db: double check file reference counts when loading file #2901
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
This change is |
Double check the file reference counts before attempting to find/create a table cache node for a file. Once a file's reference count falls to zero, the file becomes obsolete and may be deleted at any moment. Today if we have a race, break this invariant and attempt to load a file with a nonpositive reference count, it's relatively unlikely that it manifests as an error. Typically tables remain open in the table cache, allowing the table cache to serve the request even if the file is no longer linked into the data directory. Additionally, even if it's not in the table cache presently, deletion of obsolete files may be delayed due to deletion pacing, hiding the race. This commit preemptively asserts on the file reference counts. I opted for not restricting this invariant check to invariants builds because it's cheap relative to a table cache lookup, and it's a particularly tricky form of corruption to debug otherwise. Informs cockroachdb/cockroach#110645.
jbowens
changed the title
RaduBerinde
approved these changes
Sep 15, 2023
There was a problem hiding this comment.
Reviewable status: 0 of 3 files reviewed, all discussions resolved
|
TFTR! |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Double check the file reference counts before attempting to find/create a table
cache node for a file. Once a file's reference count falls to zero, the file
becomes obsolete and may be deleted at any moment.
Today if we have a race, break this invariant and attempt to load a file with a
nonpositive reference count, it's relatively unlikely that it manifests as an
error. Typically tables remain open in the table cache, allowing the table
cache to serve the request even if the file is no longer linked into the data
directory. Additionally, even if it's not in the table cache presently,
deletion of obsolete files may be delayed due to deletion pacing, hiding the
race.
This commit preemptively asserts on the file reference counts. I opted for not
restricting this invariant check to invariants builds because it's cheap
relative to a table cache lookup, and it's a particularly tricky form of
corruption to debug otherwise.
Informs cockroachdb/cockroach#110645.