Thread: [PATCH] Change "checkpoint starting" message to use "wal"

[PATCH] Change "checkpoint starting" message to use "wal"

From
Christoph Berg
Date:
A customer was complaining that the "checkpoint starting: xlog"
message was not included in the grand PG10 rename of user-visible
bits. The attached patch fixes that.

Christoph
-- 
Senior Berater, Tel.: +49 2166 9901 187
credativ GmbH, HRB Mönchengladbach 12080, USt-ID-Nummer: DE204566209
Trompeterallee 108, 41189 Mönchengladbach
Geschäftsführung: Dr. Michael Meskes, Jörg Folz, Sascha Heuer
Unser Umgang mit personenbezogenen Daten unterliegt
folgenden Bestimmungen: https://www.credativ.de/datenschutz

Attachment

Re: [PATCH] Change "checkpoint starting" message to use "wal"

From
Michael Paquier
Date:
On Thu, Nov 29, 2018 at 09:47:09AM +0100, Christoph Berg wrote:
> A customer was complaining that the "checkpoint starting: xlog"
> message was not included in the grand PG10 rename of user-visible
> bits. The attached patch fixes that.

At the same time it would make sense to rename CHECKPOINT_CAUSE_XLOG?  I
would not risk back-patching such a change as that would be annoying for
tools parsing logs like pgbadger.
--
Michael

Attachment

Re: [PATCH] Change "checkpoint starting" message to use "wal"

From
Christoph Berg
Date:
Re: Michael Paquier 2018-11-29 <20181129085902.GD9004@paquier.xyz>
> On Thu, Nov 29, 2018 at 09:47:09AM +0100, Christoph Berg wrote:
> > A customer was complaining that the "checkpoint starting: xlog"
> > message was not included in the grand PG10 rename of user-visible
> > bits. The attached patch fixes that.
> 
> At the same time it would make sense to rename CHECKPOINT_CAUSE_XLOG?  I
> would not risk back-patching such a change as that would be annoying for
> tools parsing logs like pgbadger.

There's hundreds of other internal uses of xlog that were not touched
either, only the user-facing parts were changed.

If this gets accepted, I'll submit a patch for pgbadger so it can be
updated before PG12 hits the field.

Christoph


Re: [PATCH] Change "checkpoint starting" message to use "wal"

From
Michael Paquier
Date:
On Thu, Nov 29, 2018 at 10:04:12AM +0100, Christoph Berg wrote:
> There's hundreds of other internal uses of xlog that were not touched
> either, only the user-facing parts were changed.

I have heard of them ;)
Just wondering if this one is worth renaming as the variable is
isolated.  It is not a big deal to do nothing though.

> If this gets accepted, I'll submit a patch for pgbadger so it can be
> updated before PG12 hits the field.

Your change looks acceptable to me FWIW.
--
Michael

Attachment

Re: [PATCH] Change "checkpoint starting" message to use "wal"

From
Stephen Frost
Date:
Greetings,

* Christoph Berg (christoph.berg@credativ.de) wrote:
> A customer was complaining that the "checkpoint starting: xlog"
> message was not included in the grand PG10 rename of user-visible
> bits. The attached patch fixes that.

+1

Thanks!

Stephen

Attachment

Re: [PATCH] Change "checkpoint starting" message to use "wal"

From
Robert Haas
Date:
On Thu, Nov 29, 2018 at 4:10 AM Michael Paquier <michael@paquier.xyz> wrote:
> On Thu, Nov 29, 2018 at 10:04:12AM +0100, Christoph Berg wrote:
> > There's hundreds of other internal uses of xlog that were not touched
> > either, only the user-facing parts were changed.
>
> I have heard of them ;)
> Just wondering if this one is worth renaming as the variable is
> isolated.  It is not a big deal to do nothing though.

Well, if we rename the user-visible part but not the internal part,
then they don't match, which is odd.

-- 
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company


Re: [PATCH] Change "checkpoint starting" message to use "wal"

From
Alvaro Herrera
Date:
On 2018-Nov-29, Robert Haas wrote:

> On Thu, Nov 29, 2018 at 4:10 AM Michael Paquier <michael@paquier.xyz> wrote:
> > On Thu, Nov 29, 2018 at 10:04:12AM +0100, Christoph Berg wrote:
> > > There's hundreds of other internal uses of xlog that were not touched
> > > either, only the user-facing parts were changed.
> >
> > I have heard of them ;)
> > Just wondering if this one is worth renaming as the variable is
> > isolated.  It is not a big deal to do nothing though.
> 
> Well, if we rename the user-visible part but not the internal part,
> then they don't match, which is odd.

But we already did that when we renamed all the xlog to WAL terminology
... why do we care about it now particularly?

-- 
Álvaro Herrera                https://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services


Re: [PATCH] Change "checkpoint starting" message to use "wal"

From
Stephen Frost
Date:
Greetings,

* Alvaro Herrera (alvherre@2ndquadrant.com) wrote:
> On 2018-Nov-29, Robert Haas wrote:
> > On Thu, Nov 29, 2018 at 4:10 AM Michael Paquier <michael@paquier.xyz> wrote:
> > > On Thu, Nov 29, 2018 at 10:04:12AM +0100, Christoph Berg wrote:
> > > > There's hundreds of other internal uses of xlog that were not touched
> > > > either, only the user-facing parts were changed.
> > >
> > > I have heard of them ;)
> > > Just wondering if this one is worth renaming as the variable is
> > > isolated.  It is not a big deal to do nothing though.
> >
> > Well, if we rename the user-visible part but not the internal part,
> > then they don't match, which is odd.
>
> But we already did that when we renamed all the xlog to WAL terminology
> ... why do we care about it now particularly?

I thought the idea was that we'd adjust things in the actual code as
that code was refactored or adjusted for other reasons, to minimize the
back-patching pain.  That said, in this particular case that would mean
just changing one variable when the other related ones aren't changed
and I suspect that might just be more confusing than having this
difference between the code and the user-messages.

So, at least in this instance, my feeling is that we keep the variable
as-is and just adjust the user message.  When, down the road, there's a
larger refactoring or change in this part of the code, that would be the
time to change the code to refer to WAL instead of XLOG.

Thanks!

Stephen

Attachment

RE: [PATCH] Change "checkpoint starting" message to use "wal"

From
"Tsunakawa, Takayuki"
Date:
From: Christoph Berg [mailto:christoph.berg@credativ.de]
> A customer was complaining that the "checkpoint starting: xlog"
> message was not included in the grand PG10 rename of user-visible bits.
> The attached patch fixes that.

Can we make use of this chance to change elog() to ereport(), so that the two messages are translated?


Regards
Takayuki Tsunakawa




Re: [PATCH] Change "checkpoint starting" message to use "wal"

From
Stephen Frost
Date:
Greetings,

* Tsunakawa, Takayuki (tsunakawa.takay@jp.fujitsu.com) wrote:
> From: Christoph Berg [mailto:christoph.berg@credativ.de]
> > A customer was complaining that the "checkpoint starting: xlog"
> > message was not included in the grand PG10 rename of user-visible bits.
> > The attached patch fixes that.
>
> Can we make use of this chance to change elog() to ereport(), so that the two messages are translated?

+1, and we should probably look for nearby cases to fix also..

Thanks!

Stephen

Attachment

Re: [PATCH] Change "checkpoint starting" message to use "wal"

From
Michael Paquier
Date:
On Thu, Nov 29, 2018 at 07:31:02PM -0500, Stephen Frost wrote:
> * Tsunakawa, Takayuki (tsunakawa.takay@jp.fujitsu.com) wrote:
>> From: Christoph Berg [mailto:christoph.berg@credativ.de]
>> > A customer was complaining that the "checkpoint starting: xlog"
>> > message was not included in the grand PG10 rename of user-visible bits.
>> > The attached patch fixes that.
>>
>> Can we make use of this chance to change elog() to ereport(), so that the two messages are translated?
>
> +1, and we should probably look for nearby cases to fix also..

Good point.  +1.
--
Michael

Attachment

Re: [PATCH] Change "checkpoint starting" message to use "wal"

From
Peter Eisentraut
Date:
On 29/11/2018 09:47, Christoph Berg wrote:
> A customer was complaining that the "checkpoint starting: xlog"
> message was not included in the grand PG10 rename of user-visible
> bits. The attached patch fixes that.

committed

-- 
Peter Eisentraut              http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services