Compare commits

...

4 Commits

Author SHA1 Message Date
Colter Downing 25db78e39d Fix healthcheck with separate session 2025-10-01 18:04:31 -07:00
Colter-Downing 1b37054dec Merge pull request #28 from vast-ai/bugfix/backend-timeout-infinite
Bugfix/backend timeout infinite
2025-08-28 11:22:33 -07:00
Colter-Downing 1a1e4174b8 Merge pull request #29 from vast-ai/bugfix/comfyui-json-cost-fix
Set cost to 100
2025-08-28 11:22:21 -07:00
Rob Ballantyne b8377c4081 Set cost to 100 2025-08-28 16:13:17 +01:00
3 changed files with 17 additions and 5 deletions
+13 -2
View File
@@ -190,18 +190,30 @@ class Backend:
log.debug(f"Exception in main handler loop {e}")
return web.Response(status=500)
@cached_property
def healthcheck_session(self):
"""Dedicated session for healthchecks to avoid conflicts with API session"""
log.debug("creating dedicated healthcheck session")
connector = TCPConnector(
force_close=True, # Keep this for isolation
enable_cleanup_closed=True,
)
timeout = ClientTimeout(total=10) # Reasonable timeout for healthchecks
return ClientSession(timeout=timeout, connector=connector)
async def __healthcheck(self):
health_check_url = self.benchmark_handler.healthcheck_endpoint
if health_check_url is None:
log.debug("No healthcheck endpoint defined, skipping healthcheck")
return
while True:
await sleep(10)
if self.__start_healthcheck is False:
continue
try:
log.debug(f"Performing healthcheck on {health_check_url}")
async with self.session.get(health_check_url) as response:
async with self.healthcheck_session.get(health_check_url) as response:
if response.status == 200:
log.debug("Healthcheck successful")
elif response.status == 503:
@@ -210,7 +222,6 @@ class Backend:
f"Healthcheck failed with status: {response.status}"
)
else:
# endpoint not ready yet so bail
log.debug(f"Healthcheck Endpoint not ready: {response.status}")
except Exception as e:
log.debug(f"Healthcheck failed with exception: {e}")
+3 -2
View File
@@ -13,8 +13,9 @@ with open("workers/comfyui/misc/test_prompts.txt", "r") as f:
test_prompts = f.readlines()
def count_workload() -> float:
# Always 1.0 where there is a single instance of ComfyUI handling requests
return 1.0
# Always 100.0 where there is a single instance of ComfyUI handling requests
# Results will indicate % or a job completed per second. Avoids sub 0.1 sec performance indication
return 100.0
@dataclasses.dataclass
class ComfyWorkflowData(ApiPayload):
+1 -1
View File
@@ -70,7 +70,7 @@ class ComfyWorkflowHandler(EndpointHandler[ComfyWorkflowData]):
@property
def healthcheck_endpoint(self) -> Optional[str]:
return None
return "/health"
@classmethod
def payload_cls(cls) -> Type[ComfyWorkflowData]: