Summary

Fixes #6646

Alternative to #6854 — instead of writing a .command.zone file from inside the task, this approach resolves the actual execution zone by parsing Google Batch StatusEvent descriptions, which include the zone in the format zones/<zone>/instances/<instance-id>.

This avoids creating an additional bucket file and keeps all changes within the Google plugin (no core module changes).

Changes

• GoogleBatchTaskHandler.groovy — On task completion, getMachineInfo() calls updateZoneFromEvents() which fetches the task status events and parses the zone from the event description using a regex. Updates CloudMachineInfo with the actual zone. Uses a volatile boolean zoneUpdated flag to ensure the API is called at most once.

Design decisions

• Graceful degradation — If no zone is found in the events (e.g. the event descriptions don't contain zone info), the trace record falls back to the configured region as before
• Read-once semantics — The zoneUpdated flag is set in a finally block, so even if the API call fails, we don't retry on every getMachineInfo() call
• No core module changes — Unlike the .command.zone approach, this doesn't touch TaskRun or add new bucket files

Test plan

• Data-driven test covering: zone from succeeded event, zone from preemption event, no zone in event, empty events
• Null machineInfo and not-completed guards
• Zone resolved only once across multiple getMachineInfo() calls
• API error handled gracefully with fallback to original zone
• All tests pass