Help desperatly needed - my Server's unstable and I don't know what to do!

[Mastiff]

I have a main server in my house that controls just about everything. I had a few small problems with the old setup, so I changed motherboard and installed everything anew a couple of months back. But I have one very big problem: Girder can suddenly stop responding. It doesn't receive any events, and I can't fire anything manually from it. I can still expand and collapse nodes in the GML, but if I try to open the settings, it freezes. I have to kill the process to exit. The problem is that the hardware from the old server is not possible to put back, and I really need to find out what's causing this! I have diabled everything except for the totally essential plug-ins. Unfortunately a lot of them are essential. I can rule out the Phidgets because wen I moved them to another computer, I still got the lockups.

The essential plug-ins that I can't stop using without killing my house (literally - wakeup calls, light control, changing channels on the digital dish a lot of other stuff depends on this) are these:
Comserver (which I suspect may be the actual problem)
Generic serial (and a caller ID modem that tells me if the office calls when I'm out)
Scheduler (used for wake up calls- totally essential)
Send message (control of JRMC)
Simple timer (part of many things in my setup)
Tree script
USB-UIRT (main control method from all rooms where I don't use NetRemote)
X10-CM1X (running an X10 CM-11A that controls a simple HVAC system and lights in the home theater)

I have even tried adding an action that reboots the server every night, but it's not enough. Girder will freeze up during the day. What can I do to find this? Please help me!

[Rob H]

Can you see if there are any errors in DebugView?

[Rob H]

Oh, and which version of Girder is this?

[Mastiff]

4.09, which I have had best results on (before things started to trickle down from 5). And what do I need to do to debugview to see Lua stuff going on? I don't see it.

[Rob H]

You should still see error (and other internal) messages from the Girder process.

Might be worth trying a later version of Girder though, that's about 6 releases behind the current version, there are a couple of DUI updates/fixes in later versions.

[Mastiff]

Well, the reason I tried to go back to version 9 (I originally installed the latest version) was the same error. I will try another run of Debugviiew, but furst I have to check out a sneaking suspicion: I may have found the error in another part of the server - it was set up as a DNS server, which it shouldn't be because I don't have a fully qualified domain name... I'm gonna let it go for a few days and see. Can network problems like that mess up the comserver?

[Mastiff]

Well, it wasn't the DNS, but there is something funky going on somewhere. Another clue may be that when the lockup happens the only thing that shows in the logger (even if I try to manually trigger anything and it doesn't work) is clients logging on and off. And when it happens, I get a message from any NetRemote trying to log on that says that there is a client, but it doesn't seem to run Girder 4. I'm wondering if the external network (that's directly connected to the Internet, and where I have other people sharing the connection with me) is messing up things, so I would like to isolate Girder to the internal network. But how? I started a thread about that.

[Mastiff]

Just to wrap up this thread: The firewall fixed it. Period. There's something going on either because I have two network cards in one computer or because there were malformed packages or whatever on my external network. No matter what, I have a stable Girder, at last! :)

[Ron]

That's Awesome

[Mastiff]

Just can't let this thread lie, in case somebody should stumble over it in the future: I upgraded to the latest Girder 4 right after this previous message, and the firewall needed new settings since the exefile had changed. Accidentally I opened for everything on the external network as well, and for two days I was back to the same problems as before. But I discovered the error in the firewall settings and corrected those. It's been running continuously for a week since, and no problems at all. And now I don't have the previous problems on my external network. Maybe it's just that the comserver doesn't really like running with two network cards on different series?