-
Notifications
You must be signed in to change notification settings - Fork 3.8k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
sql: DROP INDEX failure with a sandwiched DROP REGION results in irrecoverable state #63974
Comments
Hi @otan, please add branch-* labels to identify which branch(es) this release-blocker affects. 🦉 Hoot! I am a Blathers, a bot for CockroachDB. My owner is otan. |
i've tried this alternate strategy in #63976, but hitting this on DROP REGION sandwiching:
i'm still convinced we should block ADD/DROP REGION when a CREATE/DROP INDEX or ALTER TABLE LOCALITY is underway. |
i've also amended #63968 to exclude the case where this might break. |
I agree here, I like the solution proposed in #63976 more as well. That'll also serve as a solution for the case I've been thinking about:
|
Yep, so the #63976 solution seems problematic because I'm pretty the blocked thread has no knowledge of the new enum as it used an older version of it. |
#63976 works! made a silly boo boo |
Consider this test, where:
In this case, we have bad state - the index does not match the regions.
Originally I tried patching this in #63968, but it's broken in the case that if we transition from RBR to Global instead of drop index, the DROP REGION would also "accidentally" partition the global indexes -- then again, we already do that today and i think that's a bug that will fall our when arul tests it.
My suggested solution would be to repartition again during CREATE INDEX/DROP INDEX when the operation is complete. This is probably similar to arul's proposed work. Can take a stab at this tomorrow. Warrants a GA-blocker.
The text was updated successfully, but these errors were encountered: