over a minute delay between feed and realtime

[Vince Skahan]

I noticed today when the doorbell rang that the motioneye feed from the admin page was well over one minute slow, making it kind of useless as a realtime doorbell camera or the like.  I did a little research and the live stream from the wifi camera itself 'does' have the correct time, and also that the motioneyeos pi3 also 'does' have correct time.  So I can only estimate that that's a massive delay in the motioneyeos admin screen monitor.    I can see two timestamps on the screen, well over a minute apart, yet both 'ticking' and the video moving.

Is this something I should expect with a fast class-10 SD card to a pi3 ?    Would an alternate implementation with a SSD card on the pi, or even using a NAS-like storage on the home LAN help any ?

Love the program, but this one surprised/concerns me a bit.....

[BobH]

I suspect your configuration is overtasking the pi3. Just as a test, try temporarily disabling saving both still images and movie videos and reducing
the image sensor frame rate to something between 1 and 10 fps with lower image resolution (try VGA 640x480) and reduced image quality (in the 40% to 60% range).
If this speeds the display to your satisfaction, then change the configuration back, one item at a time to see how each function loads the camera system.

Another factor is your local area network speed and competing network traffic. File streaming or downloading by any device on your network will reduce the
bandwidth available for your camera. Older home wireless routers often are rated at 11 MBps or 54 MBps network speed but when you read the fine print, may only provide 5 MBps
throughput between devices on the network.

If you report your results, with speed numbers, others can learn from your work.

Best Regards,
Bob

[Vince Skahan]

On Monday, November 6, 2017 at 11:56:19 AM UTC-8, BobH wrote:

If you report your results, with speed numbers, others can learn from your work.

I'm running a wifi camera to a wifi pi3 with a fast SD card on it.   Turning still images off got it from 90 secs behind to 30 seconds behind.   Moving the pi3 to wired (rather than wireless) had no significant effect.  The pi seems unloaded, with a load average a little over 1.00 and lots of free memory.  Top reports the cpu is 25% used.  Don't see anything waiting for i/o.  But it's still waaay behind.

Moving to a Docker container on a 16 GB RAM i5 Intel NUC with a SSD with the same settings results in the two timestamps in the motioneyeos image to match within one second.  So I'm calling it a 'pi is just too slow' thing.  I was going to move to a container anyway.