 12, 2023 at 1:49 PM Amit Kapila wrote:
> >
>
> I tried the following test:
>
>
> Repeat On the publisher and subscriber:
> /* Create role regress_alice with NOSUPERUSER on
On Fri, May 12, 2023 at 10:33 AM Amit Kapila wrote:
>
> On Fri, May 12, 2023 at 7:38 AM Masahiko Sawada wrote:
> >
> > On Thu, May 11, 2023 at 2:04 PM Amit Kapila wrote:
> > >
> > > On Fri, Apr 28, 2023 at 2:35 PM Hayato Kuroda (Fujitsu)
> &
On Fri, May 12, 2023 at 7:38 AM Masahiko Sawada wrote:
>
> On Thu, May 11, 2023 at 2:04 PM Amit Kapila wrote:
> >
> > On Fri, Apr 28, 2023 at 2:35 PM Hayato Kuroda (Fujitsu)
> > wrote:
> > >
> > > Dear hackers,
> > >
> > &
On Fri, May 12, 2023 at 9:10 AM Masahiko Sawada wrote:
>
> On Fri, May 12, 2023 at 1:12 AM Robert Haas wrote:
> >
> > On Thu, May 11, 2023 at 7:38 AM Amit Kapila wrote:
> > > Do we want the initial sync to also respect 'run_as_owner' option? I
> > > might
be missing something but I don't see anything in the docs about
initial sync interaction with this option. In the commit a2ab9c06ea,
we did the permission checking during the initial sync so I thought we
should do it here as well.
--
With Regards,
Amit Kapila.
2D%3DeuO-Qs-GAg%40mail.gmail.com
--
With Regards,
Amit Kapila.
da-San came up
with a new patch with a different design based on the discussion here.
I haven't looked at it yet though.
[1] -
https://about.gitlab.com/blog/2019/02/13/delayed-replication-for-disaster-recovery-with-postgresql/
[2] - https://dev.mysql.com/doc/refman/8.0/en/replication-delayed.html
[3] -
https://www.postgresql.org/message-id/73b06a32-56ab-4056-86ff-e307f3c316f1%40www.fastmail.com
--
With Regards,
Amit Kapila.
On Tue, May 9, 2023 at 12:44 PM Drouvot, Bertrand
wrote:
>
> On 5/9/23 8:02 AM, Amit Kapila wrote:
> > On Mon, May 8, 2023 at 1:45 PM Drouvot, Bertrand
> > wrote:
>
> >
> > Why not initialize the cascading standby node just before the standby
> >
SndWakeup(switchedTLI, true);
> + ConditionVariableBroadcast(>cv);
>
> After the change, we wakeup physical walsender regardless of switchedTLI flag.
> Is this intentional ? if so, I think It would be better to update the
> comments above this.
>
This raises the question of whether we need this condition variable
logic for physical walsenders?
--
With Regards,
Amit Kapila.
appen because
apply worker should eventually update these relations state to 'r'. If
this is reproducible, we can try two things to further investigate the
issue: (a) Disable and Enable the subscription once and see if that
helps; (b) try increasing the LOG level to DEBUG2 and see if we get
any useful LOG message.
--
With Regards,
Amit Kapila.
On Tue, May 9, 2023 at 7:50 AM Masahiko Sawada wrote:
>
> On Mon, May 8, 2023 at 8:09 PM Amit Kapila wrote:
> >
> >
> > I think it is only possible for the leader apply can worker to try to
> > receive the error message from an error queue after your 0002 patch.
>
t_name=NONE. It would be better to
discuss this in a separate thread to seek the opinion of others.
--
With Regards,
Amit Kapila.
On Mon, May 8, 2023 at 1:45 PM Drouvot, Bertrand
wrote:
>
> On 5/8/23 4:42 AM, Amit Kapila wrote:
> > On Sat, May 6, 2023 at 9:33 PM Drouvot, Bertrand
> > wrote:
> >>
> >> On 5/6/23 3:28 PM, Amit Kapila wrote:
> >>> On Sat, May 6, 2023 at 1:52 PM
901 - 1000 of 5977 matches
Mail list logo