BUG: Netcam code in 3.1.18 is not stable

When you start Motion and have a netcam - even with single jpegs - Motion often stops right away.

Test case

I have used your URL and I can for sure say that it is not the URL or your camera which is the problem.

Something is rotten in the new Motion netcam code.

I can reproduce your error. I can even reproduce it when I connect to a jpg file on my own webserver.

I have made some tools to get better control of the experiments.

I have a C program that I use to connect to the Netcam and save the raw data that it returns in a file.

And I have made a small perl program which can send data via my Apache server (nph- prefixed to control all headers).

They are all packed here: http://www.lavrsen.dk/twiki/bin/view/Motion/MjpegSniffer

The result is that Motion will fail often but not all the time with a jpg from the Toshiba cam. It works better with a jpeg from my own Motion server but even this fails.

When Motion fails it fails within the first few seconds. Once it runs - it runs stable for many minutes until I kill it.

Running with daemon off here are two scenarios.

First success

[root@merlin mjprox]# motion
Processing thread 0 - config file /usr/local/etc/motion.conf
Processing thread 1 - config file /usr/local/etc/thread1.conf
Processing thread 2 - config file /usr/local/etc/thread2.conf
Processing thread 3 - config file /usr/local/etc/thread3.conf
Processing thread 4 - config file /usr/local/etc/thread4.conf
Processing thread 5 - config file /usr/local/etc/thread5.conf
Processing thread 6 - config file /usr/local/etc/thread7.conf
Processing thread 7 - config file /usr/local/etc/thread8.conf
Processing thread 8 - config file /usr/local/etc/thread9.conf
Webcam port for thread 1 is 8081
Webcam port for thread 2 is 8082
Webcam port for thread 3 is 8083
Webcam port for thread 4 is 8084
Webcam port for thread 5 is 8085
Webcam port for thread 6 is 8087
Webcam port for thread 7 is 8088
Webcam port for thread 8 is 8089
Thread1 device: http://www.lavrsen.dk/cgi-bin/nph-rawparse.pl input: -1
Thread2 device: /dev/video0 input: 1
Thread3 device: /dev/video1 input: 1
Thread4 device: /dev/video4 input: 8
Thread5 device: /dev/video3 input: 8
Netcam: thread 1: starting...
Thread6 device: /dev/video7 input: 8
Thread7 device: /dev/video2 input: 8
Thread8 device: /dev/video6 input: 8
[HTTP/1.0 200 OK]
Netcam: jpeg file
Netcam: thread 1: thread PID: 11626
0 0 0

And now failure.

[root@merlin mjprox]# motion
Processing thread 0 - config file /usr/local/etc/motion.conf
Processing thread 1 - config file /usr/local/etc/thread1.conf
Processing thread 2 - config file /usr/local/etc/thread2.conf
Processing thread 3 - config file /usr/local/etc/thread3.conf
Processing thread 4 - config file /usr/local/etc/thread4.conf
Processing thread 5 - config file /usr/local/etc/thread5.conf
Processing thread 6 - config file /usr/local/etc/thread7.conf
Processing thread 7 - config file /usr/local/etc/thread8.conf
Processing thread 8 - config file /usr/local/etc/thread9.conf
Webcam port for thread 1 is 8081
Webcam port for thread 2 is 8082
Webcam port for thread 3 is 8083
Webcam port for thread 4 is 8084
Webcam port for thread 5 is 8085
Webcam port for thread 6 is 8087
Webcam port for thread 7 is 8088
Webcam port for thread 8 is 8089
Thread1 device: http://www.lavrsen.dk/cgi-bin/nph-rawparse.pl input: -1
Thread2 device: /dev/video0 input: 1
Thread3 device: /dev/video1 input: 1
Thread4 device: /dev/video4 input: 8
Thread5 device: /dev/video3 input: 8
Thread6 device: /dev/video7 input: 8
Thread7 device: /dev/video2 input: 8
Thread8 device: /dev/video6 input: 8
Netcam: thread 1: starting...
[HTTP/1.0 200 OK]
Netcam: jpeg file
Netcam : Image size 0

So there is something timing wise that often goes wrong when the new netcam code creates the netcam thread.

Environment

Motion version:	3.1.18
ffmpeg version:	rpm 0.4.9-2004110
Shared libraries:	xmlrpc, ffmpeg, mysql, postgresql
Server OS:	FC2 with custom 2.6.10

-- KennethLavrsen - 01 Feb 2005

Follow up

I've a patch to fix this bug , i'll release as soon as i verify that there's not race condictions with any camera.

-- AngelCarpintero - 06 Feb 2005

Fix record

Fixed with post-3 patch , reopen if errors remains.

-- AngelCarpintero - 08 Feb 2005

---

Status changed to Released. The fix is in 3.1.19 (and 3.2.1_snaps)