Aws::S3::Errors::BadDigest: The CRC32 you specified did not match the calculated checksum

[sato11]

Describe the bug

As soon as I upgraded aws-sdk-s3 to 1.178.0 and higher I start to observe the error mentioned in the title.

I think it's related to #3166, where I'm ok with the change itself but I'm having a difficulty to actually migrate and the error I'm getting is rather unique (meaning I hardly find the same error online) so allow me to raise a new issue.

I am using aws-sdk-s3 as the backend for Active Storage 7.2 and "ActiveStorage::IntegrityError" is the abstracted version of the error.

Stacktrace

Aws::S3::Errors::BadDigest: The CRC32 you specified did not match the calculated checksum. (Aws::S3::Errors::BadDigest)
    from seahorse/client/plugins/raise_response_errors.rb:17:in 'call'
    from aws-sdk-s3/plugins/sse_cpk.rb:24:in 'call'
    from aws-sdk-s3/plugins/dualstack.rb:21:in 'call'
    from aws-sdk-s3/plugins/accelerate.rb:43:in 'call'
    from aws-sdk-core/plugins/checksum_algorithm.rb:169:in 'call'
    from aws-sdk-core/plugins/jsonvalue_converter.rb:16:in 'call'
    from aws-sdk-core/plugins/invocation_id.rb:16:in 'call'
    from aws-sdk-core/plugins/idempotency_token.rb:19:in 'call'
    from aws-sdk-core/plugins/param_converter.rb:26:in 'call'
    from seahorse/client/plugins/request_callback.rb:89:in 'call'
    from aws-sdk-core/plugins/response_paging.rb:12:in 'call'
    from seahorse/client/plugins/response_target.rb:24:in 'call'
    from aws-sdk-core/plugins/telemetry.rb:39:in 'block in call'
    from aws-sdk-core/telemetry/no_op.rb:29:in 'in_span'
    from aws-sdk-core/plugins/telemetry.rb:53:in 'span_wrapper'
    from aws-sdk-core/plugins/telemetry.rb:39:in 'call'
    from seahorse/client/request.rb:72:in 'send_request'
    from aws-sdk-s3/client.rb:18639:in 'put_object'
    from aws-sdk-s3/object.rb:3082:in 'block in put'
    from aws-sdk-core/plugins/user_agent.rb:92:in 'metric'
    from aws-sdk-s3/object.rb:3081:in 'put'
    from active_storage/service/s3_service.rb:134:in 'upload_with_single_part'
    from active_storage/service/s3_service.rb:33:in 'block in upload'
    from active_support/notifications.rb:210:in 'block in instrument'
    from active_support/notifications/instrumenter.rb:58:in 'instrument'
    from active_support/notifications.rb:210:in 'instrument'
    from active_storage/service.rb:165:in 'instrument'
    from active_storage/service/s3_service.rb:29:in 'upload'
    from /home/ac-user/tsunaren/shared/bundle/ruby/3.3.0/gems/activestorage-7.2.2.2/app/models/active_storage/blob.rb:273:in 'upload_without_unfurling'
    from active_storage/attached/changes/create_one.rb:26:in 'upload'
    from active_support/core_ext/object/try.rb:15:in 'public_send'
    from active_support/core_ext/object/try.rb:15:in 'try'
    from active_storage/attached/model.rb:142:in 'block in has_one_attached'
    from active_support/callbacks.rb:407:in 'instance_exec'
    from active_support/callbacks.rb:407:in 'block in make_lambda'
    from active_support/callbacks.rb:208:in 'call'
    from active_support/callbacks.rb:563:in 'block in invoke_after'
    from active_support/callbacks.rb:563:in 'each'
    from active_support/callbacks.rb:563:in 'invoke_after'
    from active_support/callbacks.rb:111:in 'run_callbacks'
    from active_support/callbacks.rb:913:in '_run_commit_callbacks'
    from active_record/transactions.rb:385:in 'committed!'
    from active_record/connection_adapters/abstract/transaction.rb:307:in 'block in commit_records'
    from active_record/connection_adapters/abstract/transaction.rb:344:in 'run_action_on_records'
    from active_record/connection_adapters/abstract/transaction.rb:306:in 'commit_records'
    from active_record/connection_adapters/abstract/transaction.rb:596:in 'block in commit_transaction'
    from active_support/concurrency/null_lock.rb:9:in 'synchronize'
    from active_record/connection_adapters/abstract/transaction.rb:584:in 'commit_transaction'
    from active_record/connection_adapters/abstract/transaction.rb:628:in 'block in within_new_transaction'
    from active_support/concurrency/null_lock.rb:9:in 'synchronize'
    from active_record/connection_adapters/abstract/transaction.rb:613:in 'within_new_transaction'
    from active_record/connection_adapters/abstract/database_statements.rb:361:in 'transaction'
    from active_record/transactions.rb:414:in 'block in with_transaction_returning_status'
    from active_record/connection_adapters/abstract/connection_pool.rb:421:in 'with_connection'
    from active_record/connection_handling.rb:296:in 'with_connection'
    from active_record/transactions.rb:410:in 'with_transaction_returning_status'
    from active_record/transactions.rb:362:in 'save'
    from active_record/suppressor.rb:52:in 'save'
    from active_storage/attached/one.rb:61:in 'attach'
    from app/jobs/operator/survey_export_csv_job.rb:31:in 'perform'
    from active_job/execution.rb:68:in 'block in _perform_job'
    from active_support/callbacks.rb:121:in 'block in run_callbacks'
    from i18n.rb:353:in 'with_locale'
    from active_job/translation.rb:9:in 'block (2 levels) in <module:Translation>'
    from active_support/callbacks.rb:130:in 'instance_exec'
    from active_support/callbacks.rb:130:in 'block in run_callbacks'
    from active_support/core_ext/time/zones.rb:65:in 'use_zone'
    from active_job/timezones.rb:9:in 'block (2 levels) in <module:Timezones>'
    from active_support/callbacks.rb:130:in 'instance_exec'
    from active_support/callbacks.rb:130:in 'block in run_callbacks'
    from active_support/callbacks.rb:141:in 'run_callbacks'
    from active_job/execution.rb:67:in '_perform_job'
    from active_job/instrumentation.rb:32:in '_perform_job'
    from active_job/execution.rb:51:in 'perform_now'
    from active_job/instrumentation.rb:26:in 'block in perform_now'
    from active_record/railties/job_runtime.rb:13:in 'block in instrument'
    from active_job/instrumentation.rb:40:in 'block in instrument'
    from active_support/notifications.rb:210:in 'block in instrument'
    from active_support/notifications/instrumenter.rb:58:in 'instrument'
    from active_support/notifications.rb:210:in 'instrument'
    from active_job/instrumentation.rb:39:in 'instrument'
    from active_record/railties/job_runtime.rb:11:in 'instrument'
    from active_job/instrumentation.rb:26:in 'perform_now'
    from active_job/logging.rb:32:in 'block in perform_now'
    from active_support/tagged_logging.rb:138:in 'block in tagged'
    from active_support/tagged_logging.rb:38:in 'tagged'
    from active_support/tagged_logging.rb:138:in 'tagged'
    from active_support/broadcast_logger.rb:241:in 'method_missing'
    from active_job/logging.rb:39:in 'tag_logger'
    from active_job/logging.rb:32:in 'perform_now'
    from sentry/rails/active_job.rb:10:in 'perform_now'
    from active_job/execution.rb:29:in 'block in execute'
    from active_support/callbacks.rb:121:in 'block in run_callbacks'
    from active_job/railtie.rb:79:in 'block (4 levels) in <class:Railtie>'
    from active_support/reloader.rb:77:in 'block in wrap'
    from active_support/execution_wrapper.rb:87:in 'wrap'
    from active_support/reloader.rb:74:in 'wrap'
    from active_job/railtie.rb:78:in 'block (3 levels) in <class:Railtie>'
    from active_support/callbacks.rb:130:in 'instance_exec'
    from active_support/callbacks.rb:130:in 'block in run_callbacks'
    from active_support/callbacks.rb:141:in 'run_callbacks'
    from active_job/execution.rb:27:in 'execute'
    from active_job/queue_adapters/sidekiq_adapter.rb:70:in 'perform'
    from sidekiq/processor.rb:202:in 'execute_job'
    from sidekiq/processor.rb:170:in 'block (2 levels) in process'
    from sidekiq/middleware/chain.rb:177:in 'block in invoke'
    from sidekiq/history/middleware.rb:54:in 'call'
    from sidekiq/middleware/chain.rb:179:in 'block in invoke'
    from sentry/sidekiq/sentry_context_middleware.rb:69:in 'call'
    from sidekiq/middleware/chain.rb:179:in 'block in invoke'
    from sidekiq/failures/middleware.rb:9:in 'call'
    from sidekiq/middleware/chain.rb:179:in 'block in invoke'
    from sidekiq/middleware/chain.rb:182:in 'invoke'
    from sidekiq/processor.rb:169:in 'block in process'
    from sidekiq/processor.rb:136:in 'block (6 levels) in dispatch'
    from sidekiq/job_retry.rb:113:in 'local'
    from sidekiq/processor.rb:135:in 'block (5 levels) in dispatch'
    from sidekiq/rails.rb:14:in 'block in call'
    from active_support/reloader.rb:77:in 'block in wrap'
    from active_support/execution_wrapper.rb:91:in 'wrap'
    from active_support/reloader.rb:74:in 'wrap'
    from sidekiq/rails.rb:13:in 'call'
    from sidekiq/processor.rb:131:in 'block (4 levels) in dispatch'
    from sidekiq/processor.rb:263:in 'stats'
    from sidekiq/processor.rb:126:in 'block (3 levels) in dispatch'
    from sidekiq/job_logger.rb:13:in 'call'
    from sidekiq/processor.rb:125:in 'block (2 levels) in dispatch'
    from sidekiq/job_retry.rb:80:in 'global'
    from sidekiq/processor.rb:124:in 'block in dispatch'
    from sidekiq/job_logger.rb:39:in 'prepare'
    from sidekiq/processor.rb:123:in 'dispatch'
    from sidekiq/processor.rb:168:in 'process'
    from sidekiq/processor.rb:78:in 'process_one'
    from sidekiq/processor.rb:68:in 'run'
    from sidekiq/component.rb:8:in 'watchdog'
    from sidekiq/component.rb:17:in 'block in safe_thread'

ActiveStorage::IntegrityError: ActiveStorage::IntegrityError (ActiveStorage::IntegrityError)
    from active_storage/service/s3_service.rb:136:in 'rescue in upload_with_single_part'
    from active_storage/service/s3_service.rb:133:in 'upload_with_single_part'
    from active_storage/service/s3_service.rb:33:in 'block in upload'
    from active_support/notifications.rb:210:in 'block in instrument'
    from active_support/notifications/instrumenter.rb:58:in 'instrument'
    from active_support/notifications.rb:210:in 'instrument'
    from active_storage/service.rb:165:in 'instrument'
    from active_storage/service/s3_service.rb:29:in 'upload'
    from /home/ac-user/tsunaren/shared/bundle/ruby/3.3.0/gems/activestorage-7.2.2.2/app/models/active_storage/blob.rb:273:in 'upload_without_unfurling'
    from active_storage/attached/changes/create_one.rb:26:in 'upload'
    from active_support/core_ext/object/try.rb:15:in 'public_send'
    from active_support/core_ext/object/try.rb:15:in 'try'
    from active_storage/attached/model.rb:142:in 'block in has_one_attached'
    from active_support/callbacks.rb:407:in 'instance_exec'
    from active_support/callbacks.rb:407:in 'block in make_lambda'
    from active_support/callbacks.rb:208:in 'call'
    from active_support/callbacks.rb:563:in 'block in invoke_after'
    from active_support/callbacks.rb:563:in 'each'
    from active_support/callbacks.rb:563:in 'invoke_after'
    from active_support/callbacks.rb:111:in 'run_callbacks'
    from active_support/callbacks.rb:913:in '_run_commit_callbacks'
    from active_record/transactions.rb:385:in 'committed!'
    from active_record/connection_adapters/abstract/transaction.rb:307:in 'block in commit_records'
    from active_record/connection_adapters/abstract/transaction.rb:344:in 'run_action_on_records'
    from active_record/connection_adapters/abstract/transaction.rb:306:in 'commit_records'
    from active_record/connection_adapters/abstract/transaction.rb:596:in 'block in commit_transaction'
    from active_support/concurrency/null_lock.rb:9:in 'synchronize'
    from active_record/connection_adapters/abstract/transaction.rb:584:in 'commit_transaction'
    from active_record/connection_adapters/abstract/transaction.rb:628:in 'block in within_new_transaction'
    from active_support/concurrency/null_lock.rb:9:in 'synchronize'
    from active_record/connection_adapters/abstract/transaction.rb:613:in 'within_new_transaction'
    from active_record/connection_adapters/abstract/database_statements.rb:361:in 'transaction'
    from active_record/transactions.rb:414:in 'block in with_transaction_returning_status'
    from active_record/connection_adapters/abstract/connection_pool.rb:421:in 'with_connection'
    from active_record/connection_handling.rb:296:in 'with_connection'
    from active_record/transactions.rb:410:in 'with_transaction_returning_status'
    from active_record/transactions.rb:362:in 'save'
    from active_record/suppressor.rb:52:in 'save'
    from active_storage/attached/one.rb:61:in 'attach'
    from app/jobs/operator/survey_export_csv_job.rb:31:in 'perform'
    from active_job/execution.rb:68:in 'block in _perform_job'
    from active_support/callbacks.rb:121:in 'block in run_callbacks'
    from i18n.rb:353:in 'with_locale'
    from active_job/translation.rb:9:in 'block (2 levels) in <module:Translation>'
    from active_support/callbacks.rb:130:in 'instance_exec'
    from active_support/callbacks.rb:130:in 'block in run_callbacks'
    from active_support/core_ext/time/zones.rb:65:in 'use_zone'
    from active_job/timezones.rb:9:in 'block (2 levels) in <module:Timezones>'
    from active_support/callbacks.rb:130:in 'instance_exec'
    from active_support/callbacks.rb:130:in 'block in run_callbacks'
    from active_support/callbacks.rb:141:in 'run_callbacks'
    from active_job/execution.rb:67:in '_perform_job'
    from active_job/instrumentation.rb:32:in '_perform_job'
    from active_job/execution.rb:51:in 'perform_now'
    from active_job/instrumentation.rb:26:in 'block in perform_now'
    from active_record/railties/job_runtime.rb:13:in 'block in instrument'
    from active_job/instrumentation.rb:40:in 'block in instrument'
    from active_support/notifications.rb:210:in 'block in instrument'
    from active_support/notifications/instrumenter.rb:58:in 'instrument'
    from active_support/notifications.rb:210:in 'instrument'
    from active_job/instrumentation.rb:39:in 'instrument'
    from active_record/railties/job_runtime.rb:11:in 'instrument'
    from active_job/instrumentation.rb:26:in 'perform_now'
    from active_job/logging.rb:32:in 'block in perform_now'
    from active_support/tagged_logging.rb:138:in 'block in tagged'
    from active_support/tagged_logging.rb:38:in 'tagged'
    from active_support/tagged_logging.rb:138:in 'tagged'
    from active_support/broadcast_logger.rb:241:in 'method_missing'
    from active_job/logging.rb:39:in 'tag_logger'
    from active_job/logging.rb:32:in 'perform_now'
    from sentry/rails/active_job.rb:10:in 'perform_now'
    from active_job/execution.rb:29:in 'block in execute'
    from active_support/callbacks.rb:121:in 'block in run_callbacks'
    from active_job/railtie.rb:79:in 'block (4 levels) in <class:Railtie>'
    from active_support/reloader.rb:77:in 'block in wrap'
    from active_support/execution_wrapper.rb:87:in 'wrap'
    from active_support/reloader.rb:74:in 'wrap'
    from active_job/railtie.rb:78:in 'block (3 levels) in <class:Railtie>'
    from active_support/callbacks.rb:130:in 'instance_exec'
    from active_support/callbacks.rb:130:in 'block in run_callbacks'
    from active_support/callbacks.rb:141:in 'run_callbacks'
    from active_job/execution.rb:27:in 'execute'
    from active_job/queue_adapters/sidekiq_adapter.rb:70:in 'perform'
    from sidekiq/processor.rb:202:in 'execute_job'
    from sidekiq/processor.rb:170:in 'block (2 levels) in process'
    from sidekiq/middleware/chain.rb:177:in 'block in invoke'
    from sidekiq/history/middleware.rb:54:in 'call'
    from sidekiq/middleware/chain.rb:179:in 'block in invoke'
    from sentry/sidekiq/sentry_context_middleware.rb:69:in 'call'
    from sidekiq/middleware/chain.rb:179:in 'block in invoke'
    from sidekiq/failures/middleware.rb:9:in 'call'
    from sidekiq/middleware/chain.rb:179:in 'block in invoke'
    from sidekiq/middleware/chain.rb:182:in 'invoke'
    from sidekiq/processor.rb:169:in 'block in process'
    from sidekiq/processor.rb:136:in 'block (6 levels) in dispatch'
    from sidekiq/job_retry.rb:113:in 'local'
    from sidekiq/processor.rb:135:in 'block (5 levels) in dispatch'
    from sidekiq/rails.rb:14:in 'block in call'
    from active_support/reloader.rb:77:in 'block in wrap'
    from active_support/execution_wrapper.rb:91:in 'wrap'
    from active_support/reloader.rb:74:in 'wrap'
    from sidekiq/rails.rb:13:in 'call'
    from sidekiq/processor.rb:131:in 'block (4 levels) in dispatch'
    from sidekiq/processor.rb:263:in 'stats'
    from sidekiq/processor.rb:126:in 'block (3 levels) in dispatch'
    from sidekiq/job_logger.rb:13:in 'call'
    from sidekiq/processor.rb:125:in 'block (2 levels) in dispatch'
    from sidekiq/job_retry.rb:80:in 'global'
    from sidekiq/processor.rb:124:in 'block in dispatch'
    from sidekiq/job_logger.rb:39:in 'prepare'
    from sidekiq/processor.rb:123:in 'dispatch'
    from sidekiq/processor.rb:168:in 'process'
    from sidekiq/processor.rb:78:in 'process_one'
    from sidekiq/processor.rb:68:in 'run'
    from sidekiq/component.rb:8:in 'watchdog'
    from sidekiq/component.rb:17:in 'block in safe_thread'

