Opened 2 years ago

Last modified 8 months ago

#9697 new defect

ffmpeg git master have isuss dash getting higher cpu each hour

Reported by: Youssef Owned by:
Priority: normal Component: undetermined
Version: git-master Keywords:
Cc: Youssef Blocked By:
Blocking: Reproduced by developer: no
Analyzed by developer: no

Description (last modified by Youssef)

ffmpeg Version git master have problem with mpd to restream it cpu became after each hour higher, there any solution for it ??

here command ffmpeg :

ffmpeg -re -i https://demo.unified-streaming.com/k8s/live/live-low-latency.isml/.mpd -c:v copy -c:a copy -sn -shortest -strict -2 -max_muxing_queue_size 400

i mean this mpd demo test is live when i restream it , cpu pid ffmpeg start 2.0 % after 1 hour be 4.0 % and each hour became higher for exmple 13 hours uptime stream became pid ffmpeg cpu 80 %, and there just copy to copy not transcoder that is strange only mpd protocol, i hope you understood what i mean !

Change History (18)

comment:1 by Youssef, 2 years ago

Cc: Youssef removed
Version: 4.4.1git-master

comment:2 by Youssef, 2 years ago

Cc: Youssef added
Description: modified (diff)
Summary: ffmpeg4 have isuss dash getting higher cpu each hourffmpeg git master have isuss dash getting higher cpu each hour

comment:3 by Marton Balint, 22 months ago

Priority: importantnormal

Not a crash or a regression, so does not qualify as important.

comment:4 by Carl Eugen Hoyos, 20 months ago

Analyzed by developer: unset
Reproduced by developer: unset

Please provide the command line you tested together with the complete, uncut console output to make this a valid ticket.

comment:5 by Youssef, 20 months ago

command is here :

ffmpeg -re -i https://demo.unified-streaming.com/k8s/live/live-low-latency.isml/.mpd -c:v copy -c:a copy -sn -shortest -strict -2 -max_muxing_queue_size 400 -f hls -hls_allow_cache 0 -hls_flags delete_segments -hls_time 6 -hls_list_size 10 output.m3u8

and every hour cpu usage get higher and my server ubuntu 18.04 and config server is Xeon E5-2640 32G ram, but after each hour cpu get higher until be cpu after 80 % for 1 stream on this machine is strange mpd do like this!

in reply to:  4 comment:6 by Youssef, 20 months ago

Replying to Carl Eugen Hoyos:

Please provide the command line you tested together with the complete, uncut console output to make this a valid ticket.

command is here :

ffmpeg -re -i https://demo.unified-streaming.com/k8s/live/live-low-latency.isml/.mpd -c:v copy -c:a copy -sn -shortest -strict -2 -max_muxing_queue_size 400 -f hls -hls_allow_cache 0 -hls_flags delete_segments -hls_time 6 -hls_list_size 10 output.m3u8

and every hour cpu usage get higher and my server ubuntu 18.04 and config server is Xeon E5-2640 32G ram, but after each hour cpu get higher until be cpu after 80 % for 1 stream on this machine is strange mpd do like this!

comment:7 by Youssef, 20 months ago

Analyzed by developer: set
Reproduced by developer: set

command is here :

ffmpeg -re -i https://demo.unified-streaming.com/k8s/live/live-low-latency.isml/.mpd -c:v copy -c:a copy -sn -shortest -strict -2 -max_muxing_queue_size 400 -f hls -hls_allow_cache 0 -hls_flags delete_segments -hls_time 6 -hls_list_size 10 output.m3u8

and every hour cpu usage get higher and my server ubuntu 18.04 and config server is Xeon E5-2640 32G ram, but after each hour cpu get higher until be cpu after 80 % for 1 stream on this machine is strange mpd do like this!

comment:8 by Youssef, 20 months ago

command is here :

ffmpeg -re -i https://demo.unified-streaming.com/k8s/live/live-low-latency.isml/.mpd -c:v copy -c:a copy -sn -shortest -strict -2 -max_muxing_queue_size 400 -f hls -hls_allow_cache 0 -hls_flags delete_segments -hls_time 6 -hls_list_size 10 output.m3u8

and every hour cpu usage get higher and my server ubuntu 18.04 and config server is Xeon E5-2640 32G ram, but after each hour cpu get higher until be cpu after 80 % for 1 stream on this machine is strange mpd do like this!

comment:9 by Carl Eugen Hoyos, 20 months ago

Please provide the command line you tested together with the complete, uncut console output to make this a valid ticket.

comment:10 by Carl Eugen Hoyos, 20 months ago

Analyzed by developer: unset
Reproduced by developer: unset

comment:11 by mkver, 19 months ago

Your mpd definitely triggered a bug in our dash demuxer. I sent a patch for it here. Can you test whether this fixes the slowdown you experience?

in reply to:  11 comment:12 by Youssef, 19 months ago

ffmpeg -re -v info -i https://demo.unified-streaming.com/k8s/live/live-low-latency.isml/.mpd -f null -

Error : No such file or directory

Last edited 19 months ago by Youssef (previous) (diff)

comment:13 by mkver, 19 months ago

If you simply copy the command line that you posted in comment 8, you will also copy some fields from after -i and before https due to this being a link. You need to delete them (or enter the url manually).

comment:14 by mkver, 19 months ago

I now applied the patches I linked to. Can you please test whether this fixes the cpu load issue, too?

comment:15 by Youssef, 19 months ago

command : ffmpeg -re -y -report -v info -i https://demo.unified-streaming.com/k8s/live/live-low-latency.isml/.mpd -c:a copy -c:v copy -f mp4 test.mp4

same issue high cpu each hour and leak memory take 1902 mb form RAM,
if you want me upload report file

Last edited 19 months ago by Youssef (previous) (diff)

comment:16 by balooshy, 11 months ago

high cpu issue remain same for any dash mpd url any patch update for this ticket ?

Last edited 11 months ago by balooshy (previous) (diff)

comment:17 by balooshy, 11 months ago

Priority: normalcritical

comment:18 by Leo Izen, 8 months ago

Priority: criticalnormal
Note: See TracTickets for help on using tickets.