openRTSP dead loop problem

[matus2]

Hello, 

we are using openRTSP tool to record IP cameras stream into file. 
Recently we have discovered, that sometimes ( maeby because of some 
network issues ? ) some of the processes stucks at 100% cpu and stops 
recording. 
We are running multiple instances of openRTSP. I have not found the real 
cause of this behavior, sometimes stucks one process per day, sometimes 
10 per day... 
Cameras are some low cost ebay style linux devices, running rtsp server 
only via TCP. 

I am running self compiled version live555-2019.03.06, on Centos7. 
I have built debug symbols package so i can somehow trace this loop with 
gdb if needed 

Command, which I am running for recording is 
/usr/bin/openRTSP -4 -V -t -w 1920 -h 1080 -P 900 -b 500000 -F /records/ 
-f 15 -D 5 rtsp://usename:password@192.168.1.5:80/0 


If i attach strace to the stucked process i can see calling only two 
syscalls calls forever: 
recvfrom(-1, 0x1953260, 65520, 0, 0x7ffc9866bc10, 0x7ffc9866baa4) = -1 
EBADF (Bad file descriptor) 
select(7, [3], [], [3], {0, 2274}) = 1 (in [3], left {0, 2273}) 
recvfrom(-1, 0x1953260, 65520, 0, 0x7ffc9866bc10, 0x7ffc9866baa4) = -1 
EBADF (Bad file descriptor) 
select(7, [3], [], [3], {0, 2215}) = 1 (in [3], left {0, 2214}) 
recvfrom(-1, 0x1953260, 65520, 0, 0x7ffc9866bc10, 0x7ffc9866baa4) = -1 
EBADF (Bad file descriptor) 
select(7, [3], [], [3], {0, 2156}) = 1 (in [3], left {0, 2155}) 
. 
. 
. 


Do you have any ideas why this loop happens, and why process is NOT 
killed when no data were received after few seconds? (-d, -D switches)? 
Because I am not very familliar with internal structure of this tool maeby 
I can provide gdb logs for you. Any help appreciated! 

Regards, 
matus2 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.live555.com/pipermail/live-devel/attachments/20190511/6ccd31b2/attachment.htm>