storage: add cluster setting to expire sticky bits

[jeffrey-xiao]

Since we represent sticky bits as HLCs, we can provide a setting to have
them expire after a certain duration. After this duration, these
manually split ranges will be eligible for automatic merging again.

This setting is useful for cases when a user might want to split the
data as it's being loaded into tables, but after the loading is
complete, the user would want the manual splits be merged as determined
by the merge queue.

Resolves: #37696

Release note (general change): Add kv.range_merge.manual_split_ttl setting

[cockroach-teamcity]

This change is 

[ajwerner]

Nice

Reviewable status: complete! 0 of 0 LGTMs obtained (waiting on @ajwerner, @jeffrey-xiao, and @nvanbenschoten)

---

pkg/storage/merge_queue.go, line 65 at r1 (raw file):

// a LHS and a RHS, then the merge queue will not consider to merge the RHS
// with the range to its left until ManualSplitTTL time has passed.
var ManualSplitTTL = settings.RegisterDurationSetting(

RegisterNonNegativeDurationSetting is probably a better choice.

---

pkg/storage/merge_queue.go, line 67 at r1 (raw file):

var ManualSplitTTL = settings.RegisterDurationSetting(
	"kv.range_merge.manual_split.ttl",
	"if nonzero, manual splits newer than this duration will not be considered for automatic merging.",

The linter will complain about that ., just drop it

---

pkg/storage/merge_queue.go, line 307 at r1 (raw file):

	if rhsDesc.StickyBit != nil {
		manualSplitTTL := ManualSplitTTL.Get(&mq.store.ClusterSettings().SV)
		manualSplitExpired := manualSplitTTL != 0 && rhsDesc.StickyBit.GoTime().Add(manualSplitTTL).Before(mq.store.Clock().Now().GoTime())

This doesn't use the logical portion of the HLC, no reason to pay for it. How about mq.store.Clock().PhysicalTime()?

[jeffrey-xiao]

Reviewable status: complete! 0 of 0 LGTMs obtained (waiting on @ajwerner and @nvanbenschoten)

---

pkg/storage/merge_queue.go, line 65 at r1 (raw file):

Previously, ajwerner wrote…

RegisterNonNegativeDurationSetting is probably a better choice.

Done.

---

pkg/storage/merge_queue.go, line 67 at r1 (raw file):

Previously, ajwerner wrote…

The linter will complain about that ., just drop it

Done.

---

pkg/storage/merge_queue.go, line 307 at r1 (raw file):

Previously, ajwerner wrote…

This doesn't use the logical portion of the HLC, no reason to pay for it. How about mq.store.Clock().PhysicalTime()?

Done.

[nvanbenschoten]

Reviewed 3 of 3 files at r1.
Reviewable status: complete! 0 of 0 LGTMs obtained (waiting on @ajwerner, @jeffrey-xiao, and @nvanbenschoten)

---

pkg/storage/client_merge_test.go, line 3288 at r1 (raw file):

	storagebase.MergeQueueEnabled.Override(sv, true)
	storage.MergeQueueInterval.Override(sv, 0) // process greedily
	storage.ManualSplitTTL.Override(sv, time.Hour)

pull time.Hour into a constant so you can use it here and below.

---

pkg/storage/client_merge_test.go, line 3474 at r1 (raw file):

	})

	t.Run("merge split ttl", func(t *testing.T) {

"sticky bit ttl"?

---

pkg/storage/merge_queue.go, line 62 at r1 (raw file):

// ManualSplitTTL is the amount of time that the merge queue will not consider
// a manually split range for merging if it is nonzero. If a range is split in

And what if it is zero? Are all manual splits considered for merges or are none?

---

pkg/storage/merge_queue.go, line 67 at r1 (raw file):

var ManualSplitTTL = settings.RegisterDurationSetting(
	"kv.range_merge.manual_split.ttl",
	"if nonzero, manual splits newer than this duration will not be considered for automatic merging.",

I think it would be clearer if we inverted this phrasing. Something like "if nonzero, manual splits older than this duration will be considered for automatic Range merging."

[nvanbenschoten]

Reviewed 3 of 3 files at r2.
Reviewable status: complete! 1 of 0 LGTMs obtained (waiting on @ajwerner)

[ajwerner]

Reviewed 3 of 3 files at r3.
Reviewable status: complete! 1 of 0 LGTMs obtained (and 1 stale) (waiting on @ajwerner)

[jeffrey-xiao]

bors r+

[craig]

Build succeeded

• GitHub CI (Cockroach)