refactor(submit): drop spark-submit retry loop; confirm raises on failure

confirm_submit_job no longer retries on SparkSubmitError. A failed
single attempt is recorded as-is:

  - pending.status = 'FAILED' with the error message in pending.error
  - the function re-raises the SparkSubmitError so the caller (and the
    agent) sees the failure immediately
  - the agent can call get_pending_job to see the persisted FAILED
    state, including the captured error

The 'retry' code path was hiding real failures behind transient
recovery: a spark-submit that died because the script is broken looks
the same as one that died because YARN was momentarily unreachable.
Without retries, the agent gets a clean FAILED state and a clear
exception to act on, instead of a delayed, ambiguous result.

Manual recovery is still possible: re-calling confirm_submit_job on a
FAILED pending resets it to PENDING and runs a single fresh attempt
(see test_confirm_resets_failed_then_succeeds). The reset path is
useful for 'I fixed the script, try again' flows; it just is no longer
the default on every failure.

Removed:
  - the for-attempt loop and time.sleep in submit.py
  - import time in submit.py and test_submit_tool.py
  - confirm_max_retries and confirm_retry_delay_seconds from
    common/config.py (env vars, snapshot, reload)
  - 2 retry tests (test_confirm_retries_spark_submit_failure_then_succeeds,
    test_confirm_exhausts_retries_and_sets_failed)
  - test_confirm_marks_failed_on_spark_submit_error updated to assert
    the new single-attempt + raise contract
  - test_confirm_resets_failed_and_retries renamed to
    test_confirm_resets_failed_then_succeeds (no more retry loop to
    exercise)

Full suite 244 passed (3 fewer than before, matching the removed
tests).

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
This commit is contained in:
Claude
2026-06-30 14:29:27 +08:00
co-authored by Claude Fable 5
parent a66991b4e3
commit 35078f0ae0
3 changed files with 68 additions and 136 deletions
+18 -61
View File
@@ -1,7 +1,6 @@
# coding=utf-8
from pathlib import Path
from unittest.mock import patch
import time
import pytest
@@ -338,6 +337,8 @@ def test_confirm_refuses_non_pending_status(monkeypatch, real_script):
def test_confirm_marks_failed_on_spark_submit_error(monkeypatch, real_script):
"""confirm_submit_job makes a single spark-submit attempt; on failure
it sets pending.status=FAILED and re-raises the SparkSubmitError."""
submit.prepare_submit_job(
connection="prod",
script_path=str(real_script),
@@ -348,71 +349,20 @@ def test_confirm_marks_failed_on_spark_submit_error(monkeypatch, real_script):
app_name="test-app",
)
pid = _last_pending_id()
call_count = [0]
def _raise(_cmd):
call_count[0] += 1
raise SparkSubmitError("boom")
monkeypatch.setattr(submit, "run_spark_submit", _raise)
settings.confirm_max_retries = 0
with pytest.raises(SparkSubmitError):
with pytest.raises(SparkSubmitError, match="boom"):
submit.confirm_submit_job(pending_id=pid)
assert call_count[0] == 1 # single attempt, no retry
p = submit.pending_store.get(pid)
assert p.status == "FAILED"
assert "boom" in (p.error or "")
# --- confirm retry / idempotency ---
def test_confirm_retries_spark_submit_failure_then_succeeds(monkeypatch, real_script):
monkeypatch.setattr(time, "sleep", lambda _s: None)
settings.confirm_max_retries = 2
submit.prepare_submit_job(
connection="prod",
script_path=str(real_script),
queue="default",
executor_memory="2G",
executor_cores=2,
num_executors=2,
app_name="test-app",
)
pid = _last_pending_id()
calls = []
def _maybe_fail(cmd):
calls.append(cmd)
if len(calls) < 3:
raise SparkSubmitError("transient")
return _spark_proc()
monkeypatch.setattr(submit, "run_spark_submit", _maybe_fail)
result = submit.confirm_submit_job(pending_id=pid)
assert result.application_id == "application_17400000001"
assert len(calls) == 3
p = submit.pending_store.get(pid)
assert p.status == "SUBMITTED"
def test_confirm_exhausts_retries_and_sets_failed(monkeypatch, real_script):
monkeypatch.setattr(time, "sleep", lambda _s: None)
settings.confirm_max_retries = 2
submit.prepare_submit_job(
connection="prod",
script_path=str(real_script),
queue="default",
executor_memory="2G",
executor_cores=2,
num_executors=2,
app_name="test-app",
)
pid = _last_pending_id()
calls = []
def _always_fail(cmd):
calls.append(cmd)
raise SparkSubmitError("persistent")
monkeypatch.setattr(submit, "run_spark_submit", _always_fail)
with pytest.raises(SparkSubmitError, match="persistent"):
submit.confirm_submit_job(pending_id=pid)
assert len(calls) == 3
p = submit.pending_store.get(pid)
assert p.status == "FAILED"
assert "persistent" in (p.error or "")
# --- confirm idempotency / manual retry ---
def test_confirm_idempotent_when_submitted(monkeypatch, real_script):
submit.prepare_submit_job(
@@ -440,8 +390,10 @@ def test_confirm_idempotent_when_submitted(monkeypatch, real_script):
assert second.tracking_url == first.tracking_url
def test_confirm_resets_failed_and_retries(monkeypatch, real_script):
monkeypatch.setattr(time, "sleep", lambda _s: None)
def test_confirm_resets_failed_then_succeeds(monkeypatch, real_script):
"""A FAILED pending can be re-confirmed after the user fixes the
underlying issue. confirm_submit_job resets FAILED -> PENDING,
then makes a single fresh attempt (no automatic retry loop)."""
submit.prepare_submit_job(
connection="prod",
script_path=str(real_script),
@@ -457,15 +409,20 @@ def test_confirm_resets_failed_and_retries(monkeypatch, real_script):
"run_spark_submit",
lambda _cmd: (_ for _ in ()).throw(SparkSubmitError("boom")),
)
with pytest.raises(SparkSubmitError):
with pytest.raises(SparkSubmitError, match="boom"):
submit.confirm_submit_job(pending_id=pid)
p = submit.pending_store.get(pid)
assert p.status == "FAILED"
assert "boom" in (p.error or "")
monkeypatch.setattr(submit, "run_spark_submit", lambda _cmd: _spark_proc())
call_count = [0]
def _counting_run(cmd):
call_count[0] += 1
return _spark_proc()
monkeypatch.setattr(submit, "run_spark_submit", _counting_run)
result = submit.confirm_submit_job(pending_id=pid)
assert result.application_id == "application_17400000001"
assert call_count[0] == 1
p = submit.pending_store.get(pid)
assert p.status == "SUBMITTED"
assert p.error is None