SARndbox Freeze?

Home Forums AR Sandbox Forum SARndbox Freeze?

This topic contains 9 replies, has 3 voices, and was last updated by  elisek 9 hours, 33 minutes ago.

Viewing 10 posts - 1 through 10 (of 10 total)
  • Author
    Posts
  • #103027

    jKrienert
    Participant

    Running into a screen-freeze error after several min. when the SARndbox is running with active fluid or activity in the box is high. The error requires a restart of the SARndbox application, and consistently occurs after only a few moments.
    When observing the SARndbox shell after the freeze, the following line is printed:

    Caught exception USB::Device:: writeControl: Device has been disconnected

    It seems like a software glitch is causing the SARndbox to detect the Kinect sensor as disconnected, however this is confirmed (at least on the hardware end) as false by immediately restarting the SARndbox application without complication.

    This forum query doesn’t quite match the circumstances in question (including a different OS), but the error message syntax is identical;
    Similar Query

    Has anyone observed a similar problem or is there any insight out there in troubleshooting this?

    #103116

    jKrienert
    Participant

    To assist troubleshooting, here are the specs for the workstation in question;

    Kernel: 3.19.0-32-generic x86_64 (64 bit gcc: 4.8.2)
    Desktop: MATE 1.12.0 (Gtk 3.10.8~8+qiana)
    Distro: Linux Mint 17.3 Rosa
    Machine:   
    System: Dell 
    product: Precision Tower 5810 v-01
    Mobo: Dell 0K240Y v-A01 
    Bios: Dell v-A07 
    date: 04/14/2015
    CPU:       
    	Quad core Intel Xeon E5-1607 v3 (-HT-MCP-) 
    	cache: 10240 KB
    	flags: (lm nx sse sse2 sse3 sse4_1 sse4_2 ssse3 vmx) 
    	bmips: 24740
    	clock speeds: 
    		max: 3100 MHz 
    		1: 1199 MHz 
    		2: 1282 MHz 
    		3: 1217 MHz
    		4: 1397 MHz
    Graphics:  
    	Card: NVIDIA GM204 [GeForce GTX 970] bus-ID: 03:00.0
    	Display Server: X.Org 1.17.1 
    	drivers: nvidia (unloaded: fbdev,vesa,nouveau)
    	Resolution: 800x600@60.3hz
    	GLX Renderer: GeForce GTX 970/PCIe/SSE2
    	GLX Version: 4.5.0 NVIDIA 367.57 Direct Rendering: Yes
    Drives:    
    	HDD Total Size: 508.0GB (2.5% used)
    	ID-1: /dev/sda model: WDC_WD5000AAKX size: 500.1GB
    	ID-2: USB /dev/sdb 

    These appear to meet the suggested components from the ‘sticky’ user guide.

    To elaborate on the problem further;
    The freezing occurs during extensive fluvial simulations, eg when lots of hands are adding water or there is already a large volume of water being simulated.
    I have reset the SARndbox resolution to 640×480 via command line which did not improve the issue.

    Does anyone have any other advice or similar experience with such a consistent freeze error?

    #103122

    Oliver Kreylos
    Moderator

    The initial error message hints at the Kinect’s USB connection not being solid. This could be due to the Kinect itself, the USB cable, the AC adapter / cable, or the PC’s USB host controller. I recommend checking all the connections and trying a different USB port.

    As an experiment, you could run KinectViewer -compress -c 0 and keep it running for a while, and see if that freezes, too. And do the same with the AR Sandbox, with water simulation temporarily disabled via the -ws 0 0 command line argument.

    #103124

    jKrienert
    Participant

    I will pursue these recommended actions and reply with an update withing the next few days.
    Thanks for rapid answers.

    Side (possible related) question;
    Will the Kinect have problems with USB 3.0 vs USB 2.0 ports?

    #103145

    jKrienert
    Participant

    Funny how a few days so easily turns into a few weeks.
    Please pardon the delay.

    @Oliver Kreylos
    Tested the two recommended actions, both of which included periodic arm waving within the sensors view space;
    > KinectViewer -compress -c 0 expressed no freeze or lag for extended runtime (~20 min.)
    > SARndbox… + -ws 0 0 cause the SARndbox to not startup with the following full command string;
    SARndbox -wts 640 480 -uhm -fpv -ws 0 0 -cp ~/.../Control.fifo
    I recently found this posting of similar context in another forum. Of interest is the following claim,

    …or hardware getting bored and switching to standby

    Could this be a possible cause for the lag and freeze? If so, are there any ways to test and mitigate?

    All the physical connections associated with the Kinect appear to be nominal. I suppose the next step is purchasing a new Kinect sensor, USB cables, and USB controller – but impoverished pockets would prefer such as a last resort…

    #103146

    jKrienert
    Participant

    Because the SARndbox program would not start when using the -ws 0 0 command via;

    SARndbox -wts 640 480 -uhm -fpv -ws 0 0 -cp ~/.../Control.fifo

    Is it possible the local fluid calculations are the source of the problem?

    Also, further testing with -wts 320 240 did not seem to help either.
    Detail on this low resolution test;
    When adding sufficient water, the dynamics are calculated for a few seconds without trouble then rapidly increasing frame rate lag freezes the software in 5 – 10 seconds. Closing out requires using the keyboard Alt+Tab combo to exit the SARndbox program via terminal manually rather than Esc in (frozen) full screen context.

    • This reply was modified 3 weeks ago by  jKrienert. Reason: Clarification
    #103170

    jKrienert
    Participant

    Further hardware testing reveals that the Kinect device appears to get reasonably warm during use, especially during the OG freeze error.
    Since the nature of the problem (repetitive software disconnect while all hardware remains connected), inclines me to figure this might be the source of trouble.

    As I understand it, there is a cooling fan inside the unit which should be activated by a thermistor when temps rise.
    At the supposed location of the fans there doesn’t appear to be indication audible running or air movement when the unit warms up simultaneous during expression of the OG freeze error.

    Can anyone chime in on their Kinects temperature characteristics during use, especially when lots of ‘fluid’ is active in the box?
    Also, does such a diagnosis sound plausible and possibly warrant purchase of a new Kinect sensor?

    Thanks.

    #103179

    jKrienert
    Participant

    Update,
    During the freeze problem, the Kinect indicates constant power (constantly blinking green led).
    This would appear to suggest that this is a software issue.

    Will yield on further posting to hear if others might be able to share insights.

    #103217

    jKrienert
    Participant

    Update;
    Problem was software based. Exact issue unknown.
    Reformat HD, reinstall of Linux Mint 18.1, and reinstall of all SARndbox utilities has stopped Ui freeze problems.

    Side note;
    Although it doesn’t appear to affect the SARndbox, the terminal fills with this while running…

    ...
    Ran out of time by 0.00455312
    Ran out of time by 0.00336655
    Ran out of time by 0.00304205
    Ran out of time by 0.0101605
    Ran out of time by 0.0139938
    Ran out of time by 0.0103207
    Ran out of time by 0.00691454
    Ran out of time by 0.00980006
    Ran out of time by 0.0101442
    Ran out of time by 0.0128205
    Ran out of time by 0.0136643
    ...

    Any insight on the cause and if its an immediate concern?

    #103227

    elisek
    Participant

    The “Run out of time” message appears when the water calculation is taking longer than allowed within the software. It won’t actually change the behaviour of the sandbox but it does mean that it may be lagging a little. If you can’t notice a significant difference I wouldn’t worry too much.

Viewing 10 posts - 1 through 10 (of 10 total)

You must be logged in to reply to this topic.

Comments are closed.