Server lock up issues since switching to z-troller - Solved !

[Thrag]

Help!

Recently I've been having issues where my homeseer server will lock up on a semi-regular basis. At first when this started I though it was a heat issue with the mini-itx atom based machine I run homeseer on, but that turned out to be a wild goose chase. I can use prime95 to get the machine hotter than it ever gets in normal operation and it doesn't skip a beat. I noticed that the great majority of my random lock ups happen overnight when the nightly optimization is running. Under further investigation I found that if I fire up z-seer within a few runs of testing the system will lock up. When I trace back the date this started, it was not long after I switched out my z-stick II for a z-troller in the quest for better reliability. So it seems to be heavy activity on the z-troller is the definite cause of my system crashes.



Has anyone had similar lockup issues? Any ideas? Since the Intel Atom D510 board I use does not have a serial port I bought the TU-S9 usb to serial adapter. I'm running all this under windows XP. Could the adapter be the problem? I've never had much luck with the quality of TrendNet's stuff so it wouldn't surprise me if that were the weak link the chain. Of course I don't want to just go blindly replacing components unless I have to.

Any help would be appreciated.

[rmasonjr]

Check to make sure that the USB power management is not turning the device off to conserve power (Device Manager).

Also, check the buffer settings for the COM port:

[Thrag]

Thanks for the reply. I've checked those settings out and it all looks good. No USB power shut off and com buffers are enabled just like the screenshot. What else should I check?

[rmasonjr]

Thats about all I know to check short of some other diagnostics. If its possible, you might try another com port or another converter.

[macromark]

I noticed that the great majority of my random lock ups happen overnight when the nightly optimization is running.

2 suggestions. 1) Don't run ZHealth nightly unless you are making changes to your Z-Wave network daily. 2) Restart your Z-Troller before and after each network optimization. I typically do this myself.

[Thrag]

I already turned off nightly optimizations once I realized I could reliably get the machine to freeze up by running optimizations or tests via z-seer, confirming that is has something to do with lots of controller activity. So it doesn't lock up every few nights as it was with nightly optimization, but I've still had the occasional lockup once a week or so with normal system use.

My most recent lockup looks to be as it was running an event that sends an off command to a long list of devices. It is a command that triggers when I leave the house to turn (almost) everything off. 15 minutes later when I checked the machine remotely from work it was non-responsive. When I arrived home later the machine was locked up (screen displaying but no response from mouse, keyboard, even the numlock light doesn't toggle).

When I have some time to test things out more I'll try resetting before optimizing or running batch tests in zseer to see if it survives the process more reliably.

[macromark]

I've seen Z-Troller lock ups before but not any that have actually locked up a PC keyboard or mouse. Sounds like you may have a hardware problem upstream somewhere.

[chewie]

How does one go about re starting the Z-Troller? Also, is it a good or a bad idea to run the nightly Z-Health that does multiple optimizations? It seems like, based on the advice above, that it would be advisable to turn that feature off?

[macromark]

How does one go about re starting the Z-Troller? Also, is it a good or a bad idea to run the nightly Z-Health that does multiple optimizations? It seems like, based on the advice above, that it would be advisable to turn that feature off?

I never use it. If I make a change to my network, I manually optimize afterwards until everything works properly.

[chewie]

Cool, thanks for the quick response Mark, I will probably just shut it off and optimize several times when I add new nodes then. Maybe turn it on for a night or 2 after adding the new nodes just to be sure that things are still working properly, but leave it off in the long term.

[Thrag]

I've seen Z-Troller lock ups before but not any that have actually locked up a PC keyboard or mouse. Sounds like you may have a hardware problem upstream somewhere.

I think you misunderstood me. The whole machine locks up. The keyboard and mouse are non-responsive because the entire machine is frozen. It's not that just the keyboard stops working but the machine is still running, it's a complete halt. The only time it locks up is when there's lots of communication with the controller. I can stress test the machine all night and day without a hiccup (prime95, various benchmarks, everything I throw at it runs fine), but if I fire up a batch test in z-seer I can crash the machine within 10 minutes. Other USB devices communicate reliably without issues. Homeseer interacting with the z-troller is unquestionably the cause of the lock up as I can reliably reproduce the crash by running things like a full network optimize or testing in z-seer.

[zap]

What build of HS are you using?

[macromark]

I think you misunderstood me. The whole machine locks up. The keyboard and mouse are non-responsive because the entire machine is frozen. It's not that just the keyboard stops working but the machine is still running, it's a complete halt. The only time it locks up is when there's lots of communication with the controller. I can stress test the machine all night and day without a hiccup (prime95, various benchmarks, everything I throw at it runs fine), but if I fire up a batch test in z-seer I can crash the machine within 10 minutes. Other USB devices communicate reliably without issues. Homeseer interacting with the z-troller is unquestionably the cause of the lock up as I can reliably reproduce the crash by running things like a full network optimize or testing in z-seer.

I still suspect something upstream is causing the problem, possibly a driver. If this was a HomeSeer/Z-Troller issue, we'd be getting more reports of this and we're not. I see you're using the Trendnet converter... we've have very good luck with that one so far. If you haven't already, check Trendnet's website to see if you have the latest driver.

[Thrag]

What build of HS are you using?

The latest 2.5.0.23.

[zap]

I would move back to .20 or .21 and see if you still have problems.
Hs changed the way they handled com in .23