You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
If I create a HTTPServer with a connection limit of 2, then I would expect only 2 connections to be made. However, the current implementation allows essentially an unlimited number of connections, subject only to OS limits.
This appears to be because the connection limit logic occurs inside HTTPRequestHandler, an InboundChannelHandler. By definition, the connection must exist for the execution to reach this point. Correct me if I'm wrong, but what I think you're trying to achieve is to limit the number of concurrent requests, not literal connections?
The limiting logic also appears to be evaluated at the end of a HTTP request. This is a fairly major issue, as it allows me to dump massive requests before you try and stop me. At the very least it would be a good idea to move the logic to the very start of "channelRead" to shut the connection down before data is sent.
Hope that helps, me and the rest of the NIO are more than happy to help at any time :)
The text was updated successfully, but these errors were encountered:
Thanks for your comments @Davidde94 !
Is this PR what you mentioned by moving it to the start of the logic?
An additional question: is channelInactive called when I close the connection?
Eg. do I need to decrease manually the counter everywhere context.close() is called or is it automatic?
If I create a HTTPServer with a connection limit of 2, then I would expect only 2 connections to be made. However, the current implementation allows essentially an unlimited number of connections, subject only to OS limits.
This appears to be because the connection limit logic occurs inside HTTPRequestHandler, an
InboundChannelHandler
. By definition, the connection must exist for the execution to reach this point. Correct me if I'm wrong, but what I think you're trying to achieve is to limit the number of concurrent requests, not literal connections?The limiting logic also appears to be evaluated at the end of a HTTP request. This is a fairly major issue, as it allows me to dump massive requests before you try and stop me. At the very least it would be a good idea to move the logic to the very start of "channelRead" to shut the connection down before data is sent.
Hope that helps, me and the rest of the NIO are more than happy to help at any time :)
The text was updated successfully, but these errors were encountered: