fix(tsm1): fix condition check for optimization of array cursor #26691
+172
−6
Conversation
This file contains hidden or 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
The array ascending cursors do not check all the required conditions before doing an optimization to copy a full 1000 points of data from the tsm read buffer to the cursors response buffer. This can result in points from tsm files appearing in cursor output multiple times. The conditions the bug appears are rare as it requires cache data that is older (or overlapping) with tsm data. The data must be in the same shard and must match for the same key. * fixes #26690
philjb
commented
Aug 12, 2025
| if c.tsm.pos < len(tvals.Timestamps) { | ||
| if pos == 0 && len(c.res.Timestamps) >= len(tvals.Timestamps) { | ||
| if pos == 0 && c.tsm.pos == 0 && len(c.res.Timestamps) >= len(tvals.Timestamps) { | ||
| // optimization: all points can be served from TSM data because |
There was a problem hiding this comment.
I'm not sure how much of an optimization this is anymore. I would expect (maybe) that go doesn't reslice (create a new slice header) if copy(dst[0:], src[0:]) is used, but i haven't looked yet.
There was a problem hiding this comment.
copy calls into runtime.memmove. No slice headers are made but setup is done to get the arguments to memmove to point to the correct place in each backing array. memmove itself handles overlap per the spec.
Looks like this optimization avoids some setup to memmove at the code of the extra conditionals. Probably worthwhile to remove the "optimization" to have less code to debug and maintain. A microbench mark might show some speed differences.
This commit adds a test that verifies the duplicate data bug fix. It will fail without the previous commit. See the test for details; in short it sets up the right cache and tsm data to expose the bug which was occuring when the cache had older data than the tsm files and then the cache was exhausted while there were still more tsm values to include in more Next() calls.
philjb
commented
Aug 14, 2025
| kc := fs.KeyCursor(context.Background(), []byte("measurement,field=value#!~#value"), 0, true) | ||
| defer kc.Close() | ||
|
|
||
| cursor := newFloatArrayAscendingCursor() |
There was a problem hiding this comment.
the test only looks at float array cursors but all types are impacted. They are made from a template so i made the assumption testing one type also covers the other datatypes.
philjb
commented
Aug 14, 2025
| // and incorrectly copies TSM data including the points from the previous Next() call. Additionally, cache data | ||
| // needs to be older than most of the TSM data. There needs to be enough tsm data younger than the cache data to completely | ||
| // fill the remaining space in the response buffer. | ||
| func TestAscendingCursorDuplicateDataBug(t *testing.T) { |
There was a problem hiding this comment.
AI helped me make this test more robust; but it also made it rather long with the setup and precondition verifications.
davidby-influx
added
kind/bug
area/flux
davidby-influx
requested changes
Aug 19, 2025
There was a problem hiding this comment.
I'd recommend using modern test helpers, like require.NoError instead of Fatalf
The assert package is used for post conditions and the require package is used for pre-conditions in the test. The library provides clearer intent of conditions and better formatting of information (message and data) in the situation an assertion fails. It also makes it easier to read by reducing ifs but some lines are longer. This commit addresses PR requests.
This was referenced Aug 20, 2025
devanbenz
pushed a commit
that referenced
this pull request
Sep 29, 2025
* fix(tsm1): fix condition check for optimization of array cursor The array ascending cursors do not check all the required conditions before doing an optimization to copy a full 1000 points of data from the tsm read buffer to the cursors response buffer. This can result in points from tsm files appearing in cursor output multiple times. The conditions the bug appears are rare as it requires cache data that is older (or overlapping) with tsm data. The data must be in the same shard and must match for the same key. * fixes #26690 * chore(tsm1): add test that covers array cursor bug This commit adds a test that verifies the duplicate data bug fix. It will fail without the previous commit. See the test for details; in short it sets up the right cache and tsm data to expose the bug which was occuring when the cache had older data than the tsm files and then the cache was exhausted while there were still more tsm values to include in more Next() calls. * chore(test): switch to using the testify library in the new test The assert package is used for post conditions and the require package is used for pre-conditions in the test. The library provides clearer intent of conditions and better formatting of information (message and data) in the situation an assertion fails. It also makes it easier to read by reducing ifs but some lines are longer. This commit addresses PR requests. (cherry picked from commit 9d1e79d)
devanbenz
added a commit
that referenced
this pull request
Sep 30, 2025
…) (#26861) Co-authored-by: Phil Bracikowski <[email protected]> fixes #26690
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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.
The array ascending cursors do not check all the required conditions before doing an optimization to copy a full 1000 points of data from the tsm read buffer to the cursors response buffer. This can result in points from tsm files appearing in cursor output multiple times.
The conditions the bug appears are rare as it requires cache data that is older (or overlapping) with tsm data. The data must be in the same shard and must match for the same key.