Errorcode 1040 on master server

Phena

Hi there,

While running our game from the office, we often encounter the 1040 client timeout on different machines.

I would like to be able to find the reason for this. Most timeouts happen quickly after a machine has hosted a session
and a second player joins this session. At first I thought this might have to do with the outgoing data ( which was ~30kB/s at that point ). I've
managed to get this down to around 8kB/s. Now it still happens, but less often. ( Extra question, what's a good amount of kB/s to strive for? How much does Photon support? )

Does anyone know what could cause this issue? Some machines hardly ever get a timeout, or only after a longer session (15min+). Also at the moment of the timeout there doesn't seem to be any weird network activity.

Thanks in advance.

Tobias

There is no fixed limit for traffic per second that will work or won't. It depends on the connection, your hard- and software. It depends on what you target as platform, basically.
Sorry for being so vague but good values are those that work for you.

You should try to use another wireless router or different machines just to test. We found this is more often a problem than the actual bytes/sec.

In general, less is better.
What are you sending? Is there room to improve this more?

Phena

Thanks for your reply Tobias,

I synch a large amount of Agents over the network, there is a lot of optimization to be done on this part but as it doesn't seem the issue
this is not the main priority at the moment.

I've been running a test this afternoon, by setting the Send and Serialize rate of the network to a higher value. I now send 50 times and Serialize 25 times per second. (opposed to 20:10) Obviously this results in more traffic and should not be kept this way, but it did solve my 1040 error. Before, the client would consistantly get the timeout, now it didn't occur over the course of 2 hours.

Does this mean that before, a packet loss or two could result in a timeout, but because there are way more being sent now, the chance of this happening is a lot lower?

Tobias

I would say "no" for "sending more makes a timeout less likely". But for some reason, it seems to help you at the moment.

By design, the send code should send as many packages as necessary to get all your updates out. The sendRate is more or less the target frequency but more packages could be created on demand.
You should try the 10 serialize/sec with the higher send rate. This won't cause more content to be created and sent but it will raise the frequency for creating packages.

Phena

Alright, i'll be testing that now and comparing it to the old settings.

Is there a way to set a timeout time on the client? Or is it recommended not to change other default settings?

Tobias

I wouldn't change the timeout in most cases.
You can't change the server-side timeout in the Photon Cloud anyways, so having a longer client timeout is not particularly helpful.