Server crashes after linking in to Gahreesen #4 cloth

[Charura]

I read somewhere that the wall is being testing...is this the reason?

I get this:
There seems to be a problem with the online connection. Please quit and try again. #03 (TERMINATED. Server LogOff. Reason. Timed Out

Then a few seconds later: Crash Detection: It seems you lost your connection to the Server - please restart the game.

[Sirius]

#4 ?... Is that the one in the second building observation room ?
Anyway, it shouldn't do it. Crashes should occur when you try using the Wall.

However, I had a same crash two times yesterday. I thought it was because somehow Gahreesen took too much time to load (and I sometime get this in other Ages too).

Maybe that's just a random crash ? I don't know if changing PRPs can crash the game online...

[diafero]

Indeed, I also got some crashes, but accounted them to my slow PC. If I type "/!ping" right after linking in, it takes quite a while until I get the reply back from the server - the same also happens for other ages, like Noloben, it is caused by the client sending a *whole lot* of network messages. The client does not do any other networking during that process (which is totally broken, btw).
In Noloben, this is caused by tons (several hundreds) of SDL-synchronized animations, as the age is not built for networking where usually, one animation controls the movement of several objects. In Gahreesen, after the update, the client send hundreds of plNotifyMsgs to the server. It seems almost all of them are related to objects called "(s,n)PanelResponder_<number>". Sirius, are these messages really necessary, all of them?

Oh, and PRP files can very easily crash the client, Cyan's file parser does not do any kind of sanity checking, most of the time.

[Charura]

The cloth I call #4 is the one on the rock between the two buildings. I just tried going to the initial link-in spot (the picture in the book) and had the same result.

[D'Lanor]

Strange. I had no problems going to the initial Gahreesen link-in spot today on DI. I wanted to have a quick look at the new wall. Then I realized I had not even started to solve the age and left immediately...

[4runner]

When I went to the rock between buildings, I almost got an immediate error-03(terminated):setver logoff:timed out.

[D'Lanor]

In Gahreesen, after the update, the client send hundreds of plNotifyMsgs to the server.

According to my client log files it also sends a pfClimbingWallMsg a few times. Since Alcugs has never needed these could they be the problem?

[Charura]

Tried going to the prison cell (#6 cloth). Same result. Drat I wanted to try the wall.

[diafero]

According to my client log files it also sends a pfClimbingWallMsg a few times. Since Alcugs has never needed these could they be the problem?

No, Alcugs knows about pfClimbingWallMsg and accepts them. It seems I already had to add it earlier to prevent Gahreesen from crashing, as even plain POTS would send them.
I could not find any error on the server indicating new unknown message types. But I know for sure that hundreds of messages sent in a patch can be a huge problem for the old networking engine, so I am quite sure that's the problem here. For some reason, for each of the "buttons" that are used to set the blockers, a plNotifyMsg is sent. This should be done locally on each client instead, without netforcing anything.

[Sirius]

Ok, I'll check these damn responders.
I noticed too that Gahreesen seemed to take more time to load after the addition. I first attributed it to my bad Internet connection...

Strange, the more I look at the Wall, the more objects I find using wrong netForce and all that. I think checking the WHOLE file again to remove useless messages wouldn't be a loss of time. Maybe I'll try doing what you said and make as many objects and responders run locally. Not even to speak about some PtClimbingWallMsg that are send by every client, while one should be enough.

Drat I wanted to try the wall.

Don't bother with this... blockers are still crashing Uru... Will have to find why.


EDIT:

In Gahreesen, after the update, the client send hundreds of plNotifyMsgs to the server. It seems almost all of them are related to objects called "(s,n)PanelResponder_<number>". Sirius, are these messages really necessary, all of them?

Exactly 171*2=342 for the blockers. OnServerInitComplete asks to reset the panels, and these responders are not fastforwarded and are netforced. Most of the responders in OnServerInitComplete are not fastforwarded anyway. I think I'll update a lot of objects, it seems more and more messy.