There are still broken pipe errors #2337
Comments
|
Could you please provide code and medias to reproduce the problem ? |
|
My below error doesn't happen with source from I got the same error with 10 minutes tutorial, it happens when preview every time(on Mac15.3 (24D60)), but it works in notebook with display_in_notebook: Lets import moviepy, lets also import numpy we will use it a some pointfrom moviepy import * ################# VIDEO LOADING################# We load our videovideo = VideoFileClip("./resources/bbb.mp4") ##################### SCENES EXTRACTION##################### We extract the scenes we want to useFirst the charactersintro_clip = video.subclipped(1, 11) ##################### SCENES PREVIEWING##################### Now, lets have a first look at our clipsWarning: you need ffplay installed for preview to workWe set a low fps so our machine can render in real time without slowing downintro_clip.preview(fps=20) Traceback (most recent call last): During handling of the above exception, another exception occurred: Traceback (most recent call last): |
|
I will also get occasional broken pipes: @shared_task(bind=True)
def process_video_moviepy(self, excavation_video_id):
self.update_state(state='PROGRESS', meta={'current': 1, 'total': 4, 'description': 'Start processing'})
try:
excavation_video = ExcavationVideo.objects.get(id=excavation_video_id)
excavation_video.error_message = None
excavation_video.save(update_fields=['error_message'])
target_fps = None
target_bitrate = None
if excavation_video.metadata['fps'] < 25 and excavation_video.metadata['bitrate'] < 1500000:
excavation_video.finished = True
self.update_state(state='PROGRESS', meta={'current': 4, 'total': 4, 'description': 'Saving file'})
excavation_video.save(update_fields=['video', 'finished'])
return {
"status": "success",
"output_video": {
"url": excavation_video.video.url,
"name": excavation_video.video.name,
"size": excavation_video.video.size
}
}
elif excavation_video.metadata['fps'] > 25:
target_fps = 25
elif excavation_video.metadata['bitrate'] > 1500000:
target_bitrate = 1500000
if not excavation_video.video:
raise ValueError("No source video attached")
storage = excavation_video.video.storage
self.update_state(state='PROGRESS', meta={'current': 2, 'total': 4, 'description': 'Loading file'})
with tempfile.TemporaryDirectory() as tmpdir:
tmp_path = Path(tmpdir)
input_path = tmp_path / excavation_video.video.name.split('/')[-1]
output_filename = f"processed_{Path(excavation_video.video.name).stem}.mp4"
output_path = tmp_path / output_filename
# Download source video to temp location
with storage.open(excavation_video.video.name, 'rb') as source_file:
with open(input_path, 'wb') as temp_file:
temp_file.write(source_file.read())
# MoviePy processing
self.update_state(state='PROGRESS', meta={'current': 3, 'total': 4, 'description': 'Processing with MoviePy'})
clip = VideoFileClip(str(input_path))
# Calculate target resolution
original_width, original_height = clip.size
cheeky_width, cheeky_height = video_wh(str(input_path))
if cheeky_width != original_width or cheeky_height != original_height:
original_width, original_height = original_height, original_width
if original_width < original_height: # Portrait
if original_height >= 3840:
max_w, max_h = 720, 1280
else:
max_w, max_h = original_width, original_height
else: # Landscape
if original_width >= 3840:
max_w, max_h = 1280, 720
else:
max_w, max_h = original_width, original_height
ratio = min(max_w / original_width, max_h / original_height)
if ratio < 1: # Only apply scaling if needed
new_w = round(original_width * ratio)
new_h = round(original_height * ratio)
else: # Keep original but ensure even
new_w, new_h = original_width, original_height
# Ensure even dimensions
new_w = new_w // 2 * 2
new_h = new_h // 2 * 2
if original_width >= 3840:
processed = clip.resize(width = new_w)
elif original_height >= 3840:
processed = clip.resize(height = new_h)
else:
processed = clip
if target_fps is not None:
processed = processed.with_fps(25)
else:
pass
# Export with specified parameters
if target_bitrate is not None:
processed.write_videofile(
str(output_path),
codec='libx264',
bitrate='1500k',
preset='fast',
audio_codec='aac',
ffmpeg_params=[
'-movflags', '+faststart',
# '-pix_fmt', 'yuv420p',
],
threads = 1
)
else:
processed.write_videofile(
str(output_path),
codec='libx264', # Explicitly set codec to ensure compatibility
preset='fast',
audio_codec='aac', # Ensure audio is retained
ffmpeg_params=[
'-movflags', '+faststart',
# '-pix_fmt', 'yuv420p',
],
threads = 1
)
clip.close()
processed.close()
del clip, processed
gc.collect()
# if not output_path.exists():
# raise FileNotFoundError("Output video file was not generated")
# Save processed video
with open(output_path, 'rb') as processed_file:
excavation_video.video.save(
output_filename,
File(processed_file)
)
excavation_video.finished = True
self.update_state(state='PROGRESS', meta={'current': 4, 'total': 4, 'description': 'Saving file'})
excavation_video.save(update_fields=['video', 'finished'])
counter = 0
for proc in psutil.process_iter(attrs=['pid', 'name']):
if "ffmpeg" in proc.info['name']:
counter += 1
proc.kill() # Kill stale processes
return {
"status": "success",
"output_video": {
"url": excavation_video.video.url,
"name": excavation_video.video.name,
"size": excavation_video.video.size,
"counter": counter
}
}Tried to mitigate the error, by killing all ffmpeg processes after I do the conversion |
There should be no more broken pipes
There are still broken pipe errors
I come to rescue this FFMPEG error from the broken pipe, because I have noticed when I have been doing tests and deployments that the error occurs at a certain memory usage of the system, the initial system I have is 8GB of RAM and if I pass the 6.5GB of use it throws this same error:
And investigating a little more it turns out that the continuous use of final_video.write_videofile(
str(output_path),
fps=24,
threads=64,
codec='libx264',
audio_codec='aac',
)
The memory is not cleaned correctly to be able to make a continuous execution of my App, so I had to implement an automatic memory cleaning to avoid that, but sometimes I still get the same error because my cleaning functions are not sufficient.
-->
Used medias
Specifications
The text was updated successfully, but these errors were encountered: