Обсуждение: pg_rewind race condition just after promotion

There's a race condition between the checkpoint at promotion and 
pg_rewind. When a server is promoted, the startup process writes an 
end-of-recovery checkpoint that includes the new TLI, and the server is 
immediate opened for business. The startup process requests the 
checkpointer process to perform a checkpoint, but it can take a few 
seconds or more to complete. If you run pg_rewind, using the just 
promoted server as the source, pg_rewind will think that the server is 
still on the old timeline, because it only looks at TLI in the control 
file's copy of the checkpoint record. That's not updated until the 
checkpoint is finished.

This isn't a new issue. Stephen Frost first reported it back 2015 [1]. 
Back then, it was deemed just a small annoyance, and we just worked 
around it in the tests by issuing a checkpoint command after promotion, 
to wait for the checkpoint to finish. I just ran into it again today, 
with the new pg_rewind test, and silenced it in the similar way.

I think we should fix this properly. I'm not sure if it can lead to a 
broken cluster, but at least it can cause pg_rewind to fail 
unnecessarily and in a user-unfriendly way. But this is actually pretty 
simple to fix. pg_rewind looks at the control file to find out the 
timeline the server is on. When promotion happens, the startup process 
updates minRecoveryPoint and minRecoveryPointTLI fields in the control 
file. We just need to read it from there. Patch attached.

I think we should also backpatch this. Back in 2015, we decided that we 
can live with this, but it's always been a bit bogus, and seems simple 
enough to fix.

Thoughts?

[1] 
https://www.postgresql.org/message-id/20150428180253.GU30322%40tamriel.snowman.net

- Heikki

At Mon, 7 Dec 2020 20:13:25 +0200, Heikki Linnakangas <hlinnaka@iki.fi> wrote in 
> There's a race condition between the checkpoint at promotion and
> pg_rewind. When a server is promoted, the startup process writes an
> end-of-recovery checkpoint that includes the new TLI, and the server
> is immediate opened for business. The startup process requests the
> checkpointer process to perform a checkpoint, but it can take a few
> seconds or more to complete. If you run pg_rewind, using the just
> promoted server as the source, pg_rewind will think that the server is
> still on the old timeline, because it only looks at TLI in the control
> file's copy of the checkpoint record. That's not updated until the
> checkpoint is finished.
> 
> This isn't a new issue. Stephen Frost first reported it back 2015
> [1]. Back then, it was deemed just a small annoyance, and we just
> worked around it in the tests by issuing a checkpoint command after
> promotion, to wait for the checkpoint to finish. I just ran into it
> again today, with the new pg_rewind test, and silenced it in the
> similar way.

I (or we) faced that and avoided that by checking for history file on
the primary.

> I think we should fix this properly. I'm not sure if it can lead to a
> broken cluster, but at least it can cause pg_rewind to fail
> unnecessarily and in a user-unfriendly way. But this is actually
> pretty simple to fix. pg_rewind looks at the control file to find out
> the timeline the server is on. When promotion happens, the startup
> process updates minRecoveryPoint and minRecoveryPointTLI fields in the
> control file. We just need to read it from there. Patch attached.

Looks fine to me.  A bit concerned about making sourceHistory
needlessly file-local but on the other hand unifying sourceHistory and
targetHistory looks better.

For the test part, that change doesn't necessariry catch the failure
of the current version, but I *believe* the prevous code is the result
of an actual failure in the past so the test probablistically (or
dependently on platforms?) hits the failure if it happned.

> I think we should also backpatch this. Back in 2015, we decided that
> we can live with this, but it's always been a bit bogus, and seems
> simple enough to fix.

I don't think this changes any successful behavior and it just saves
the failure case so +1 for back-patching.

> Thoughts?
> 
> [1]
> https://www.postgresql.org/message-id/20150428180253.GU30322%40tamriel.snowman.net

regards.

-- 
Kyotaro Horiguchi
NTT Open Source Software Center

On 08/12/2020 06:45, Kyotaro Horiguchi wrote:
> At Mon, 7 Dec 2020 20:13:25 +0200, Heikki Linnakangas <hlinnaka@iki.fi> wrote in
>> I think we should fix this properly. I'm not sure if it can lead to a
>> broken cluster, but at least it can cause pg_rewind to fail
>> unnecessarily and in a user-unfriendly way. But this is actually
>> pretty simple to fix. pg_rewind looks at the control file to find out
>> the timeline the server is on. When promotion happens, the startup
>> process updates minRecoveryPoint and minRecoveryPointTLI fields in the
>> control file. We just need to read it from there. Patch attached.
> 
> Looks fine to me.  A bit concerned about making sourceHistory
> needlessly file-local but on the other hand unifying sourceHistory and
> targetHistory looks better.

Looking closer, findCommonAncestorTimeline() was freeing sourceHistory, 
which was pretty horrible when it's a file-local variable. I changed it 
so that both the source and target histories are passed to 
findCommonAncestorTimeline() as arguments. That seems more clear.

> For the test part, that change doesn't necessariry catch the failure
> of the current version, but I *believe* the prevous code is the result
> of an actual failure in the past so the test probablistically (or
> dependently on platforms?) hits the failure if it happned.

Right. I think the current test coverage is good enough. We've been 
bitten by this a few times by now, when we've forgotten to add the 
manual checkpoint commands to new tests, and the buildfarm has caught it 
pretty quickly.

>> I think we should also backpatch this. Back in 2015, we decided that
>> we can live with this, but it's always been a bit bogus, and seems
>> simple enough to fix.
> 
> I don't think this changes any successful behavior and it just saves
> the failure case so +1 for back-patching.

Thanks for the review! New patch version attached.

- Heikki

The following review has been posted through the commitfest application:
make installcheck-world:  tested, passed
Implements feature:       tested, passed
Spec compliant:           tested, passed
Documentation:            tested, passed

The v3 patch LGTM. I wonder if we should explicitly say in pg_rewind tests that
they _don't_ have to call `checkpoint`, or otherwise, we will lose the test
coverage for this scenario. But I don't have a strong opinion on this one.

The new status of this patch is: Ready for Committer

> On 14 Jul 2021, at 14:03, Aleksander Alekseev <aleksander@timescale.com> wrote:
>
> The following review has been posted through the commitfest application:
> make installcheck-world:  tested, passed
> Implements feature:       tested, passed
> Spec compliant:           tested, passed
> Documentation:            tested, passed
>
> The v3 patch LGTM. I wonder if we should explicitly say in pg_rewind tests that
> they _don't_ have to call `checkpoint`, or otherwise, we will lose the test
> coverage for this scenario. But I don't have a strong opinion on this one.
>
> The new status of this patch is: Ready for Committer

Heikki, do you have plans to address this patch during this CF?

--
Daniel Gustafsson        https://vmware.com/

2021年11月9日(火) 20:31 Daniel Gustafsson <daniel@yesql.se>:
>
> > On 14 Jul 2021, at 14:03, Aleksander Alekseev <aleksander@timescale.com> wrote:
> >
> > The following review has been posted through the commitfest application:
> > make installcheck-world:  tested, passed
> > Implements feature:       tested, passed
> > Spec compliant:           tested, passed
> > Documentation:            tested, passed
> >
> > The v3 patch LGTM. I wonder if we should explicitly say in pg_rewind tests that
> > they _don't_ have to call `checkpoint`, or otherwise, we will lose the test
> > coverage for this scenario. But I don't have a strong opinion on this one.
> >
> > The new status of this patch is: Ready for Committer
>
> Heikki, do you have plans to address this patch during this CF?

Friendly reminder ping one year on; I haven't looked at this patch in
detail but going by the thread contents it seems it should be marked
"Ready for Committer"? Moved to the next CF anyway.

Regards

Ian Barwick

On 11/12/2022 02:01, Ian Lawrence Barwick wrote:
> 2021年11月9日(火) 20:31 Daniel Gustafsson <daniel@yesql.se>:
>>
>>> On 14 Jul 2021, at 14:03, Aleksander Alekseev <aleksander@timescale.com> wrote:
>>>
>>> The following review has been posted through the commitfest application:
>>> make installcheck-world:  tested, passed
>>> Implements feature:       tested, passed
>>> Spec compliant:           tested, passed
>>> Documentation:            tested, passed
>>>
>>> The v3 patch LGTM. I wonder if we should explicitly say in pg_rewind tests that
>>> they _don't_ have to call `checkpoint`, or otherwise, we will lose the test
>>> coverage for this scenario. But I don't have a strong opinion on this one.
>>>
>>> The new status of this patch is: Ready for Committer
>>
>> Heikki, do you have plans to address this patch during this CF?
> 
> Friendly reminder ping one year on; I haven't looked at this patch in
> detail but going by the thread contents it seems it should be marked
> "Ready for Committer"? Moved to the next CF anyway.

Here's an updated version of the patch.

I renamed the arguments to findCommonAncestorTimeline() so that the 
'targetHistory' argument doesn't shadow the global 'targetHistory' 
variable. No other changes, and this still looks good to me, so I'll 
wait for the cfbot to run on this and commit in the next few days.

- Heikki

On 22/02/2023 16:00, Heikki Linnakangas wrote:
> On 11/12/2022 02:01, Ian Lawrence Barwick wrote:
>> 2021年11月9日(火) 20:31 Daniel Gustafsson <daniel@yesql.se>:
>>>
>>>> On 14 Jul 2021, at 14:03, Aleksander Alekseev <aleksander@timescale.com> wrote:
>>>>
>>>> The following review has been posted through the commitfest application:
>>>> make installcheck-world:  tested, passed
>>>> Implements feature:       tested, passed
>>>> Spec compliant:           tested, passed
>>>> Documentation:            tested, passed
>>>>
>>>> The v3 patch LGTM. I wonder if we should explicitly say in pg_rewind tests that
>>>> they _don't_ have to call `checkpoint`, or otherwise, we will lose the test
>>>> coverage for this scenario. But I don't have a strong opinion on this one.
>>>>
>>>> The new status of this patch is: Ready for Committer
>>>
>>> Heikki, do you have plans to address this patch during this CF?
>>
>> Friendly reminder ping one year on; I haven't looked at this patch in
>> detail but going by the thread contents it seems it should be marked
>> "Ready for Committer"? Moved to the next CF anyway.
> 
> Here's an updated version of the patch.
> 
> I renamed the arguments to findCommonAncestorTimeline() so that the
> 'targetHistory' argument doesn't shadow the global 'targetHistory'
> variable. No other changes, and this still looks good to me, so I'll
> wait for the cfbot to run on this and commit in the next few days.

Pushed. I decided not to backpatch this, after all. We haven't really 
been treating this as a bug so far, and the patch didn't apply cleanly 
to v13 and before.

- Heikki