changefeedccl: add ignore_disable_changefeed_replication option

[andyyang890]

Fixes #119891

Release note (enterprise change): A new boolean changefeed option
named ignore_disable_changefeed_replication has been added.
When set, the changefeed will not filter events even if CDC
filtering is configured, which is currently available via the
disable_changefeed_replication session variable,
sql.ttl.changefeed_replication.disabled cluster setting, and the
ttl_disable_changefeed_replication table storage parameter.

[cockroach-teamcity]

This change is 

[rharding6373]

Nice work! I have fairly minor comments.

Reviewed 3 of 4 files at r1.
Reviewable status: complete! 0 of 0 LGTMs obtained (waiting on @andyyang890, @herkolategan, @jayshrivastava, @miraradeva, and @renatolabs)

---

-- commits line 5 at r1:
I think the description of changefeedReplicationDisabled should be expanded to say that changefeeds with this option will not filter TTL delete events.

---

pkg/cmd/roachtest/tests/cdc_filtering.go line 271 at r1 (raw file):

		// Produce a canonical format that we can assert on. The format is of the
		// form: id@exp_at|<deleted> (before: id@exp_at).
		eventToString := func(before *state, after *state) string {

Can this function be moved into a shared helper function? It seems like there's duplicate logic here.

[andyyang890]

Reviewable status: complete! 0 of 0 LGTMs obtained (waiting on @herkolategan, @jayshrivastava, @miraradeva, @renatolabs, and @rharding6373)

---

-- commits line 5 at r1:

Previously, rharding6373 (Rachael Harding) wrote…

I think the description of changefeedReplicationDisabled should be expanded to say that changefeeds with this option will not filter TTL delete events.

Done.

---

pkg/cmd/roachtest/tests/cdc_filtering.go line 271 at r1 (raw file):

Previously, rharding6373 (Rachael Harding) wrote…

Can this function be moved into a shared helper function? It seems like there's duplicate logic here.

Good call. I've refactored the code into a makeEventString function.

[rharding6373]

Looks great!

Reviewable status: complete! 1 of 0 LGTMs obtained (waiting on @andyyang890, @herkolategan, @jayshrivastava, @miraradeva, and @renatolabs)

---

pkg/cmd/roachtest/tests/cdc_filtering.go line 271 at r1 (raw file):

Previously, andyyang890 (Andy Yang) wrote…

Good call. I've refactored the code into a makeEventString function.

This was an even cleaner refactor than what I considered. Very nice! Thanks for doing that.

---

pkg/sql/ttl/ttlbase/ttl_helpers.go line 78 at r2 (raw file):

		"sql.ttl.changefeed_replication.disabled",
		"if true, deletes issued by TTL will not be replicated via changefeeds "+
			"(except changefeeds that have the ignore_disable_changefeed_replication option set)",

nit: add something like ", which will always replicate TTL deletes" to the text in parens for clarity.

[andyyang890]

Reviewable status: complete! 0 of 0 LGTMs obtained (and 1 stale) (waiting on @herkolategan, @jayshrivastava, @miraradeva, @renatolabs, and @rharding6373)

---

pkg/cmd/roachtest/tests/cdc_filtering.go line 271 at r1 (raw file):

Previously, rharding6373 (Rachael Harding) wrote…

This was an even cleaner refactor than what I considered. Very nice! Thanks for doing that.

Thanks!

---

pkg/sql/ttl/ttlbase/ttl_helpers.go line 78 at r2 (raw file):

Previously, rharding6373 (Rachael Harding) wrote…

nit: add something like ", which will always replicate TTL deletes" to the text in parens for clarity.

Updated, hopefully it's more clear now.

[andyyang890]

TFTR!

bors r=rharding6373

[andyyang890]

Seems like it didn't get added to the bors queue

bors r=rharding6373

[craig]

Build succeeded:

• Bazel Essential CI (Cockroach)
• macos_amd64_build
• macos_arm64_build
• windows_build