mirror of
https://github.com/trustgraph-ai/trustgraph.git
synced 2026-04-25 00:16:23 +02:00
fix: prevent duplicate dispatcher creation race condition in invoke_global_service (#715)
* fix: prevent duplicate dispatcher creation race condition in invoke_global_service Concurrent coroutines could all pass the `if key in self.dispatchers` check before any of them wrote the result back, because `await dispatcher.start()` yields to the event loop. This caused multiple Pulsar consumers to be created on the same shared subscription, distributing responses round-robin and dropping ~2/3 of them — manifesting as a permanent spinner in the Workbench UI. Apply a double-checked asyncio.Lock in both `invoke_global_service` and `invoke_flow_service` so only one dispatcher is ever created per service key. * test: add concurrent-dispatch tests for race condition fix Add asyncio.gather-based tests that verify invoke_global_service and invoke_flow_service create exactly one dispatcher under concurrent calls, preventing the duplicate Pulsar consumer bug.
This commit is contained in:
parent
10a931f04c
commit
d4723566cb
2 changed files with 133 additions and 53 deletions
|
|
@ -118,6 +118,7 @@ class DispatcherManager:
|
|||
|
||||
self.flows = {}
|
||||
self.dispatchers = {}
|
||||
self.dispatcher_lock = asyncio.Lock()
|
||||
|
||||
async def start_flow(self, id, flow):
|
||||
logger.info(f"Starting flow {id}")
|
||||
|
|
@ -165,30 +166,28 @@ class DispatcherManager:
|
|||
|
||||
key = (None, kind)
|
||||
|
||||
if key in self.dispatchers:
|
||||
return await self.dispatchers[key].process(data, responder)
|
||||
if key not in self.dispatchers:
|
||||
async with self.dispatcher_lock:
|
||||
if key not in self.dispatchers:
|
||||
request_queue = None
|
||||
response_queue = None
|
||||
if kind in self.queue_overrides:
|
||||
request_queue = self.queue_overrides[kind].get("request")
|
||||
response_queue = self.queue_overrides[kind].get("response")
|
||||
|
||||
# Get queue overrides if specified for this service
|
||||
request_queue = None
|
||||
response_queue = None
|
||||
if kind in self.queue_overrides:
|
||||
request_queue = self.queue_overrides[kind].get("request")
|
||||
response_queue = self.queue_overrides[kind].get("response")
|
||||
dispatcher = global_dispatchers[kind](
|
||||
backend = self.backend,
|
||||
timeout = 120,
|
||||
consumer = f"{self.prefix}-{kind}-request",
|
||||
subscriber = f"{self.prefix}-{kind}-request",
|
||||
request_queue = request_queue,
|
||||
response_queue = response_queue,
|
||||
)
|
||||
|
||||
dispatcher = global_dispatchers[kind](
|
||||
backend = self.backend,
|
||||
timeout = 120,
|
||||
consumer = f"{self.prefix}-{kind}-request",
|
||||
subscriber = f"{self.prefix}-{kind}-request",
|
||||
request_queue = request_queue,
|
||||
response_queue = response_queue,
|
||||
)
|
||||
await dispatcher.start()
|
||||
self.dispatchers[key] = dispatcher
|
||||
|
||||
await dispatcher.start()
|
||||
|
||||
self.dispatchers[key] = dispatcher
|
||||
|
||||
return await dispatcher.process(data, responder)
|
||||
return await self.dispatchers[key].process(data, responder)
|
||||
|
||||
def dispatch_flow_import(self):
|
||||
return self.process_flow_import
|
||||
|
|
@ -299,36 +298,35 @@ class DispatcherManager:
|
|||
|
||||
key = (flow, kind)
|
||||
|
||||
if key in self.dispatchers:
|
||||
return await self.dispatchers[key].process(data, responder)
|
||||
if key not in self.dispatchers:
|
||||
async with self.dispatcher_lock:
|
||||
if key not in self.dispatchers:
|
||||
intf_defs = self.flows[flow]["interfaces"]
|
||||
|
||||
intf_defs = self.flows[flow]["interfaces"]
|
||||
if kind not in intf_defs:
|
||||
raise RuntimeError("This kind not supported by flow")
|
||||
|
||||
if kind not in intf_defs:
|
||||
raise RuntimeError("This kind not supported by flow")
|
||||
qconfig = intf_defs[kind]
|
||||
|
||||
qconfig = intf_defs[kind]
|
||||
if kind in request_response_dispatchers:
|
||||
dispatcher = request_response_dispatchers[kind](
|
||||
backend = self.backend,
|
||||
request_queue = qconfig["request"],
|
||||
response_queue = qconfig["response"],
|
||||
timeout = 120,
|
||||
consumer = f"{self.prefix}-{flow}-{kind}-request",
|
||||
subscriber = f"{self.prefix}-{flow}-{kind}-request",
|
||||
)
|
||||
elif kind in sender_dispatchers:
|
||||
dispatcher = sender_dispatchers[kind](
|
||||
backend = self.backend,
|
||||
queue = qconfig,
|
||||
)
|
||||
else:
|
||||
raise RuntimeError("Invalid kind")
|
||||
|
||||
if kind in request_response_dispatchers:
|
||||
dispatcher = request_response_dispatchers[kind](
|
||||
backend = self.backend,
|
||||
request_queue = qconfig["request"],
|
||||
response_queue = qconfig["response"],
|
||||
timeout = 120,
|
||||
consumer = f"{self.prefix}-{flow}-{kind}-request",
|
||||
subscriber = f"{self.prefix}-{flow}-{kind}-request",
|
||||
)
|
||||
elif kind in sender_dispatchers:
|
||||
dispatcher = sender_dispatchers[kind](
|
||||
backend = self.backend,
|
||||
queue = qconfig,
|
||||
)
|
||||
else:
|
||||
raise RuntimeError("Invalid kind")
|
||||
|
||||
await dispatcher.start()
|
||||
await dispatcher.start()
|
||||
self.dispatchers[key] = dispatcher
|
||||
|
||||
self.dispatchers[key] = dispatcher
|
||||
|
||||
return await dispatcher.process(data, responder)
|
||||
return await self.dispatchers[key].process(data, responder)
|
||||
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue