refactor metrics event processing

[harp-intel]

eliminate race condition in perf event processing

This pull request refactors how perf event output is collected and processed in the metrics package, shifting from line-by-line string handling to batching by interval using byte slices. The changes improve the accuracy and efficiency of event grouping, simplify channel usage, and add robust parsing and testing for interval extraction. Most changes are in cmd/metrics/metrics.go, with supporting updates to target and script streaming interfaces.

Perf event batching and processing improvements:

• Refactored perf event output handling to batch events by interval using byte slices ([][]byte), replacing previous timer-based grouping and string channels. This ensures each batch corresponds to a specific perf interval, improving accuracy and reducing race conditions. (cmd/metrics/metrics.go, internal/target/target.go, internal/target/local_target.go, internal/target/remote_target.go, internal/target/helpers.go, internal/script/script.go) [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14]

• Added the extractInterval function to parse the interval value from perf event JSON lines, with robust error handling for missing or malformed data. (cmd/metrics/event_frame.go)

• Updated the perf output processing pipeline (processPerfOutput) to consume batches from a channel, process each batch per interval, and drain remaining batches after context cancellation, ensuring no events are lost. (cmd/metrics/metrics.go) [1] [2] [3] [4]

Testing and error handling enhancements:

• Added comprehensive unit tests for the new extractInterval function, covering valid, missing, and malformed interval cases. (cmd/metrics/event_frame_test.go)

• Improved error messages and parsing robustness for event JSON, and removed unnecessary error checks for CPU range parsing. (cmd/metrics/event_frame.go) [1] [2]

Supporting interface and signature changes:

• Updated all relevant interfaces and method signatures to use chan []byte instead of chan string for stdout and stderr streaming, ensuring consistency across local and remote target command execution and script streaming. (internal/target/target.go, internal/target/local_target.go, internal/target/remote_target.go, internal/target/helpers.go, internal/script/script.go) [1] [2] [3] [4] [5] [6]

[Copilot]

Pull Request Overview

This pull request refactors the metrics perf event processing system to eliminate race conditions and improve thread safety. The main change replaces a timer-based approach for batching perf events with an interval-based batching system that groups events by their interval values using channel-based communication.

• Introduced interval-based event batching to replace timer-based frame detection
• Changed channel types from string to []byte for better performance and memory management
• Added comprehensive test coverage for the new interval extraction functionality

Reviewed Changes

Copilot reviewed 8 out of 8 changed files in this pull request and generated 3 comments.

Show a summary per file

File	Description
internal/target/target.go	Updated interface signature to use []byte channels instead of string
internal/target/remote_target.go	Updated method signature to match new interface
internal/target/local_target.go	Updated method signature to match new interface
internal/target/helpers.go	Modified to send []byte data directly and create buffer copies to avoid race conditions
internal/script/script.go	Updated function signature to use []byte channels
cmd/metrics/metrics.go	Major refactor implementing interval-based batching with new channel communication and processing logic
cmd/metrics/event_frame_test.go	Added comprehensive tests for the new extractInterval function
cmd/metrics/event_frame.go	Added extractInterval function and cleaned up error handling

---

_{Tip: Customize your code reviews with copilot-instructions.md. Create the file or learn how to get started.}

[adgubrud]

Looks good to me