Increasing RTMP Timeout Period

I’ve been playing around with VHost.xml’s values for Application/PingTimeout, Application/ValidationFrequency and Client/ClientTimeout and blocking off the RTMP port using IPSec on the Wowza machine. With values of 90000 everywhere, I’m still getting Wowza disconnection events around 30 secs after blocking the port. Are there any other settings I should check or is my port-blocking test flawed?

My end goal is to increase the timeout period to cater for unstable mobile connections.

I’m getting the same behavior by simply pulling out the LAN cable from the client machine, so I don’t think the IPSec method is causing problems.

Your VHost.xml changes might not be having any effect because the Application.xml settings override the VHost.xml settings for an application.

Rihcared

There are PingTimeout and ValidationFreqency settings in the Application.xml. If these are set to the default, that is what your application is going to be using.

<Root>
	<Application>
		<!-- Uncomment to set application level timeout values
		<ApplicationTimeout>60000</ApplicationTimeout>
		<PingTimeout>12000</PingTimeout>
		<ValidationFrequency>8000</ValidationFrequency>

PingTimeout and ValidationFrequency are only relevant to RTMP connections. That might be the problem, if you are looking for effect in other client types.

Richard

Checked my Application.xml settings and there aren’t any overrides. I’ve also tried placing higher timeout settings in Application.xml (I based it on the commented-out section in the vod sample’s Application.xml) but I’m still getting disconnect events about 20 seconds after a plug pull.

Could it be that I’m just misusing the events? Should I be catching some sort of timeout event instead of this disconnect event that may be fired multiple times before a timeout?

Could I get confirmation that things are working as intended and I’m not hitting some misconfiguration or event misuse? If this is how things are, I’ll go on ahead and hand-code a solution to simulate a longer “timeout” period.

Already tried editing Application.xml a few posts back (retried it just now as a sanity check, too) and I’m still getting the disconnect event after 20ish seconds. And, yes, I’m using RTMP as the protocol. Does it matter if I’m using the RTMPe flavor?

Well, I’ll see if I can get this working on some other servers. If not, I guess I’m left with hand-coding a solution that persists behavior across multiple disconnect events.