fix: avoid int32 overflow when maxTrafficWeight is large

[semx]

Fixes canary replica counts being computed from an overflowed int32 product when maxTrafficWeight is large.

The problem

Several places multiply a replica count by a traffic weight while both are still int32, and only widen the result afterwards:

ceilWeightedCanaryCount := int32(math.Ceil(float64(specReplicas*desiredWeight) / float64(maxWeight)))

maxTrafficWeight has no upper bound (the CRD only says format: int32), and the repo's own tests use 100000000. At that scale the product exceeds math.MaxInt32 for quite ordinary replica counts, and the arithmetic wraps.

approximateWeightedCanaryStableReplicaCounts with maxWeight=100000000, maxSurge=0:

replicas	weight	canary (before)	canary (expected)
43	50%	-21	22
100	50%	8	50
43	100%	1	43

A negative canary count at 50%, and a single canary replica when the rollout is fully shifted to canary.

checkReplicasAvailable is affected the same way: the overflowed product goes negative, so availableReplicas < desiredReplicas is never true and it reports that there are enough replicas. With 43 replicas at 50% weight and zero of them available it returns true, which lets traffic shift to pods that are not ready.

The threshold is replicas * weight > MaxInt32, so with maxTrafficWeight=100000000 it starts at about 22 replicas.

The fix

Widen the operands before multiplying. trafficWeightToReplicas in the same file already does exactly this:

func trafficWeightToReplicas(replicas, weight, maxWeight int32) int32 {
	return int32(math.Ceil(float64(weight) * float64(replicas) / float64(maxWeight)))
}

The same shape is applied to approximateWeightedCanaryStableReplicaCounts, weightDelta, checkReplicasAvailable, the dynamic-stable-scale weight in trafficrouting.go, and the actual-weight display in rollout_info.go. Behaviour for the default maxTrafficWeight=100 is unchanged — the existing table cases all still pass.

Tests

Added cases to the existing tables. Both fail on master and pass with this change:

• TestApproximateWeightedNewStableReplicaCounts — expected: 22, actual: -21 and expected: 50, actual: 8
• TestCheckReplicasAvailableWithCustomMaxTrafficWeight — 43 replicas, 0 available, 50% weight must not report available

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud

[github-actions]

Published E2E Test Results

  4 files    4 suites   3h 52m 55s ⏱️
123 tests 114 ✅  7 💤 2 ❌
496 runs  464 ✅ 28 💤 4 ❌

For more details on these failures, see this check.

Results for commit c3cd007.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 85.07%. Comparing base (d1dd467) to head (c3cd007).

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #4923      +/-   ##
==========================================
- Coverage   85.12%   85.07%   -0.06%     
==========================================
  Files         166      166              
  Lines       19238    19239       +1     
==========================================
- Hits        16376    16367       -9     
- Misses       2019     2024       +5     
- Partials      843      848       +5     

Flag	Coverage Δ
e2e	52.42% <83.33%> (-0.10%)	⬇️
unit-tests	81.55% <100.00%> (+<0.01%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[github-actions]

Published Unit Test Results

2 584 tests   2 584 ✅  3m 24s ⏱️
  130 suites      0 💤
    1 files        0 ❌

Results for commit c3cd007.