• Replace panic calls with structured error logging in threadSafeRand
  - Invalid bounds and crypto/rand failures now return safe zero-values
  - Prevents uncontrolled crashes and internal error detail leakage
• Check pairingPurge in IsStaticProvider for epoch-boundary subscriptions
  - Providers in pairingPurge may still serve active WS subscriptions
  - Prevents misclassification after epoch transitions
• Stop ticker in PeriodicProbeProviders to prevent resource leak
  - Add defer ticker.Stop() to release underlying timer
• Skip per-candidate log allocations when debug disabled
  - Gate selectionProbabilities and logAttrs behind utils.IsDebugEnabled()
  - Eliminates O(n) allocations per relay in production

Fix epoch-boundary subscriptions and resource leak

• Extended IsStaticProvider to check pairingPurge map for providers from previous epoch
• Added defer ticker.Stop() in PeriodicProbeProviders to prevent timer resource leak
• Updated documentation to clarify handling of purged providers in subscriptions

protocol/lavasession/consumer_session_manager.go

Gate debug log allocations behind debug flag

• Wrapped selectionProbabilities map and logAttrs slice construction in IsDebugEnabled() check
• Moved expensive per-candidate log payload building behind debug gate
• Eliminates O(n) allocations on every provider selection in production

protocol/provideroptimizer/weighted_selector.go

Add IsDebugEnabled utility function

• Added IsDebugEnabled() function to check if debug-level logging is active
• Provides utility for guarding expensive computations only needed for debug logs
• Compares defaultGlobalLogLevel against zerolog.DebugLevel

utils/lavalog.go

Description

threadSafeRand methods now log and return 0 on invalid bounds or crypto/rand failures, which
can silently produce incorrect/non-random outputs and mask bugs. The logs also omit the underlying
err, reducing debuggability for these failure points.

Code

utils/rand/rand.go[R45-52]

+		utils.LavaFormatError("Intn called with invalid bound", nil, utils.LogAttr("n", n))
+		return 0
  }
  maxVal := big.NewInt(int64(n))
  result, err := cryptorand.Int(cryptorand.Reader, maxVal)
  if err != nil {
-		panic("crypto/rand failed: " + err.Error())
+		utils.LavaFormatError("crypto/rand failure in Intn", nil)
+		return 0

Evidence

Compliance requires handling failure points with meaningful context and without silent failures. The
updated code replaces panics with LavaFormatError(...) and return 0, and for crypto/rand
failures does not include the actual err, making failures both silent to callers and lacking
actionable context.

Rule 3: Generic: Robust Error Handling and Edge Case Management
utils/rand/rand.go[45-52]
utils/rand/rand.go[114-121]

Agent prompt

The issue below was found during a code review. Follow the provided context and guidance below and implement a solution

## Issue description
`threadSafeRand` methods (`Intn`, `Int63n`, and other RNG helpers changed similarly) now log errors and return `0`, which silently changes behavior and can cause incorrect randomness-dependent behavior. Additionally, `crypto/rand` failures are logged without the underlying `err`, reducing diagnostic value.
## Issue Context
These methods previously `panic`ed on invalid bounds and `crypto/rand` failures. If callers rely on correct randomness, returning `0` is a silent failure and can bias selection logic or conceal bugs.
## Fix Focus Areas
- utils/rand/rand.go[42-55]
- utils/rand/rand.go[62-66]
- utils/rand/rand.go[75-79]
- utils/rand/rand.go[88-92]
- utils/rand/rand.go[101-105]
- utils/rand/rand.go[111-124]

ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.