test_abort_metrics_reset is flaky due to hardware-dependent
fixed sleeps: replace fixed sleeps with polling.
test_metrics_exist_run_batch passes even when the engine crashes
on startup (false positive): add subprocess lifecycle guards.
Signed-off-by: Andreas Karatzas <akaratza@amd.com>
This PR completes the removal of the deprecated vllm:time_per_output_token_seconds
metric that was deprecated in v0.11, hidden in v0.12, scheduled for removal in v0.13,
but delayed until v0.15.
Signed-off-by: carlory <baofa.fan@daocloud.io>
Co-authored-by: Claude Haiku 4.5 <noreply@anthropic.com>