Re: [naviserver-devel] revproxy timeouts

[Gustaf N. <ne...@wu...>]

Just to be sure: timeouts during connection creation work fine, but read 
timeouts from the back-end lead to the problem. ... and this happened 
with an https backend. And you are experiencing this under Linux. Did i 
understand this correctly?

actually, ns_connchan has no own timeout semantics, but relies on the 
Ns_SockCallback* infrastructure. In general, the Tcl callback can signal 
with a result "0", that the current connection channel should be closed 
automatically.

-g
Am 31.03.17 um 16:47 schrieb David Osborne:
> Hi,
>
> I've been trying to get revproxy to work with a timeout. So if the 
> backend doesn't reply within $timeout secs, then the connection is closed.
>
> What I thought we could do was set a timeout when creating the 
> callback to call backendReply.
> Then when the callback is triggered with $when eq "t" we could return 
> a 504 then close the connection.
>
> This is an outline if what I am trying:
> https://bitbucket.org/davidqc/revproxy/commits/658704673bcb07d27962a9b4d4a89a07a64c5843
>
> This appears to work initally. When accessing a backend page which 
> takes over 10 seconds to respond, the client gets a 504.
> But it appears that things aren't cleaned up properly and the 2 socket 
> callbacks persist after the timeout.
>
> Inline images 2
>
> This has the effect that the proxy is essentially unresponsive and 
> doesn't forward any more requests (even the Control Ports are not 
> accessible locally).
> If I restart the backend server, that brings the proxy back to life. 
> The proxy and the backend are running on different ports on the same 
> server in this case.
>
> Should this approach work or have I misunderstood how ns_connchan 
> timeouts should work?
>
> Thanks,
> -- 
> David