[NDMII-1268] Add unit tests for flowAccumulator.detectHashCollision #29504
Conversation
|
|
Test changes on VMUse this command from test-infra-definitions to manually test this PR changes on a VM: inv create-vm --pipeline-id=45122137 --os-family=ubuntuNote: This applies to commit 4a96443 |
| // Then | ||
| assert.Equal(t, uint64(0), acc.hashCollisionFlowCount.Load()) | ||
|
|
||
| // test hash collision (same flow object) does not increment flow count |
There was a problem hiding this comment.
heya! to clarify my understanding further for the test cases:
- this first case and the second one are "hash collisions" but in this case valid ones given they are essentially the same "flow" and thus the collision is valid because they have the same hash (as expected, since they have the same flow context)
- the last case has different fields in the flow context but end up at the same hash and thus a "collision" that we're not expecting, thus the increment in the metric / log emitted?
lemme know if i've misunderstood, just absorbing some knowledge on the journey of reviewing your PR :-)
There was a problem hiding this comment.
Yes, except the hash isnt actually considered in the function. It appears to just be passed along to be included in the warning log and only used if and only if the two flows have the same values.
There was a problem hiding this comment.
gotcha, that makes sense :-)
in this case, in the nittiest of picks, i wonder if we can amend the comments to label as valid vs. invalid hash collisions for flows? or instead of using "hash collision" for the valid cases, a different label for visibility that these are valid matching hashes?
Regression DetectorRegression Detector ResultsRun ID: 8aeb3a2e-8cf9-491e-9976-cf73f48e307a Metrics dashboard Target profiles Baseline: ba4cc7a Performance changes are noted in the perf column of each table:
No significant changes in experiment optimization goalsConfidence level: 90.00% There were no significant changes in experiment optimization goals at this confidence level and effect size tolerance.
|
| perf | experiment | goal | Δ mean % | Δ mean % CI | trials | links |
|---|---|---|---|---|---|---|
| ➖ | uds_dogstatsd_to_api_cpu | % cpu utilization | +1.23 | [+0.49, +1.97] | 1 | Logs |
| ➖ | pycheck_lots_of_tags | % cpu utilization | +0.87 | [-1.77, +3.50] | 1 | Logs |
| ➖ | uds_dogstatsd_to_api | ingress throughput | +0.02 | [-0.06, +0.10] | 1 | Logs |
| ➖ | idle | memory utilization | +0.00 | [-0.04, +0.05] | 1 | Logs |
| ➖ | tcp_dd_logs_filter_exclude | ingress throughput | +0.00 | [-0.01, +0.01] | 1 | Logs |
| ➖ | otel_to_otel_logs | ingress throughput | -0.01 | [-0.82, +0.80] | 1 | Logs |
| ➖ | file_tree | memory utilization | -0.12 | [-0.21, -0.03] | 1 | Logs |
| ➖ | tcp_syslog_to_blackhole | ingress throughput | -0.16 | [-0.22, -0.11] | 1 | Logs |
| ➖ | basic_py_check | % cpu utilization | -1.31 | [-4.20, +1.59] | 1 | Logs |
Bounds Checks
| perf | experiment | bounds_check_name | replicates_passed |
|---|---|---|---|
| ✅ | idle | memory_usage | 10/10 |
Explanation
A regression test is an A/B test of target performance in a repeatable rig, where "performance" is measured as "comparison variant minus baseline variant" for an optimization goal (e.g., ingress throughput). Due to intrinsic variability in measuring that goal, we can only estimate its mean value for each experiment; we report uncertainty in that value as a 90.00% confidence interval denoted "Δ mean % CI".
For each experiment, we decide whether a change in performance is a "regression" -- a change worth investigating further -- if all of the following criteria are true:
-
Its estimated |Δ mean %| ≥ 5.00%, indicating the change is big enough to merit a closer look.
-
Its 90.00% confidence interval "Δ mean % CI" does not contain zero, indicating that if our statistical model is accurate, there is at least a 90.00% chance there is a difference in performance between baseline and comparison variants.
-
Its configuration does not mark it "erratic".
| aggHash3 := flowB1.AggregationHash() | ||
| acc.detectHashCollision(aggHash3, *flowA1, *flowB1) | ||
| assert.Equal(t, uint64(1), acc.hashCollisionFlowCount.Load()) |
There was a problem hiding this comment.
@vicweiss Curious, how did you find a good flowB1 candidate that produces the exact same hash?
There was a problem hiding this comment.
oh, the hash aggHash1 (flowA1) and aggHash3 (flowB1) are actually not equal.
In theory, when acc.detectHashCollision() is called, the hash (1st param) is expected the hash of both flowA1, and flowB1.
For the purpose of this test, faking that part seems to works and might be enough, if we go that route, maybe we can add some comment about the fact it's not a realistic case (flowA1 and flowB2) but is enough for unit testing.
Ideally, we should find a real candidate flows that share the same hash, but might not be easy.
|
/merge |
|
🚂 MergeQueue: pull request added to the queue The median merge time in Use |
dd-mergequeue
bot
deleted the
vic.weiss/flowAccumulator_detectHashCollision_test
branch
What does this PR do?
Adds missing unit test to flowAccumulator.detectHashCollision
Motivation
Increase test coverage
Describe how to test/QA your changes
invoke test --targets=./comp/netflowPossible Drawbacks / Trade-offs
N/A
Additional Notes