@andrey.vasnetsov What is the server HTTP timeout value?

There is a client_timeout, which defines the timeout of client request transmission. Default is 5 seconds. But it is not the same as service timeout, which is unlimited. So if requests is executed even a minute, server won't break the connection

I see. I was reading the qdrant timeouts. I found grpc ones for p2p and so on. I couldn't find any for HTTP requests. Setting the clients to any value which is lower than the server one would be useless. That's the reason for the question.

So it's unlimited. Does it handle possible scenarios of HTTP attacks which try to hold connection to overload the web workers?

qdrant is not expected to be exposed to the internet on it's own. Usually you either have a backend service or a proxy.

Sure. Nginx, Kong or other API or application gateway can handle this. Just curious as this absence of limit has some extra implications on security handling.
Thank you!

im using fastapi to manage qdrant, qdrant its not exposed to internet, i have 3 servers and to avoid to get stuck when a node is down i put a timeout in the requests