My api endpoint process the request for more than 60 seconds. Then empty reply is being sent to the client. How would I increase the request timeout? Thanks
Are you trying to increase the timeout or reduce the 60 processing time? 60 seconds seems long but I also have no idea what your endpoint is doing.
In the slack a similar question has been asked today. There the long response time was the result of querying an external API. So if this is related, then I fear, that there is nothing we can do about the actual processing time, but instead need to increase the timeout somehow.
I am trying to increase the timeout. I am sending a file to the server and processing the data and sending the data to another server. It may take more than one minute based on the size of file.
Since this was already in the slack today with another user, I got permission to post his minimal repository:
The 65 second sleep in the action is to simulate the foreign API call and not subject to change.
2 Likes
it’s a new default in Cowboy 2 - can be changed using protocol_options - idle_timeout in the endpoint config
config :hello, HelloWeb.Endpoint,
http: [ port: 4000,
protocol_options: [idle_timeout: 5_000_000]
],...
8 Likes
This worked on my local clone of the linked repository:
$ time curl -v localhost:4001/api/test
* Trying ::1...
* TCP_NODELAY set
* connect to ::1 port 4001 failed: Connection refused
* Trying 127.0.0.1...
* TCP_NODELAY set
* Connected to localhost (127.0.0.1) port 4001 (#0)
> GET /api/test HTTP/1.1
> Host: localhost:4001
> User-Agent: curl/7.64.0
> Accept: */*
>
< HTTP/1.1 200 OK
< cache-control: max-age=0, private, must-revalidate
< content-length: 17
< content-type: application/json; charset=utf-8
< date: Wed, 13 Mar 2019 11:45:38 GMT
< server: Cowboy
< x-request-id: 2m5o42cfnvn7nmsujk0000e3
<
* Connection #0 to host localhost left intact
[{"test":"test"}]
The changes I did:
diff --git a/config/dev.exs b/config/dev.exs
index 6e25348..3e4b96d 100644
--- a/config/dev.exs
+++ b/config/dev.exs
@@ -7,7 +7,7 @@ use Mix.Config
# watchers to your application. For example, we use it
# with webpack to recompile .js and .css sources.
config :demo, DemoWeb.Endpoint,
- http: [port: 4001],
+ http: [port: 4001, protocol_options: [idle_timeout: 70_000]],
debug_errors: true,
code_reloader: true,
check_origin: false,
5 Likes
yep, :infinity also works…
[idle_timeout: :infinity]
this issue also popped up here Video streaming connection closed automatically - one could argue for phoenix having :infinity as a default… on the other hand 60 secs is a “sane” default…
would say there might be a need for an explicit config showing 60_000 as the configured… the hidden default kinda falls into the implicit category - that I’m not a fan of…
4 Likes
Are there any reverse-proxies (nginx, HAProxy, Apache, etc.) sitting in front of the app when you experience the timeouts? They typically have their own backend timeout settings to watch out for.
Thanks. It worked.
2 Likes
I know I’m late to the party, but had a similar issue and found out about the solution independently and wanted to see if somebody else had the issue (that’s how I found this thread); in my case I stumbled upon it when using IEx.pry inside a controller’s action to do some debugging. Luckily last year when I was learning Plug, had the same issue and eventually learned about that Cowboy2 option after a long while looking around the docs of Plug and Cowboy.
2 Likes
