Actor Lifecycle terminated abruptly in Akka.Dotnet.Streams #6275
Replies: 1 comment
-
Have you tried adding some try/catch logging to your flow logic to see where things are failing? Unless I'm missing something there may be a lack of framing in the bytestrings received, which could cause some issues in your parse logic:
If you don't have proper framing on your reads, it is possible for you to receive 'incomplete' data at this stage. Normally, when sending data to/from Streams this way, I use In the case of HTTP, you may need to write your own framing logic, but |
Beta Was this translation helpful? Give feedback.
-
Hello.
This question is related to #6246. But this is a different problem. (Does it makes sense to merge these discussions ?)
Basically I want to be able to receive a really big amount of HTTP requests and every second send somewhere else request with the combined content of the requests received in that second.
I thought that Using Akka.Streams would allow me to create an actor (sourceTask from the code snippet) per request received and that sending the HTTP response as fast as possible would be key.
Here's my attempt. This fails for many requests but works if you send like 6 requests and use .Grouped(4), for example.
This error happens when I attempt to send 600 HTTP requests at the same time. It can be related to the limit of actors on a system, but I tried to reconfigure akka.stream.materializer.max-input-buffer-size from 16 to 512 and I didn't get anywhere.
This Warning message happens when I send 600 requests but with 0.1 seconds between each other. In here it also seems like the .post1 Flow is "attached" the TCP connection, even after a HTTP response has been sent.
Thank you for reading.
Hope I am not "wasting" anyone's time with too "basic" questions.
Beta Was this translation helpful? Give feedback.
All reactions