We have upgraded our docker image from thumbor 6.4.1 to 6.5.1 yesterday. Both CPU and response time are increasing considerably ever since, that's the only change that happened.

Expected behaviour

Nothing should dramatically change...

Actual behaviour

Both CPU usage and response-time went up considerably.

• Orange - time to fetch
• Blue - p95 of response time
• Yellow dashed - Max response time
• Red - deployment (first ones to upgrade, second one to downgrade)

Once I've downgraded back to 6.4.1 the response time went down again...

• CPU Usage per pod after the deployments...

The spike matches the deployment time, the CPU usage went down over night because of the less intensive traffic but the response time was still higher (1.5x) the one before.

Operating system

Ubuntu 16.04 on K8s

Your thumbor.conf

    import re

    ALLOWED_SOURCES=[
      # https://regexper.com/https://regexper.com/#^https:\/\/*(.+)?hey[.]car\/*(.+)
      re.compile(r'^https://*(.+)?hey[.]car/*(.+)'),
    ]

    MAX_SOURCE_SIZE = 500
    SECURITY_KEY = "REDACTED"
    MAX_WIDTH = 2000
    MAX_HEIGHT = 2000

    # Cache
    MAX_AGE = 365 * 24 * 60 * 60
    MAX_AGE_TEMP_IMAGE = 240

    # Defaults
    QUALITY = 100

    # Engines
    HTTP_LOADER_CURL_ASYNC_HTTP_CLIENT = True

    # Metrics
    METRICS = "thumbor.metrics.statsd_metrics"
    STATSD_PORT = 8125
    STATSD_HOST = "datadog-statsd.kube-system"
    STATSD_PREFIX = "image_proxy_service"

    # GC Options
    GC_INTERVAL = 1800

    # AutoWebp
    AUTO_WEBP = True

    # Sentry
    ERROR_HANDLER_MODULE = "thumbor.error_handlers.sentry"
    SENTRY_DSN = "REDACTED"

Any thoughts on what could've caused this? Is there any new config that I need to take into account when upgrading? I've been reading the changelog and haven't found anything clearly related to that.

We are experiencing memory usage explosion on Ubuntu 14.04.1 LTS with Thumbor v6.0.1 and Pillow 3.2.0. Our memory usage can hit ~97%. Then it will return to ~60% memory usage and after some time it will hit again ~97% memory usage.

After some investigation, there is possility that resize method in thumbor/engines/pil.py is the root cause. If we run gc.collect() in resize method, our memory usage is never hit 97% again.

Thumbor request URL

Sample url http://localhost:7001/unsafe/1920x1080/filename.jpg

Script to reproduce

#! /usr/bin/env ruby

require 'open-uri'

@total = 0
count = 100000

(5000..(5000+count)).each do |i|
  size = 2000 + rand(2000)
  wxh = [size, size].join('x')
  url = "http://localhost:7001/unsafe/#{wxh}/filename.jpg"
  start_time = Time.now
  open url
  end_time = Time.now
  duration = end_time - start_time
  @total += duration
  puts "got: #{url} in #{duration.round(2)}"
end

puts "That took: #{@total.round(2)} seconds"
puts "Average #{(@total / count).round(2)} per image"

Expected behaviour

Memory usage should below 90%

Actual behaviour

Memory usage hit ~97%

Operating system

14.04.1 LTS Thumbor v6.0.1 Pillow 3.2.0

thumbor.conf

MAX_AGE = 31536000
ALLOW_ANIMATED_GIFS = False
LOADER = 'thumbor.loaders.file_loader'

STORAGE = 'thumbor.storages.mixed_storage'
SECURITY_KEY = 'example'

ALLOW_UNSAFE_URL = True
FILE_LOADER_ROOT_PATH = '/vagrant/images'

REDIS_STORAGE_SERVER_HOST = 'example.com'
REDIS_STORAGE_SERVER_PORT = 6379
REDIS_STORAGE_SERVER_DB = 0
REDIS_STORAGE_SERVER_PASSWORD = None

MIXED_STORAGE_CRYPTO_STORAGE = 'tc_redis.storages.redis_storage'
MIXED_STORAGE_DETECTOR_STORAGE = 'tc_redis.storages.redis_storage'

DETECTORS = [
   'thumbor.detectors.face_detector',
   'thumbor.detectors.profile_detector',
   'thumbor.detectors.glasses_detector',
   'thumbor.detectors.feature_detector'
]

I was reading this article http://www.igvita.com/2012/12/18/deploying-new-image-formats-on-the-web/ and found myself thinking that it should be fairly easy for us to support webp format (supported currently by chrome and opera).

Do you guys think we should provide some flag that would make thumbor return the smallest possible image? The only downside is that browsers (as is said on the article) do NOT send proper Accept headers.

We could allow the users to specify what formats they want to support. Thumbor would generate the image in all the formats, then pick the smallest one.

What do you guys think?

Allow the value 'center' to be passed to the filter's parameters x or/and y in order to automatically center the filigrane. ex : watermark(watermark.png,center,center,10) watermark(watermark.png,10,center,10) watermark(watermark.png,center,-10,10)

I don't know what I did wrong but the diff are not showing properly :

•    regex = r'(?:watermark((?P<url>._?),(?P<x>-?[\d]_?),(?P<y>-?[\d]_?),(?P<alpha>[\d]_?)))'
  

•    regex = r'(?:watermark\((?P<url>.*?),(?P<x>(?:-?\d+)|center),(?P<y>(?:-?\d+)|center),(?P<alpha>[\d]*?)\))'
  

•    inv_x = self.x[0] == '-'       
  

•    inv_y = self.y[0] == '-'  
  

•    x, y = int(self.x), int(self.y)  
  

•    center_x = self.x == 'center'
  

•    center_y = self.y == 'center'
  

•    if not center_x :
  

•        inv_x = self.x[0] == '-'
  

•        x = int(self.x)
  

•    if not center_y :
  

•        inv_y = self.y[0] == '-'
  

•        y = int(self.y)
  

•    if inv_x:
  

•        x = (sz[0] - watermark_sz[0]) + x
  

•    if inv_y:
  

•        y = (sz[1] - watermark_sz[1]) + y
  

•    if center_x :
  

•        x = (sz[0] - watermark_sz[0]) /2
  

•    elif inv_x:
  

•        x = (sz[0] - watermark_sz[0]) + x  
  

•    if center_y :
  

•        y = (sz[1] - watermark_sz[1]) /2
  

•    elif inv_y:
  

•        y = (sz[1] - watermark_sz[1]) + y
  

• @filter_method(BaseFilter.String, r'-?[\d]+', r'-?[\d]+', BaseFilter.PositiveNumber, async=True)
• @filter_method(BaseFilter.String, r'(?:-?\d+)|center', r'(?:-?\d+)|center', BaseFilter.PositiveNumber, async=True)

Original:

With trim (wrong / no transparency):

Without trim (transparency = good):

I had the idea maybe this could be the problem: https://github.com/thumbor/thumbor/blob/master/thumbor/transformer.py#L96

Setup: DigitalOcean Machine 1: Ubuntu 14.11 Thumbor 5.05

Machine 2 Ubuntu 15.04 Thumbor 4.12.2

Both machines were with the default config. Only thing changed was the Security key.

supervisord is starting 2 instances of thumber. And nginx is load balancing them. There is nothing else on the machine running.

Every couple of hours. For some reason one of our cores on a 2core machine goes up to 100% and stays that way.

If we dont notice the problem after a while the second core will go up to 100%

And then the images will load slowly.

Looking at logs (debug is enabled) it all looked good. Except in the syslog there there are thousands of these:

Jul 24 18:15:23 packer-ubuntu-15-04-amd64 kernel: [41464.159540] thumbor[25784]: segfault at 1ff1b58 ip 00007f4906cf9783 sp 00007fffe096d3c0 error 6 in libc-2.21.so[7f4906c77000+1c0000]

Any ideas why this is happening.

https://www.dropbox.com/s/ugfvd01vjii8q18/Screenshot%202015-07-25%2021.12.45.png?dl=0

Hello,

Here is a contrib to make thumbor able to store images in Hbase storage (1 row per path) (like at Yfrog for example http://www.slideshare.net/jacque74/hug-hbase-presentation )

Very inspire by the redis storage module.

Hbase is part of the Apache Hadoop project http://hbase.apache.org/ a k/v store column oriented based on HDFS.

it depends on hbase-thrift and thrift (available via pip).

Cheers,

Damien

Hi,

First, let me start by thanking you for this beautiful and elegant solution. We are using Thumbor for last 6 months in production, serving on average 10,000,000 images/hour with zero issues. Now that we know the product is very stable, we would like to optimize the EC2 instances to further enhance Thumbor on processing more images/minute.

We are using a Thumbor stack deployed with AWS CloudFormation and an autoscaling group of c5.2xlarge CentOS7 EC2 instances, fully optimized to take advantage of the 10Gbit network infrastructure AWS offers. All images are cached with Akamai and whatever missed, is cached locally for an hour into a Redis cache.

Current infrastructure (image served with Thumbor🥇): The thumbor-6.4.2 RPM is available into AXIVO repository. We use python-pillow-simd-4.3.0, as is 2.5 times faster compared to python-pillow-4.3.0 and proxy all calls through nginx-1.13.10. This allowed us to reduce the CPU usage by approximately 20%.

While CPU is currently floating around 75% values on each instance, I noticed the memory is partially used. This is the Thumbor configuration we use, on 8 processors and 16GB of RAM:

$ cat /etc/thumbor/thumbor.conf
ALLOW_OLD_URLS = False
ALLOW_UNSAFE_URL = False
AUTO_WEBP = True
DETECTORS = [
	'thumbor.detectors.face_detector',
	'thumbor.detectors.feature_detector',
	'thumbor.detectors.glasses_detector',
	'thumbor.detectors.profile_detector'
]
OPTIMIZERS = [
	'thumbor.optimizers.jpegtran'
]
ENGINE_THREADPOOL_SIZE = 16
JPEGTRAN_PATH = '/bin/jpegtran'
HTTP_LOADER_MAX_CLIENTS = 16384
MAX_AGE = 2592000
MAX_AGE_TEMP_IMAGE = 60
STORAGE = 'tc_redis.storages.redis_storage'
STORAGE_EXPIRATION_SECONDS = 3600
REDIS_STORAGE_IGNORE_ERRORS = True
REDIS_STORAGE_SERVER_HOST = '0001.usw2.cache.amazonaws.com'
REDIS_STORAGE_SERVER_PORT = 6379
REDIS_STORAGE_SERVER_DB = 0
REDIS_STORAGE_SERVER_PASSWORD = None
RESULT_STORAGE = 'tc_redis.result_storages.redis_result_storage'
RESULT_STORAGE_EXPIRATION_SECONDS = 3600
REDIS_RESULT_STORAGE_IGNORE_ERRORS = True
REDIS_RESULT_STORAGE_SERVER_HOST = '0001.usw2.cache.amazonaws.com'
REDIS_RESULT_STORAGE_SERVER_PORT = 6379
REDIS_RESULT_STORAGE_SERVER_DB = 1
REDIS_RESULT_STORAGE_SERVER_PASSWORD = None

$ free -m
              total        used        free      shared  buff/cache   available
Mem:          15309        7604        5871         284        1834        7024
Swap:             0           0           0

$ systemctl --type=service --state=running list-units | grep thumbor
[email protected]             loaded active running Thumbor ([email protected])
[email protected]             loaded active running Thumbor ([email protected])
[email protected]             loaded active running Thumbor ([email protected])
[email protected]             loaded active running Thumbor ([email protected])
[email protected]             loaded active running Thumbor ([email protected])
[email protected]             loaded active running Thumbor ([email protected])
[email protected]             loaded active running Thumbor ([email protected])
[email protected]             loaded active running Thumbor ([email protected])

The ENGINE_THREADPOOL_SIZE value is 2 x nproc.

Are there any other settings I should take advantage of, to optimize the CPU usage and also increase the memory usage? @heynemann, @guilhermef, @cezarsa, your input is greatly appreciated on this matter.

Regards,

Floren Munteanu

Chrome 46 is shipping with support for Client Hints. Specifically, we'll send them on an opt-in basis and can advertise device DPR, viewport width, and resource display width (when it's known) to the server to enable automated delivery and optimization of responsive images.

For additional context on motivation and developer use cases, see: https://developers.google.com/web/updates/2015/09/automating-resource-selection-with-client-hints

It would be great to have CH support in thumbor! :)

After a straightforward 200x200 resize, this black and white GIF image turns into a pixelated mess.

Operation I am attempting: /unsafe/200x200/https://square-production.s3.amazonaws.com/files/547b05546a7fc457dfc17b0c55132119133460c5/original.gif

The result image looks like:

I noticed a huge bottleneck when using the thumbor storages and result storages:

The BaseStorage interface in https://github.com/thumbor/thumbor/blob/master/thumbor/result_storages/init.py#L22

class BaseStorage(object):
    ...
    def put(self, bytes):
        ...

    def get(self):
        ...

supplies only synchronous methods with return values instead of callbacks like the loader interface in https://github.com/thumbor/thumbor/blob/master/thumbor/loaders/file_loader.py

def load(context, path, callback):

(perfectly fine)

https://github.com/thumbor/thumbor/blob/master/thumbor/result_storages/init.py Should look more like

#!/usr/bin/python
# -*- coding: utf-8 -*-
...
import os
from os.path import exists

class BaseStorage(object):
    def __init__(self, context):
        self.context = context

    def put(self, bytes, callback):
        raise NotImplementedError()

    def get(self, callback):
        raise NotImplementedError()

    def last_updated(self, callback):
        raise NotImplementedError()

    def ensure_dir(self, path, callback):
        raise NotImplementedError()

Expected behaviour

Thumbor should process gifs in a faster way.

Actual behaviour

Thumbor is taking too long to process some gifs. So we run some tests using gifsicle (our actual process gifs lib) and pil to compare time processing on both.

Conclusion

We thought, although Gifsicle indeed had a better performance in flips, pil had better processing times in all the others. We don't think the flip filter is heavily used, so the trade off pays itself.

Expected behaviour

Installing thumbor on windows server.

Actual behaviour

UnicodeDecodeError: 'charmap' codec can't decode byte 0x9d in position 148: character maps to

Operating system

Windows server 2016 Standard

Your thumbor.conf

C:\Users\Administrator>pip install thumbor Collecting thumbor Using cached thumbor-7.3.1.tar.gz (29.7 MB) Installing build dependencies ... done Getting requirements to build wheel ... error error: subprocess-exited-with-error

× Getting requirements to build wheel did not run successfully. │ exit code: 1 ╰─> [27 lines of output] Traceback (most recent call last): File "C:\Users\Administrator\AppData\Local\Programs\Python\Python311\Lib\site-packages\pip_vendor\pep517\in_process_in_process.py", line 351, in main() File "C:\Users\Administrator\AppData\Local\Programs\Python\Python311\Lib\site-packages\pip_vendor\pep517\in_process_in_process.py", line 333, in main json_out['return_val'] = hook(**hook_input['kwargs']) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "C:\Users\Administrator\AppData\Local\Programs\Python\Python311\Lib\site-packages\pip_vendor\pep517\in_process_in_process.py", line 118, in get_requires_for_build_wheel return hook(config_settings) ^^^^^^^^^^^^^^^^^^^^^ File "C:\Users\Administrator\AppData\Local\Temp\2\pip-build-env-_5qg5a97\overlay\Lib\site-packages\setuptools\build_meta.py", line 338, in get_requires_for_build_wheel return self._get_build_requires(config_settings, requirements=['wheel']) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "C:\Users\Administrator\AppData\Local\Temp\2\pip-build-env-_5qg5a97\overlay\Lib\site-packages\setuptools\build_meta.py", line 320, in _get_build_requires self.run_setup() File "C:\Users\Administrator\AppData\Local\Temp\2\pip-build-env-_5qg5a97\overlay\Lib\site-packages\setuptools\build_meta.py", line 485, in run_setup self).run_setup(setup_script=setup_script) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "C:\Users\Administrator\AppData\Local\Temp\2\pip-build-env-_5qg5a97\overlay\Lib\site-packages\setuptools\build_meta.py", line 335, in run_setup exec(code, locals()) File "", line 20, in File "C:\Users\Administrator\AppData\Local\Programs\Python\Python311\Lib\pathlib.py", line 1059, in read_text return f.read() ^^^^^^^^ File "C:\Users\Administrator\AppData\Local\Programs\Python\Python311\Lib\encodings\cp1252.py", line 23, in decode return codecs.charmap_decode(input,self.errors,decoding_table)[0] ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ UnicodeDecodeError: 'charmap' codec can't decode byte 0x9d in position 148: character maps to [end of output]

note: This error originates from a subprocess, and is likely not a problem with pip. error: subprocess-exited-with-error

× Getting requirements to build wheel did not run successfully. │ exit code: 1 ╰─> See above for output.

note: This error originates from a subprocess, and is likely not a problem with pip.

How can I install thumbor on windows server? Please help me...

Thumbor request URL

The request url for the issue you are having, you can swap the host name with a fake This image below shows fine. mydomain<<:unsafe/debug/620x620/smart/https://images.healthshots.com/healthshots/en/uploads/2021/09/01154247/ice-on-face-1-770x433.jpg This image doesn't work. mydomain<<:/unsafe/debug/1920x620/smart/https://cdn.filestackcontent.com/pRGFW9N5RsCAJtoEqC9m Also got the Jpeg hosted and passed so it can't be the CDN. Try downloading the image. ( from https://newsroom.notified.com/gul-pr/media/68282/vilda-rosenblad-foto-anna-lena-ahlstromjpg)

Expected behaviour

Tell us what should happen I'd expect smart cropping.

Actual behaviour

Tell us what happens instead When cropped it cuts off the head and doesn't smart crop.

Operating system

This is run from the Docker container provided. https://github.com/thumbor/thumbor/blob/master/docker/Dockerfile with additions. Can provide if like.

Your thumbor.conf

Your config file (WARNING: replace your SECURITY_KEY key with a dummy key or your key will become public)

Thumber.conf `from os.path import expanduser, join home = expanduser("~") PRESERVE_EXIF_INFO = False SEND_IF_MODIFIED_LAST_MODIFIED_HEADERS = False LOADER = "thumbor.loaders.http_loader" MAX_SOURCE_SIZE = 0 UPLOAD_ENABLED = False ENGINE = "thumbor.engines.pil" DETECTORS = [ 'thumbor.detectors.face_detector', 'thumbor.detectors.feature_detector', ]

SECURITY_KEY = "*********" ALLOW_UNSAFE_URL = True HTTP_LOADER_CURL_ASYNC_HTTP_CLIENT = False MAX_AGE = 360000 MAX_AGE_TEMP_IMAGE = 60 MAX_WIDTH = 1920 QUALITY = 70 AUTO_WEBP = True `

------------- Below environment Variables ---------------------

ENV VARS. AWS_STORAGE_ROOT_PATH=storage AUTO_WEBP=True AWS_RESULT_STORAGE_BUCKET_NAME=s3-j****** STORAGE=thumbor_aws.storage AWS_RESULT_STORAGE_S3_SECRET_ACCESS_KEY=************************** RESULT_STORAGE=thumbor_aws.result_storage AWS_STORAGE_BUCKET_NAME=s3-j*********** AWS_RESULT_STORAGE_S3_ACCESS_KEY_ID=**************** ALLOW_UNSAFE_URL=True AWS_STORAGE_S3_ACCESS_KEY_ID=******************** AWS_STORAGE_REGION_NAME=******** LOADER=thumbor.loaders.http_loader AWS_DEFAULT_LOCATION=https://s3-***************** HTTP_LOADER_VALIDATE_CERTS=False AWS_STORAGE_S3_SECRET_ACCESS_KEY=*********************** SECURITY_KEY=**************** AWS_STORAGE_S3_ENDPOINT_URL=https://s3-j**************** AWS_RESULT_STORAGE_ROOT_PATH=storage AWS_RESULT_STORAGE_S3_ENDPOINT_URL=https://s3-************** ALLOW_OLD_URLS=True AWS_RESULT_STORAGE_REGION_NAME=********

Thumbor v7.3.0 (of 01-Nov-2022)

Thumbor doctor will analyze your install and verify if everything is working as expected.

Verifying libraries support...

❎ pycurl is not installed. Thumbor works much better with PyCurl. For more information visit http://pycurl.io/.

❎ cairosvg is not installed. Thumbor uses CairoSVG for reading SVG files. For more information check https://cairosvg.org/.

❎ pyexiv2 is not installed. If you do not need EXIF metadata, you can safely ignore this. For more information visit https://python3-exiv2.readthedocs.io/en/latest/.

Verifying thumbor compiled extensions...

✅ _alpha ✅ _bounding_box ✅ _brightness ✅ _colorize ✅ _composite ✅ _contrast ✅ _convolution ✅ _curve ✅ _equalize ✅ _fill ✅ _nine_patch ✅ _noise ✅ _rgb ✅ _round_corner ✅ _saturation ✅ _sharpen

Verifying extension programs...

✅ jpegtran is installed correctly. ❎ ffmpeg is not installed. Thumbor uses ffmpeg for rendering animated images as GIFV. For more information visit https://www.ffmpeg.org/.

❎ gifsicle is not installed. Thumbor uses gifsicle for better processing of GIF images. For more information visit https://www.lcdf.org/gifsicle/.

Verifying security...

😞 Oh no! We found some things that could improve... 😞

⚠️Warnings⚠️

• pyexiv2 Error Message: No module named 'pyexiv2'

  Error Description: If you do not need EXIF metadata, you can safely ignore this. For more information visit https://python3-exiv2.readthedocs.io/en/latest/.

⛔Errors⛔

• pycurl Error Message: pycurl: libcurl link-time ssl backends (secure-transport, openssl) do not include compile-time ssl backend (none/other)

  Error Description: Thumbor works much better with PyCurl. For more information visit http://pycurl.io/.

• cairosvg Error Message: No module named 'cairosvg'

  Error Description: Thumbor uses CairoSVG for reading SVG files. For more information check https://cairosvg.org/.

• ffmpeg Error Message: Could not find ffmpeg.

  Error Description: Thumbor uses ffmpeg for rendering animated images as GIFV. For more information visit https://www.ffmpeg.org/.

• gifsicle Error Message: Could not find gifsicle.

  Error Description:

The section "Ubuntu/Debian using aptitude (apt-get)" in the documentation seems very outdated, as the PPA hasn't been updated since 2014.

I suggest it be removed.

Thanks

Thumbor request URL

example.com

Expected behaviour

File uploaded to S3 bucket.

Actual behaviour

_RROR ERROR: Traceback (most recent call last): File "/home/thumbor/venv/lib/python3.9/site-packages/tornado/web.py", line 1713, in _execute result = await result File "/home/thumbor/venv/lib/python3.9/site-packages/thumbor/handlers/imaging.py", line 119, in get return await self.check_image(kw) File "/home/thumbor/venv/lib/python3.9/site-packages/thumbor/handlers/imaging.py", line 28, in check_image exists = await self.context.modules.storage.exists( File "/home/thumbor/venv/lib/python3.9/site-packages/tc_aws/storages/s3_storage.py", line 116, in exists file_key = self.storage.get_key(file_abspath) File "/home/thumbor/venv/lib/python3.9/site-packages/boto/s3/bucket.py", line 193, in get_key key, resp = self._get_key_internal(key_name, headers, query_args_l) File "/home/thumbor/venv/lib/python3.9/site-packages/boto/s3/bucket.py", line 231, in get_key_internal raise self.connection.provider.storage_response_error( boto.exception.S3ResponseError: S3ResponseError: 403 Forbidden

Operating system

Base Image. python:3.9.15-slim-bullseye

Your thumbor.conf

HTTP_LOADER_VALIDATE_CERTS = "False" ALLOW_OLD_URLS = "True" ALLOW_UNSAFE_URL = "True" FORWARDED_ALLOWED_IPS = "0.0.0.0" SECURITY_KEY = "******" HTTP_ADDRESS = "0.0.0.0" AUTO_WEBP = "True" QUALITY = "70" MAX_AGE_TEMP_IMAGE = "60" AWS_ACCESS_KEY_ID = "*********" AWS_SECRET_ACCESS_KEY = "************" TC_AWS_REGION = "******" TC_AWS_STORAGE_BUCKET = "s3-bucket-name" TC_AWS_STORAGE_ROOT_PATH = "storage" TC_AWS_RESULT_STORAGE_BUCKET = "s3-bucket-name" TC_AWS_RESULT_STORAGE_ROOT_PATH = "result_storage" STORAGE = "tc_aws.storages.s3_storage" UPLOAD_PHOTO_STORAGE = "tc_aws.storages.s3_storage" COMPATIBILITY_LEGACY_LOADER = "tc_aws.loaders.s3_loader" LOADER = "thumbor.compatibility.loader" RESULT_STORAGE = "tc_aws.result_storages.s3_storage"