Error creating Timelapses #849
-
|
Hi All, I have this problem: all of sudden allsky stopped creating timelapses. I'm using a Raspi 3B+ with a Raspicam HQ and raspbian buster, logs says this: Any idea? |
Beta Was this translation helpful? Give feedback.
Replies: 10 comments 20 replies
-
|
@gorante no idea which version you are using But I also had problems with timelapse before. There was not enough memory for high resolution images try this: pi@allsky:~/allsky $ ./scripts/generateForDay.sh -t 20211206 then try: cat /home/pi/allsky/tmp/timelapseTMP.txt Here you can see the problem: x264 [error]: malloc of size 38600544 failed Fix: resize images |
Beta Was this translation helpful? Give feedback.
-
|
You can also simply change the size of the timelapse video but leave each individual frame the original size. See the TIMELAPSE_VIDEO_X and _Y variables in config/config.sh (I think that the correct names). |
Beta Was this translation helpful? Give feedback.
-
Hi @AndreasLMeg, thanks but the log i've posted is already what i found in the log file. It just stops rendering the timelapse for some reasons. |
Beta Was this translation helpful? Give feedback.
-
|
@gorante what is in the timelapseTMP.txt file? |
Beta Was this translation helpful? Give feedback.
-
|
Sorry for the delay but i wanted to check with a new full-night capture (and still it fails).
This is what allsky.log in /var/log reports instead:
|
Beta Was this translation helpful? Give feedback.
-
|
@gorante that ffmpeg command gives us a lot of headaches. Are you on the Buster or Bullseye operating system? |
Beta Was this translation helpful? Give feedback.
-
Just tried with remove bad images:
I launched generatefortheday manualli thought, but still it crashed: Could be a buster related issue? |
Beta Was this translation helpful? Give feedback.
-
|
@gorante would you please attach (don't copy / paste) your config/config.sh file? Try creating an Allsky runs on Bullseye now, but a lot of people have it running on Buster so it's not a Buster problem. |
Beta Was this translation helpful? Give feedback.
-
|
with I get the same values, but the conversion is OK: I have no clue. Maybe SD or memory problem ? |
Beta Was this translation helpful? Give feedback.
-
|
next (last) try:
|
Beta Was this translation helpful? Give feedback.
-
|
btw: some values are different - (is your system uptodate ?) |
Beta Was this translation helpful? Give feedback.
-
|
hi @AndreasLMeg here it is in order:
console output:
shell output is the same of above, here below the tail output (I see oom-killer there)
Thanks! |
Beta Was this translation helpful? Give feedback.
-
|
Ah, yes the system is up-to-date, I have updated allsky a couple of weeks ago (the timelapse problem was still there, i updated to chek if update would resolve the that) |
Beta Was this translation helpful? Give feedback.
-
|
@gorante it looks like your Pi is running out of memory and / or swap space. Please look in the Wiki for the section on troubleshooting timelapse issues. It describes how to increase swap space. I suggest increasing to 4 GB. Then try to create a timelapse with the "test" directory. If that works, try a real directory. |
Beta Was this translation helpful? Give feedback.
-
Hi @EricClaeys looking at the logfiles i reached the same conclusion (thanks to pointing me to the right ones), I changed swap file size to 2048 (on 32b it is limited to 2048) and it works, but... but it keep killing allsky service. Any time I run the script, i find the service stopped,then i need to restart it via web interface. Also: this cam is up and running from more or less one year, going through updates and stuff, but it has always worked fine. Then it stopped generating thumbnails (not right after and update). Could it be that something changed in RPi memory management? Or some component uses more memory? |
Beta Was this translation helpful? Give feedback.
-
|
@gorante: sorry I have lost the overview. Can you please bring me up to date |
Beta Was this translation helpful? Give feedback.
-
|
@gorante Interesting that the timelapse.sh script stops the allsky service and you have to manually restart it. The allsky.sh command should automatically restart. Next time your run the script, right before doing so, run The next time allsky stops, attach the last 50 lines from /var/log/allsky.sh. I'm not sure why things aren't working now after having worked fine for a year. We likely introduced a bug somewhere, or a change we made has an unintended impact on something else that causes a problem. |
Beta Was this translation helpful? Give feedback.
-
|
The OOM Killer is something very bad as far as I know. Any programs are simply killed by the OS. More information can be found in /var/log/messages (not /var/log/allsky.log) The output of it is probably better to reduce the image size BEFORE creating the timelapse. |
Beta Was this translation helpful? Give feedback.
-
|
made a test with my "old" Raspi Pi Model 3B V1.2, 1GB RAM, 16GB-SD, bullseye
pi@allsky:~/allsky $ cat /proc/meminfo
|
Beta Was this translation helpful? Give feedback.
-
|
I guess that makes sense. If there's not enough memory+swap space the process can't continue and the operating system kills it. Unfortunately we only get a "killed" message and no indication that the process is trying to use more memory+swap than exists. If the resolution of the video is too large, the ffmpeg command tries to allocate more memory to handle the large video, and can't so we get the malloc() failed error. ffmpeg apparently can't use much memory, even if there's a lot of memory on the system. This may be because it's a 32-bit application. I get the "malloc" error on my 8-GB Pi 4 with the RPiHQ camera @ 4096x3040 |
Beta Was this translation helpful? Give feedback.
-
|
@gorante, @AndreasLMeg |
Beta Was this translation helpful? Give feedback.
@gorante, @AndreasLMeg
I did addition testing from what Andreas did, and updated the Wiki page on troubleshooting Timelapse issues.
The "malloc" errors are because the image's resolution is greater than what the libx264 VCODEC supports, whereas the "Kill" errors are due to not enough RAM memory+swap.