-
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
release-24.2: storage: fix comparison of suffixes #129605
release-24.2: storage: fix comparison of suffixes #129605
Conversation
Thanks for opening a backport. Please check the backport criteria before merging:
If your backport adds new functionality, please ensure that the following additional criteria are satisfied:
Also, please add a brief release justification to the body of your PR to justify this |
Note: once this lands, if we can, we'll want to cherry-pick this onto |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Reviewed 2 of 2 files at r1, all commit messages.
Reviewable status: complete! 1 of 0 LGTMs obtained (waiting on @nicktrav, @RaduBerinde, and @tbg)
pkg/storage/pebble.go
line 499 at r1 (raw file):
panic(errors.AssertionFailedf("malformed suffix: %x", a)) } // Strip off sentinel byte.
nit: "sentinel byte" is confusing. the sentinel byte is the 0x00
byte at the end of prefix, whereas this is the version-length byte we're trimming.
This is a partial backport of cockroachdb#128043. Release justification: this fixes an issue hit in a real cluster where a range key is never cleaned up. The EngineComparer has a bug when comparing bare suffixes - we call the normalization function on a slice without the sentinel byte; that function's logic is based on length so it will not function properly. We recently observed that this can cause problems when a RangeKeySet has a suffix with the (now obsolete) synthetic bit set, and the corresponding `RangeKeyUnset` is issued by a newer version and doesn't have the bit set. These suffixes are supposed to equal each other but this is not currently the case. We fix the comparator to account for this case correctly and add a more comprehensive test. On the old code, the updated test failed with: ``` Compare(0000000000000002000000010d, 000000000000000200000001010e) = 1, expected 0 ``` Fixes: None Release note: None
b30849f
to
5daa9f6
Compare
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Reviewable status: complete! 0 of 0 LGTMs obtained (and 1 stale) (waiting on @jbowens, @nicktrav, and @tbg)
pkg/storage/pebble.go
line 499 at r1 (raw file):
Previously, jbowens (Jackson Owens) wrote…
nit: "sentinel byte" is confusing. the sentinel byte is the
0x00
byte at the end of prefix, whereas this is the version-length byte we're trimming.
Yeah, I keep confusing the terminology, fixed.
Based on the specified backports for this PR, I applied new labels to the following linked issue(s). Please adjust the labels as needed to match the branches actually affected by the issue(s), including adding any known older branches. Issue #129592: branch-release-23.1, branch-release-23.2, branch-release-24.2.1-rc. 🦉 Hoot! I am a Blathers, a bot for CockroachDB. My owner is dev-inf. |
Encountered an error creating backports. Some common things that can go wrong:
You might need to create your backport manually using the backport tool. error creating merge commit from 5daa9f6 to blathers/backport-release-23.1-129605: POST https://api.github.com/repos/cockroachdb/cockroach/merges: 409 Merge conflict [] you may need to manually resolve merge conflicts with the backport tool. Backport to branch 23.1.x failed. See errors above. 🦉 Hoot! I am a Blathers, a bot for CockroachDB. My owner is dev-inf. |
This is a partial backport of #128043. CC @cockroachdb/release
Release justification: this fixes an issue hit in a real cluster where
a range key is never cleaned up.
The EngineComparer has a bug when comparing bare suffixes - we call
the normalization function on a slice without the sentinel byte; that
function's logic is based on length so it will not function properly.
We recently observed that this can cause problems when a
RangeKeySet has a suffix with the (now obsolete) synthetic bit set,
and the corresponding
RangeKeyUnset
is issued by a newer version anddoesn't have the bit set. These suffixes are supposed to equal each
other but this is not currently the case.
We fix the comparator to account for this case correctly and add a
more comprehensive test. On the old code, the updated test failed
with:
Fixes: #129592
Release note: None