Обсуждение: Improve read_local_xlog_page_guts by replacing polling with latch-based waiting

Hi hackers,

During a performance run [1], I observed heavy polling in
read_local_xlog_page_guts(). Heikki’s comment from a few months ago
suggests replacing the current check–sleep–repeat loop with the
condition-variable (CV) infrastructure used by the walsender:

1) Problem and Background
/*
 * Loop waiting for xlog to be available if necessary
 *
 * TODO: The walsender has its own version of this function, which uses a
 * condition variable to wake up whenever WAL is flushed. We could use the
 * same infrastructure here, instead of the check/sleep/repeat style of
 * loop.
 */

Because read_local_xlog_page_guts() waits for a specific flush or
replay LSN, polling becomes inefficient when waits are long. I built a
POC patch that swaps polling for CVs, but a single global CV (or even
separate “flush” and “replay” CVs) isn’t ideal:
• The wake-up routines don’t know which LSN each waiter cares about,
so they would need to broadcast on every flush/replay.

• Caching the minimum outstanding target LSN could reduce spurious
wake-ups but won’t eliminate them when multiple backends wait for
different LSNs simultaneously.

• The walsender accepts some broadcast overhead via two CVs for
different waiters. A more precise approach would require a request
queue that maps waiters to target LSNs and issues targeted
wake-ups—adding complexity.

2) Proposal
I came across the thread “Implement waiting for WAL LSN replay:
reloaded” [2] by Alexander. The “Implement WAIT FOR” patch in that
thread provides a well-established infrastructure for waiting on WAL
replay in backends. With modest adjustments, it could be generalized.

Main changes in patch v1 Improve read_local_xlog_page_guts by replacing polling
with latch-based waiting:
• Introduce WaitForLSNFlush, analogous to WaitForLSNReplay from the
“WAIT FOR” work.

• Replace the busy-wait in read_local_xlog_page_guts() with
WaitForLSNFlush and WaitForLSNReplay.

• Add wake-up calls in XLogFlush and XLogBackgroundFlush.

Edge Case: Timeline Switch During Wait
/*
* Check which timeline to get the record from.
*
* We have to do it each time through the loop because if we're in
* recovery as a cascading standby, the current timeline might've
* become historical. We can't rely on RecoveryInProgress() because in
* a standby configuration like
*
* A => B => C
*
* if we're a logical decoding session on C, and B gets promoted, our
* timeline will change while we remain in recovery.
*
* We can't just keep reading from the old timeline as the last WAL
* archive in the timeline will get renamed to .partial by
* StartupXLOG().

read_local_xlog_page_guts() re-evaluates the active timeline on each
loop iteration because, on a cascading standby, the current timeline
can become historical. Once that happens, there’s no need to keep
waiting for that timeline. A timeline switch could therefore render an
in-progress wait unnecessary.

One option is to add a wake-up at the point where the timeline switch
occurs, so waiting processes exit promptly. The current approach
chooses not to do this, given that most waits are short and timeline
changes in cascading standby are rare.  Supporting timeline-switch
wake-ups would also require additional handling in both
WaitForLSNFlush and WaitForLSNReplay, increasing complexity.

Comments and suggestions are welcome.

[1] https://www.postgresql.org/message-id/CABPTF7VuFYm9TtA9vY8ZtS77qsT+yL_HtSDxUFnW3XsdB5b9ew@mail.gmail.com
[2] https://www.postgresql.org/message-id/flat/CAPpHfdsjtZLVzxjGT8rJHCYbM0D5dwkO%2BBBjcirozJ6nYbOW8Q%40mail.gmail.com

Best,
Xuneng

Hi,

Attached the wrong patch
v1-0001-Improve-read_local_xlog_page_guts-by-replacing-po.patch. The
correct one is attached again.

On Wed, Aug 27, 2025 at 11:23 PM Xuneng Zhou <xunengzhou@gmail.com> wrote:
>
> Hi hackers,
>
> During a performance run [1], I observed heavy polling in
> read_local_xlog_page_guts(). Heikki’s comment from a few months ago
> suggests replacing the current check–sleep–repeat loop with the
> condition-variable (CV) infrastructure used by the walsender:
>
> 1) Problem and Background
> /*
>  * Loop waiting for xlog to be available if necessary
>  *
>  * TODO: The walsender has its own version of this function, which uses a
>  * condition variable to wake up whenever WAL is flushed. We could use the
>  * same infrastructure here, instead of the check/sleep/repeat style of
>  * loop.
>  */
>
> Because read_local_xlog_page_guts() waits for a specific flush or
> replay LSN, polling becomes inefficient when waits are long. I built a
> POC patch that swaps polling for CVs, but a single global CV (or even
> separate “flush” and “replay” CVs) isn’t ideal:
> • The wake-up routines don’t know which LSN each waiter cares about,
> so they would need to broadcast on every flush/replay.
>
> • Caching the minimum outstanding target LSN could reduce spurious
> wake-ups but won’t eliminate them when multiple backends wait for
> different LSNs simultaneously.
>
> • The walsender accepts some broadcast overhead via two CVs for
> different waiters. A more precise approach would require a request
> queue that maps waiters to target LSNs and issues targeted
> wake-ups—adding complexity.
>
> 2) Proposal
> I came across the thread “Implement waiting for WAL LSN replay:
> reloaded” [2] by Alexander. The “Implement WAIT FOR” patch in that
> thread provides a well-established infrastructure for waiting on WAL
> replay in backends. With modest adjustments, it could be generalized.
>
> Main changes in patch v1 Improve read_local_xlog_page_guts by replacing polling
> with latch-based waiting:
> • Introduce WaitForLSNFlush, analogous to WaitForLSNReplay from the
> “WAIT FOR” work.
>
> • Replace the busy-wait in read_local_xlog_page_guts() with
> WaitForLSNFlush and WaitForLSNReplay.
>
> • Add wake-up calls in XLogFlush and XLogBackgroundFlush.
>
> Edge Case: Timeline Switch During Wait
> /*
> * Check which timeline to get the record from.
> *
> * We have to do it each time through the loop because if we're in
> * recovery as a cascading standby, the current timeline might've
> * become historical. We can't rely on RecoveryInProgress() because in
> * a standby configuration like
> *
> * A => B => C
> *
> * if we're a logical decoding session on C, and B gets promoted, our
> * timeline will change while we remain in recovery.
> *
> * We can't just keep reading from the old timeline as the last WAL
> * archive in the timeline will get renamed to .partial by
> * StartupXLOG().
>
> read_local_xlog_page_guts() re-evaluates the active timeline on each
> loop iteration because, on a cascading standby, the current timeline
> can become historical. Once that happens, there’s no need to keep
> waiting for that timeline. A timeline switch could therefore render an
> in-progress wait unnecessary.
>
> One option is to add a wake-up at the point where the timeline switch
> occurs, so waiting processes exit promptly. The current approach
> chooses not to do this, given that most waits are short and timeline
> changes in cascading standby are rare.  Supporting timeline-switch
> wake-ups would also require additional handling in both
> WaitForLSNFlush and WaitForLSNReplay, increasing complexity.
>
> Comments and suggestions are welcome.
>
> [1] https://www.postgresql.org/message-id/CABPTF7VuFYm9TtA9vY8ZtS77qsT+yL_HtSDxUFnW3XsdB5b9ew@mail.gmail.com
> [2] https://www.postgresql.org/message-id/flat/CAPpHfdsjtZLVzxjGT8rJHCYbM0D5dwkO%2BBBjcirozJ6nYbOW8Q%40mail.gmail.com
>
> Best,
> Xuneng

Hi,

Some changes in v3:
1) Update the note of xlogwait.c to reflect the extending of its use
for flush waiting and internal use for both flush and replay waiting.
2) Update the comment above logical_read_xlog_page which describes the
prior-change behavior of read_local_xlog_page.

Best,
Xuneng

Hi,

On Thu, Aug 28, 2025 at 4:22 PM Xuneng Zhou <xunengzhou@gmail.com> wrote:
>
> Hi,
>
> Some changes in v3:
> 1) Update the note of xlogwait.c to reflect the extending of its use
> for flush waiting and internal use for both flush and replay waiting.
> 2) Update the comment above logical_read_xlog_page which describes the
> prior-change behavior of read_local_xlog_page.

In an off-list discussion, Alexander pointed out potential issues with
the current single-heap design for replay and flush when promotion
occurs concurrently with WAIT FOR. The following is a simple example
illustrating the problem:

During promotion, there's a window where we can have mixed waiter
types in the same heap:

  T1: Process A calls read_local_xlog_page_guts on standby
  T2: RecoveryInProgress() = TRUE, adds to heap as replay waiter
  T3: Promotion begins
  T4: EndRecovery() calls WaitLSNWakeup(InvalidXLogRecPtr)
  T5: SharedRecoveryState = RECOVERY_STATE_DONE
  T6: Process B calls read_local_xlog_page_guts
  T7: RecoveryInProgress() = FALSE, adds to SAME heap as flush waiter

The problem is that replay LSNs and flush LSNs represent different
positions in the WAL stream. Having both types in the same heap can
lead to:
  - Incorrect wakeup logic (comparing incomparable LSNs)
  - Processes waiting forever
  - Wrong waiters being woken up

To avoid this problem, patch v4 is updated to utilize two separate
heaps for flush and replay like Alexander suggested earlier.  It also
introduces a new separate min LSN tracking field for flushing.

Best,
Xuneng

Hi,

On Sun, Sep 28, 2025 at 9:47 PM Xuneng Zhou <xunengzhou@gmail.com> wrote:
>
> Hi,
>
> On Thu, Aug 28, 2025 at 4:22 PM Xuneng Zhou <xunengzhou@gmail.com> wrote:
> >
> > Hi,
> >
> > Some changes in v3:
> > 1) Update the note of xlogwait.c to reflect the extending of its use
> > for flush waiting and internal use for both flush and replay waiting.
> > 2) Update the comment above logical_read_xlog_page which describes the
> > prior-change behavior of read_local_xlog_page.
>
> In an off-list discussion, Alexander pointed out potential issues with
> the current single-heap design for replay and flush when promotion
> occurs concurrently with WAIT FOR. The following is a simple example
> illustrating the problem:
>
> During promotion, there's a window where we can have mixed waiter
> types in the same heap:
>
>   T1: Process A calls read_local_xlog_page_guts on standby
>   T2: RecoveryInProgress() = TRUE, adds to heap as replay waiter
>   T3: Promotion begins
>   T4: EndRecovery() calls WaitLSNWakeup(InvalidXLogRecPtr)
>   T5: SharedRecoveryState = RECOVERY_STATE_DONE
>   T6: Process B calls read_local_xlog_page_guts
>   T7: RecoveryInProgress() = FALSE, adds to SAME heap as flush waiter
>
> The problem is that replay LSNs and flush LSNs represent different
> positions in the WAL stream. Having both types in the same heap can
> lead to:
>   - Incorrect wakeup logic (comparing incomparable LSNs)
>   - Processes waiting forever
>   - Wrong waiters being woken up
>
> To avoid this problem, patch v4 is updated to utilize two separate
> heaps for flush and replay like Alexander suggested earlier.  It also
> introduces a new separate min LSN tracking field for flushing.
>

v5-0002 separates the waitlsn_cmp() comparator function into two distinct
functions (waitlsn_replay_cmp and waitlsn_flush_cmp) for the replay
and flush heaps, respectively.

Best,
Xuneng

On Thu, Oct 02, 2025 at 11:06:14PM +0800, Xuneng Zhou wrote:
> v5-0002 separates the waitlsn_cmp() comparator function into two distinct
> functions (waitlsn_replay_cmp and waitlsn_flush_cmp) for the replay
> and flush heaps, respectively.

The primary goal that you want to achieve here is a replacement of the
wait/sleep logic of read_local_xlog_page_guts() with a condition
variable, and design a new facility to make the callback more
responsive on polls.  That's a fine idea in itself.  However I would
suggest to implement something that does not depend entirely on WAIT
FOR, which is how your patch is presented.  Instead of having your
patch depend on an entirely different feature, it seems to me that you
should try to extract from this other feature the basics that you are
looking for, and make them shared between the WAIT FOR patch and what
you are trying to achieve here.  You should not need something as
complex as what the other feature needs for a page read callback in
the backend.

At the end, I suspect that you will reuse a slight portion of it (or
perhaps nothing at all, actually, but I did not look at the full scope
of it).  You should try to present your patch so as it is in a
reviewable state, so as others would be able to read it and understand
it.  WAIT FOR is much more complex than what you want to do here
because it covers larger areas of the code base and needs to worry
about more cases.  So, you should implement things so as the basic
pieces you want to build on top of are simpler, not more complicated.
Simpler means easier to review and easier to catch problems, designed
in a way that depends on how you want to fix your problem, not
designed in a way that depends on how a completely different feature
deals with its own problems.
--
Michael

Hi Michael,

Thanks for your review.

On Fri, Oct 3, 2025 at 2:24 PM Michael Paquier <michael@paquier.xyz> wrote:
>
> On Thu, Oct 02, 2025 at 11:06:14PM +0800, Xuneng Zhou wrote:
> > v5-0002 separates the waitlsn_cmp() comparator function into two distinct
> > functions (waitlsn_replay_cmp and waitlsn_flush_cmp) for the replay
> > and flush heaps, respectively.
>
> The primary goal that you want to achieve here is a replacement of the
> wait/sleep logic of read_local_xlog_page_guts() with a condition
> variable, and design a new facility to make the callback more
> responsive on polls.  That's a fine idea in itself.  However I would
> suggest to implement something that does not depend entirely on WAIT
> FOR, which is how your patch is presented.  Instead of having your
> patch depend on an entirely different feature, it seems to me that you
> should try to extract from this other feature the basics that you are
> looking for, and make them shared between the WAIT FOR patch and what
> you are trying to achieve here.  You should not need something as
> complex as what the other feature needs for a page read callback in
> the backend.
>
> At the end, I suspect that you will reuse a slight portion of it (or
> perhaps nothing at all, actually, but I did not look at the full scope
> of it).  You should try to present your patch so as it is in a
> reviewable state, so as others would be able to read it and understand
> it.  WAIT FOR is much more complex than what you want to do here
> because it covers larger areas of the code base and needs to worry
> about more cases.  So, you should implement things so as the basic
> pieces you want to build on top of are simpler, not more complicated.
> Simpler means easier to review and easier to catch problems, designed
> in a way that depends on how you want to fix your problem, not
> designed in a way that depends on how a completely different feature
> deals with its own problems.

The core infrastructure shared by both this patch and the WAIT FOR
command patch is primarily in xlogwait.c, which provides the mechanism
for waiting until a given LSN is reached. Other parts of the code in
the WAIT FOR patch—covering the SQL command implementation,
documentation, and tests—is not relevant for the current patch. What
we need is only the infrastructure in xlogwait.c, on which we can
implement the optimization for read_local_xlog_page_guts.

Regarding complexity: the initial optimization idea was to introduce
condition-variable based waiting, as Heikki suggested in his comment:

/*
 * Loop waiting for xlog to be available if necessary
 *
 * TODO: The walsender has its own version of this function, which uses a
 * condition variable to wake up whenever WAL is flushed. We could use the
 * same infrastructure here, instead of the check/sleep/repeat style of
 * loop.
 */

I reviewed the relevant code in WalSndWakeup and WalSndWait. While
these mechanisms reduce polling overhead, they don’t prevent false
wakeups. Addressing that would likely require a request queue that
maps waiters to their target LSNs and issues targeted wakeups—a much
more complex design. Given that read_local_xlog_page_guts is not as
performance-sensitive as its equivalents, this added complexity may
not be justified. So I implemented the initial version of the
optimization like  WalSndWakeup and WalSndWait.

After this, I came across  the WAIT FOR patch in the mailing list and
noticed that the infrastructure in xlogwait.c aligns well with our
needs. Based on that, I built the current patch using this shared
infra.

If we prioritise simplicity and can tolerate occasional false wakeups,
then waiting in read_local_xlog_page_guts can be implemented in a much
simpler way than the current version. At the same time, the WAIT FOR
command seems to be a valuable feature in its own right, and both
patches can naturally share the same infrastructure. We could also
extract the infra and implement the current patch on it, then Wait for
could utilize it as well. Personally, I don’t have a strong preference
between the two approaches.

Best,
Xuneng

Hi,

On Fri, Oct 3, 2025 at 9:50 PM Xuneng Zhou <xunengzhou@gmail.com> wrote:
>
> Hi Michael,
>
> Thanks for your review.
>
> On Fri, Oct 3, 2025 at 2:24 PM Michael Paquier <michael@paquier.xyz> wrote:
> >
> > On Thu, Oct 02, 2025 at 11:06:14PM +0800, Xuneng Zhou wrote:
> > > v5-0002 separates the waitlsn_cmp() comparator function into two distinct
> > > functions (waitlsn_replay_cmp and waitlsn_flush_cmp) for the replay
> > > and flush heaps, respectively.
> >
> > The primary goal that you want to achieve here is a replacement of the
> > wait/sleep logic of read_local_xlog_page_guts() with a condition
> > variable, and design a new facility to make the callback more
> > responsive on polls.  That's a fine idea in itself.  However I would
> > suggest to implement something that does not depend entirely on WAIT
> > FOR, which is how your patch is presented.  Instead of having your
> > patch depend on an entirely different feature, it seems to me that you
> > should try to extract from this other feature the basics that you are
> > looking for, and make them shared between the WAIT FOR patch and what
> > you are trying to achieve here.  You should not need something as
> > complex as what the other feature needs for a page read callback in
> > the backend.
> >
> > At the end, I suspect that you will reuse a slight portion of it (or
> > perhaps nothing at all, actually, but I did not look at the full scope
> > of it).  You should try to present your patch so as it is in a
> > reviewable state, so as others would be able to read it and understand
> > it.  WAIT FOR is much more complex than what you want to do here
> > because it covers larger areas of the code base and needs to worry
> > about more cases.  So, you should implement things so as the basic
> > pieces you want to build on top of are simpler, not more complicated.
> > Simpler means easier to review and easier to catch problems, designed
> > in a way that depends on how you want to fix your problem, not
> > designed in a way that depends on how a completely different feature
> > deals with its own problems.
>
> The core infrastructure shared by both this patch and the WAIT FOR
> command patch is primarily in xlogwait.c, which provides the mechanism
> for waiting until a given LSN is reached. Other parts of the code in
> the WAIT FOR patch—covering the SQL command implementation,
> documentation, and tests—is not relevant for the current patch. What
> we need is only the infrastructure in xlogwait.c, on which we can
> implement the optimization for read_local_xlog_page_guts.
>
> Regarding complexity: the initial optimization idea was to introduce
> condition-variable based waiting, as Heikki suggested in his comment:
>
> /*
>  * Loop waiting for xlog to be available if necessary
>  *
>  * TODO: The walsender has its own version of this function, which uses a
>  * condition variable to wake up whenever WAL is flushed. We could use the
>  * same infrastructure here, instead of the check/sleep/repeat style of
>  * loop.
>  */
>
> I reviewed the relevant code in WalSndWakeup and WalSndWait. While
> these mechanisms reduce polling overhead, they don’t prevent false
> wakeups. Addressing that would likely require a request queue that
> maps waiters to their target LSNs and issues targeted wakeups—a much
> more complex design. Given that read_local_xlog_page_guts is not as
> performance-sensitive as its equivalents, this added complexity may
> not be justified. So I implemented the initial version of the
> optimization like  WalSndWakeup and WalSndWait.
>
> After this, I came across  the WAIT FOR patch in the mailing list and
> noticed that the infrastructure in xlogwait.c aligns well with our
> needs. Based on that, I built the current patch using this shared
> infra.
>
> If we prioritise simplicity and can tolerate occasional false wakeups,
> then waiting in read_local_xlog_page_guts can be implemented in a much
> simpler way than the current version. At the same time, the WAIT FOR
> command seems to be a valuable feature in its own right, and both
> patches can naturally share the same infrastructure. We could also
> extract the infra and implement the current patch on it, then Wait for
> could utilize it as well. Personally, I don’t have a strong preference
> between the two approaches.
>

Another potential use for this infra could be static XLogRecPtr
WalSndWaitForWal(XLogRecPtr loc), I'm planning to hack a version as
well.

Best,
Xuneng

Hi,

On Sat, Oct 4, 2025 at 10:25 AM Xuneng Zhou <xunengzhou@gmail.com> wrote:
>
> Hi,
>
> On Fri, Oct 3, 2025 at 9:50 PM Xuneng Zhou <xunengzhou@gmail.com> wrote:
> >
> > Hi Michael,
> >
> > Thanks for your review.
> >
> > On Fri, Oct 3, 2025 at 2:24 PM Michael Paquier <michael@paquier.xyz> wrote:
> > >
> > > On Thu, Oct 02, 2025 at 11:06:14PM +0800, Xuneng Zhou wrote:
> > > > v5-0002 separates the waitlsn_cmp() comparator function into two distinct
> > > > functions (waitlsn_replay_cmp and waitlsn_flush_cmp) for the replay
> > > > and flush heaps, respectively.
> > >
> > > The primary goal that you want to achieve here is a replacement of the
> > > wait/sleep logic of read_local_xlog_page_guts() with a condition
> > > variable, and design a new facility to make the callback more
> > > responsive on polls.  That's a fine idea in itself.  However I would
> > > suggest to implement something that does not depend entirely on WAIT
> > > FOR, which is how your patch is presented.  Instead of having your
> > > patch depend on an entirely different feature, it seems to me that you
> > > should try to extract from this other feature the basics that you are
> > > looking for, and make them shared between the WAIT FOR patch and what
> > > you are trying to achieve here.  You should not need something as
> > > complex as what the other feature needs for a page read callback in
> > > the backend.
> > >
> > > At the end, I suspect that you will reuse a slight portion of it (or
> > > perhaps nothing at all, actually, but I did not look at the full scope
> > > of it).  You should try to present your patch so as it is in a
> > > reviewable state, so as others would be able to read it and understand
> > > it.  WAIT FOR is much more complex than what you want to do here
> > > because it covers larger areas of the code base and needs to worry
> > > about more cases.  So, you should implement things so as the basic
> > > pieces you want to build on top of are simpler, not more complicated.
> > > Simpler means easier to review and easier to catch problems, designed
> > > in a way that depends on how you want to fix your problem, not
> > > designed in a way that depends on how a completely different feature
> > > deals with its own problems.
> >
> > The core infrastructure shared by both this patch and the WAIT FOR
> > command patch is primarily in xlogwait.c, which provides the mechanism
> > for waiting until a given LSN is reached. Other parts of the code in
> > the WAIT FOR patch—covering the SQL command implementation,
> > documentation, and tests—is not relevant for the current patch. What
> > we need is only the infrastructure in xlogwait.c, on which we can
> > implement the optimization for read_local_xlog_page_guts.
> >
> > Regarding complexity: the initial optimization idea was to introduce
> > condition-variable based waiting, as Heikki suggested in his comment:
> >
> > /*
> >  * Loop waiting for xlog to be available if necessary
> >  *
> >  * TODO: The walsender has its own version of this function, which uses a
> >  * condition variable to wake up whenever WAL is flushed. We could use the
> >  * same infrastructure here, instead of the check/sleep/repeat style of
> >  * loop.
> >  */
> >
> > I reviewed the relevant code in WalSndWakeup and WalSndWait. While
> > these mechanisms reduce polling overhead, they don’t prevent false
> > wakeups. Addressing that would likely require a request queue that
> > maps waiters to their target LSNs and issues targeted wakeups—a much
> > more complex design. Given that read_local_xlog_page_guts is not as
> > performance-sensitive as its equivalents, this added complexity may
> > not be justified. So I implemented the initial version of the
> > optimization like  WalSndWakeup and WalSndWait.
> >
> > After this, I came across  the WAIT FOR patch in the mailing list and
> > noticed that the infrastructure in xlogwait.c aligns well with our
> > needs. Based on that, I built the current patch using this shared
> > infra.
> >
> > If we prioritise simplicity and can tolerate occasional false wakeups,
> > then waiting in read_local_xlog_page_guts can be implemented in a much
> > simpler way than the current version. At the same time, the WAIT FOR
> > command seems to be a valuable feature in its own right, and both
> > patches can naturally share the same infrastructure. We could also
> > extract the infra and implement the current patch on it, then Wait for
> > could utilize it as well. Personally, I don’t have a strong preference
> > between the two approaches.
> >
>
> Another potential use for this infra could be static XLogRecPtr
> WalSndWaitForWal(XLogRecPtr loc), I'm planning to hack a version as
> well.
>
> Best,
> Xuneng

v6 refactors and extends the infrastructure from the WAIT FOR command
patch, applying it to read_local_xlog_page_guts. I'm also thinking of
creating a standalone patch/commit for the extended
infra in xlogwait, so it can be reused in different threads.

Best,
Xuneng