Closed crbean closed 1 year ago
There is no related record in the log. When you run the py_spy command to view the current stack, you can find that the thread is blocked.
Thread 33994 (idle) select (selectors.py:469) _run_once (asyncio/base_events.py:2381) run_forever (asyncio/base_events.py:857) run_until_complete (asyncio/base_events.py:634) run (asyncio/runners.py:300) run_async (jina/helper.py:1854) post (jina/clients/mixin.py:1041)
Hello, can you share ur Flow and code to reproduce?
f = Flow(port=JINA_PORT, host=JINA_IP, timeout_send=TIMEOUT_SEND).add(uses=XXX, replicas=int(2), host=JINA_IP)
what is your JINA_IP
is it localhost?
Also, it is important to know what is the Executor doing.
What do you see when setting JINA_LOG_LEVEL=DEBUG
?
JINA_IP is 127.0.0.1 how to know what is the Executor doing? Does the value of JUNA_LOG_LEVEL need to be set to debug?
sorry, Debug not True
Is thr Executor something you develop or u use one from the Hub?
Yeah,we develop a executor like clip-as-service——using bert to inference which is the model trained by transformers
.
crbean and i use the same service.
one thing that is weird is that in the post
it seems that you are passing a Document
class and not a Document
instance
I would need to see some logs to understand the problem. It is hard to take actions with the information provided
@JoanFM
one thing that is weird is that in the
post
it seems that you are passing aDocument
class and not aDocument
instance
No,it is DocumentArray([Document(text=text) for text in data_list])
I would need to see some logs to understand the problem. It is hard to take actions with the information provided
Trying to collect.
Hello @wqh17101 ,
is there any progress on this? have u been able to collect more information?
@JoanFM No,just close it now, and i will open it if i need.
Will close it for the time being, do not hesitate to reopen an issue for any problem
Describe the bug
When the GRPC protocol is used to invoke the jina server, the synchronization request is always blocked.
Describe how you solve it
client.post('/', Document, parameters=XXX, results_in_order=True, continue_on_error=True,max_attempts=3)
Environment
Screenshots