sascha/esphome
1
0
Fork 0
mirror of https://github.com/esphome/esphome.git synced 2025-09-02 03:12:20 +01:00
Code Issues Releases Wiki Activity

[scheduler] Fix retry race condition on cancellation (#9788)

Browse Source
This commit is contained in:
J. Nick Koston
2025-07-25 08:14:15 -10:00
committed by GitHub
parent 9ac10d7276
commit 88ccde4ba1
4 changed files with 104 additions and 33 deletions
Download Patch File Download Diff File Expand all files Collapse all files

78
tests/integration/fixtures/scheduler_retry_test.yaml
View File

@@ -10,7 +10,7 @@ esphome:
host:
api:
logger:
level: VERBOSE
level: VERY_VERBOSE
globals:
- id: simple_retry_counter
@@ -19,6 +19,9 @@ globals:
- id: backoff_retry_counter
type: int
initial_value: '0'
- id: backoff_last_attempt_time
type: uint32_t
initial_value: '0'
- id: immediate_done_counter
type: int
initial_value: '0'
@@ -35,20 +38,55 @@ globals:
type: int
initial_value: '0'
# Using different component types for each test to ensure isolation
sensor:
- platform: template
name: Test Sensor
id: test_sensor
name: Simple Retry Test Sensor
id: simple_retry_sensor
lambda: return 1.0;
update_interval: never
- platform: template
name: Backoff Retry Test Sensor
id: backoff_retry_sensor
lambda: return 2.0;
update_interval: never
- platform: template
name: Immediate Done Test Sensor
id: immediate_done_sensor
lambda: return 3.0;
update_interval: never
binary_sensor:
- platform: template
name: Cancel Retry Test Binary Sensor
id: cancel_retry_binary_sensor
lambda: return false;
- platform: template
name: Empty Name Test Binary Sensor
id: empty_name_binary_sensor
lambda: return true;
switch:
- platform: template
name: Script Retry Test Switch
id: script_retry_switch
optimistic: true
- platform: template
name: Multiple Same Name Test Switch
id: multiple_same_name_switch
optimistic: true
script:
- id: run_all_tests
then:
# Test 1: Simple retry
- logger.log: "=== Test 1: Simple retry ==="
- lambda: |-
auto *component = id(test_sensor);
auto *component = id(simple_retry_sensor);
App.scheduler.set_retry(component, "simple_retry", 50, 3,
[](uint8_t retry_countdown) {
id(simple_retry_counter)++;
@@ -65,19 +103,19 @@ script:
# Test 2: Backoff retry
- logger.log: "=== Test 2: Retry with backoff ==="
- lambda: |-
auto *component = id(test_sensor);
static uint32_t backoff_start_time = 0;
static uint32_t last_attempt_time = 0;
backoff_start_time = millis();
last_attempt_time = backoff_start_time;
auto *component = id(backoff_retry_sensor);
App.scheduler.set_retry(component, "backoff_retry", 50, 4,
[](uint8_t retry_countdown) {
id(backoff_retry_counter)++;
uint32_t now = millis();
uint32_t interval = now - last_attempt_time;
last_attempt_time = now;
uint32_t interval = 0;
// Only calculate interval after first attempt
if (id(backoff_retry_counter) > 1) {
interval = now - id(backoff_last_attempt_time);
}
id(backoff_last_attempt_time) = now;
ESP_LOGI("test", "Backoff retry attempt %d (countdown=%d, interval=%dms)",
id(backoff_retry_counter), retry_countdown, interval);
@@ -100,7 +138,7 @@ script:
# Test 3: Immediate done
- logger.log: "=== Test 3: Immediate done ==="
- lambda: |-
auto *component = id(test_sensor);
auto *component = id(immediate_done_sensor);
App.scheduler.set_retry(component, "immediate_done", 50, 5,
[](uint8_t retry_countdown) {
id(immediate_done_counter)++;
@@ -111,8 +149,8 @@ script:
# Test 4: Cancel retry
- logger.log: "=== Test 4: Cancel retry ==="
- lambda: |-
auto *component = id(test_sensor);
App.scheduler.set_retry(component, "cancel_test", 25, 10,
auto *component = id(cancel_retry_binary_sensor);
App.scheduler.set_retry(component, "cancel_test", 30, 10,
[](uint8_t retry_countdown) {
id(cancel_retry_counter)++;
ESP_LOGI("test", "Cancel test retry attempt %d", id(cancel_retry_counter));
@@ -121,7 +159,7 @@ script:
// Cancel it after 100ms
App.scheduler.set_timeout(component, "cancel_timer", 100, []() {
bool cancelled = App.scheduler.cancel_retry(id(test_sensor), "cancel_test");
bool cancelled = App.scheduler.cancel_retry(id(cancel_retry_binary_sensor), "cancel_test");
ESP_LOGI("test", "Retry cancellation result: %s", cancelled ? "true" : "false");
ESP_LOGI("test", "Cancel retry ran %d times before cancellation", id(cancel_retry_counter));
});
@@ -129,7 +167,7 @@ script:
# Test 5: Empty name retry
- logger.log: "=== Test 5: Empty name retry ==="
- lambda: |-
auto *component = id(test_sensor);
auto *component = id(empty_name_binary_sensor);
App.scheduler.set_retry(component, "", 100, 5,
[](uint8_t retry_countdown) {
id(empty_name_retry_counter)++;
@@ -139,7 +177,7 @@ script:
// Try to cancel after 150ms
App.scheduler.set_timeout(component, "empty_cancel_timer", 150, []() {
bool cancelled = App.scheduler.cancel_retry(id(test_sensor), "");
bool cancelled = App.scheduler.cancel_retry(id(empty_name_binary_sensor), "");
ESP_LOGI("test", "Empty name retry cancel result: %s",
cancelled ? "true" : "false");
ESP_LOGI("test", "Empty name retry ran %d times", id(empty_name_retry_counter));
@@ -169,7 +207,7 @@ script:
# Test 7: Multiple same name
- logger.log: "=== Test 7: Multiple retries with same name ==="
- lambda: |-
auto *component = id(test_sensor);
auto *component = id(multiple_same_name_switch);
// Set first retry
App.scheduler.set_retry(component, "duplicate_retry", 100, 5,
@@ -200,7 +238,7 @@ script:
ESP_LOGI("test", "Simple retry counter: %d (expected 2)", id(simple_retry_counter));
ESP_LOGI("test", "Backoff retry counter: %d (expected 4)", id(backoff_retry_counter));
ESP_LOGI("test", "Immediate done counter: %d (expected 1)", id(immediate_done_counter));
ESP_LOGI("test", "Cancel retry counter: %d (expected ~3-4)", id(cancel_retry_counter));
ESP_LOGI("test", "Cancel retry counter: %d (expected 2-4)", id(cancel_retry_counter));
ESP_LOGI("test", "Empty name retry counter: %d (expected 1-2)", id(empty_name_retry_counter));
ESP_LOGI("test", "Component retry counter: %d (expected 2)", id(script_retry_counter));
ESP_LOGI("test", "Multiple same name counter: %d (expected 20+)", id(multiple_same_name_counter));

14
tests/integration/test_scheduler_retry_test.py
View File

@@ -148,16 +148,16 @@ async def test_scheduler_retry_test(
f"Expected at least 2 intervals, got {len(backoff_intervals)}"
)
if len(backoff_intervals) >= 3:
# First interval should be ~50ms
assert 30 <= backoff_intervals[0] <= 70, (
# First interval should be ~50ms (very wide tolerance for heavy system load)
assert 20 <= backoff_intervals[0] <= 150, (
f"First interval {backoff_intervals[0]}ms not ~50ms"
)
# Second interval should be ~100ms (50ms * 2.0)
assert 80 <= backoff_intervals[1] <= 120, (
assert 50 <= backoff_intervals[1] <= 250, (
f"Second interval {backoff_intervals[1]}ms not ~100ms"
)
# Third interval should be ~200ms (100ms * 2.0)
assert 180 <= backoff_intervals[2] <= 220, (
assert 100 <= backoff_intervals[2] <= 500, (
f"Third interval {backoff_intervals[2]}ms not ~200ms"
)
@@ -175,7 +175,7 @@ async def test_scheduler_retry_test(
# Wait for cancel retry test
try:
await asyncio.wait_for(cancel_retry_done.wait(), timeout=2.0)
await asyncio.wait_for(cancel_retry_done.wait(), timeout=3.0)
except TimeoutError:
pytest.fail(
f"Cancel retry test did not complete. Count: {cancel_retry_count}"
@@ -195,8 +195,8 @@ async def test_scheduler_retry_test(
)
# Empty name retry should run at least once before being cancelled
assert 1 <= empty_name_retry_count <= 2, (
f"Expected 1-2 empty name retry attempts, got {empty_name_retry_count}"
assert 1 <= empty_name_retry_count <= 3, (
f"Expected 1-3 empty name retry attempts, got {empty_name_retry_count}"
)
assert empty_cancel_result is True, (
"Empty name retry cancel should have succeeded"