Open nopmop opened 5 years ago
It seems I have exactly same problem with kerberos machinery in docker container.
Restarting container does not help, the only thing that works is system reboot.
After the configuration update I get:
kerberos_1 | [OMX.broadcom.video_encode]Wating state to be 3
kerberos_1 | 21/04/2020 10:12:59.509 INFO [trivial] Capture: Succesfully opened Raspberry Pi camera module.
kerberos_1 | 21/04/2020 10:13:04.510 INFO [trivial] Capture: Start streaming thread
kerberos_1 | 21/04/2020 10:13:04.511 INFO [trivial] Stream: Configured stream on port 8889 with quality: 75
kerberos_1 | 21/04/2020 10:13:04.511 INFO [trivial] Starting cloud service: S3
kerberos_1 | 21/04/2020 10:13:04.528 INFO [trivial] Starting conditions: Enabled
kerberos_1 | 21/04/2020 10:13:04.529 INFO [trivial] Starting algorithm: DifferentialCollins
kerberos_1 | 21/04/2020 10:13:04.529 INFO [trivial] Starting expositor: Hull
kerberos_1 | 21/04/2020 10:13:04.923 INFO [trivial] Starting heuristic: Sequence
kerberos_1 | 21/04/2020 10:13:04.924 INFO [trivial] Starting io devices: Video,Disk,MQTT
kerberos_1 | 21/04/2020 10:13:05.833 INFO [trivial] HullExpositor: activity detected from (9,8) to (208,45)
kerberos_1 | 21/04/2020 10:13:09.512 INFO [trivial] Capture: devices is blocking, and not grabbing any more frames.
kerberos_1 | /container-runner.sh: line 96: 32701 Aborted (core dumped) kerberosio
System info:
[2020-04-21 10:16:06][garda] Hardware: Raspberry Pi 3 Model B Plus Rev 1.3
[2020-04-21 10:16:06][garda] Kernel: Linux camback 4.19.66-v7+ #1253 SMP Thu Aug 15 11:49:46 BST 2019 armv7l GNU/Linux
[2020-04-21 10:16:06][garda] OS: PRETTY_NAME="Raspbian GNU/Linux 9 (stretch)"
[2020-04-21 10:16:06][garda] Raspberry version for kerberos: 3
Steps to reproduce:
@ChieftainY2k You didnt had this with the 2.6.1?
@cedricve
Indeed I did have it but I thought it was some system misconfiguration on my part.
After thorough testing I did yesterday it's present on RPI2/3/4 (I was able to test it with RPI4 which I got yesterday , significant upgrade, yay! :-) )
However it does not happen every time (usually after 3rd or 4th configuration update), strange....
@ChieftainY2k You didnt had this with the 2.6.1?
@cedricve Indeed I did have it but I thought it was some system misconfiguration on my part. After thorough testing I did yesterday it's present on RPI2/3/4 (I was able to test it with RPI4 which I got yesterday , significant upgrade, yay! :-) )
However it does not happen every time, strange....
@ChieftainY2k You didnt had this with the 2.6.1?
Cool ;) Does it occur when you dont use the script function?
Cool ;) Does it occur when you dont use the script function?
Unfortunately yes, it does :( here's the config I use:
@cedricve I can open up ssh via the ngrok tunnel for you to look inside the kerberos container to fiddle around and check it out if you want :)
thanks, just wondering if you are using latest version of the Raspbian OS?
I'm using the newest Raspbian Buster with the latest apt-get updates.
thanks, just wondering if you are using latest version of the Raspbian OS?
May be unrelated but sometimes this appears in the /var/log/kern.log (machinery in docker container)
Apr 21 15:15:14 raspberry4 kernel: [ 7739.571559] INFO: task kerberosio:16407 blocked for more than 120 seconds.
Apr 21 15:15:14 raspberry4 kernel: [ 7739.571570] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
Apr 21 15:15:14 raspberry4 kernel: [ 7739.571575] kerberosio D 0 16407 2501 0x00000004
Apr 21 15:15:14 raspberry4 kernel: [ 7739.571606] [<c09c95cc>] (schedule) from [<c09cd5b4>] (schedule_timeout+0x200/0x428)
Apr 21 15:15:14 raspberry4 kernel: [ 7739.571622] [<c09cbb48>] (__down) from [<c0278cd0>] (down+0x54/0x68)
Apr 21 15:15:14 raspberry4 kernel: [ 7739.571643] [<c086d590>] (vchiq_release) from [<c03d66fc>] (__fput+0xa4/0x1e8)
Apr 21 15:15:14 raspberry4 kernel: [ 7739.571661] [<c03d68b0>] (____fput) from [<c02426dc>] (task_work_run+0xbc/0xe0)
Apr 21 15:15:14 raspberry4 kernel: [ 7739.571679] [<c0226bd0>] (do_exit) from [<c0227434>] (do_group_exit+0x4c/0xe4)
Apr 21 15:15:14 raspberry4 kernel: [ 7739.571695] [<c0233780>] (get_signal) from [<c020ca1c>] (do_signal+0x364/0x474)
Apr 21 15:15:14 raspberry4 kernel: [ 7739.571709] [<c020cd38>] (do_work_pending) from [<c0201034>] (slow_work_pending+0xc/0x20)
Apr 21 15:15:14 raspberry4 kernel: [ 7739.571718] ffa0: aa434b40 00000000 00000002 00000000
Apr 21 15:15:14 raspberry4 kernel: [ 7739.571724] ffc0: 00000018 00000000 00000000 000000a2 03b7a338 b0680860 aa434ca8 b06808d8
Apr 21 15:15:14 raspberry4 kernel: [ 7739.571729] ffe0: 00000000 aa434b38 00000000 b633b030 80000010 aa434b40
Apr 21 15:15:14 raspberry4 kernel: [ 7739.571736] INFO: task kworker/2:1:14577 blocked for more than 120 seconds.
Description When I set the option to call a bash script after a motion event
kerberosio
keeps crashing (being respawned bysystemd
). Reverting back the config doesn't work (settings are not stored anymore), and running the process manually shows only the following:After reboot & config revert things start to operate normally again. It looks like after the first crash (maybe happening for trivial reasons) a deadlock is happening because the camera is left in an unclean state.
To Reproduce Steps to reproduce the behavior:
Configuration
->Motion
--- (click next screen * 2) ---> enableSCRIPT
outputExpected behavior Crashes can happen, but a reboot should not be required to fix things, I guess.
Screenshots
Operating System and Device:
Additional context
A strace shows the following: