Azure / karpenter-provider-azure / 23127799296

Pull #1530

github

feat: detect Machine API quota errors earlier via sync polling window

After PUT Machine returns 201, poll Machine status for up to 15s
(configurable via MACHINE_SYNC_WAIT_TIMEOUT). If quota/capacity error
surfaces during this window, handle it synchronously — core retries
on same NodeClaim instead of delete+recreate cycle.

Exit conditions:
- ProvisioningState=Failed → handle error immediately
- ProvisioningState=Succeeded → return early (no wasted wait)
- Timeout while Creating → hand off to background LRO as today
- MACHINE_SYNC_WAIT_TIMEOUT=0 → skip entirely (backward compat)

New env vars: MACHINE_SYNC_WAIT_TIMEOUT (default 15s),
MACHINE_SYNC_POLL_INTERVAL (default 3s).

8 new sync wait tests + 7 new duration utility tests.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

120 of 143 new or added lines in 6 files covered. (83.92%)

58879 of 62746 relevant lines covered (93.84%)

0.95 hits per line