At the level of HTTP request it's a PUT request to https://{bucket}.s3.{region}.amazonaws.com/* resulting in 400 Bad Request, when "The CRC32 you specified did not match the calculated checksum" looks like the error message S3 responds with.

At first I assumed that it is a problem with Active Storage not being ready for the upgrade, but I was not able to reproduce the bad request at all. Below is the script that I thought might reproduce it, but it just works for some reasons.

Details

# frozen_string_literal: true

require "bundler/inline"

gemfile(true) do
  source "https://rubygems.org"

  gem "rails", "~> 7.2.0"
  # If you want to test against edge Rails replace the previous line with this:
  # gem "rails", github: "rails/rails", branch: "main"

  gem "sqlite3"
  gem "aws-sdk-s3", "1.178.0"
end

require "active_record/railtie"
require "active_storage/engine"
require "minitest/autorun"

ENV["DATABASE_URL"] = "sqlite3::memory:"

class TestApp < Rails::Application
  config.load_defaults Rails::VERSION::STRING.to_f

  config.root = __dir__
  config.hosts << "example.org"
  config.eager_load = false
  config.session_store :cookie_store, key: "cookie_store_key"
  config.secret_key_base = "secret_key_base"

  config.logger = Logger.new($stdout)
  Rails.logger  = config.logger

  config.active_storage.service = :amazon
  config.active_storage.service_configurations = {
    amazon: {
      service: "S3",
      bucket: ENV["AWS_S3_BUCKET"],
      region: ENV["AWS_S3_REGION"],
      access_key_id: ENV["AWS_ACCESS_KEY_ID"],
      secret_access_key: ENV["AWS_SECRET_ACCESS_KEY"],
    }
  }

  config.active_job.queue_adapter = :inline
end
Rails.application.initialize!

require ActiveStorage::Engine.root.join("db/migrate/20170806125915_create_active_storage_tables.rb").to_s

ActiveRecord::Schema.define do
  CreateActiveStorageTables.new.change

  create_table :users, force: true
end

class User < ActiveRecord::Base
  has_one_attached :profile
end

class BugTest < ActiveSupport::TestCase
  def test_upload_and_download
    user = User.create!(
      profile: {
        content_type: "text/plain",
        filename: "dummy.txt",
        io: ::StringIO.new("dummy"),
      }
    )

    assert_equal "dummy", user.profile.download
  end
end

Configuring the client to opt out of the data integrity protections by setting request_checksum_calculation and response_checksum_validation to when_required resolves the issue.

However, disabling the better protection feels like the ultimate decision and it would be great if I can avoid that. But the problem is now that the situation is very unclear so far and I though I should create an issue rather than suffering from it alone.

There are a couple of preceding issues around like these

• Upgrading from aws-sdk-s3 (1.176.1) to aws-sdk-s3 (1.178.0) breaks uploads to Cloudflare R2 #3174
• ActiveStorage creates conflicting checksums when using S3-compatible services on Blob.create_and_upload! rails/rails#54374

Note that both of these issues target Cloudflare R2, while mine targets AWS S3 itself. Also, the exact error message I'm observing which goes like this;

The CRC32 you specified did not match the calculated checksum

and it appears very rarely online, except for this one instance: awslabs/aws-sdk-rust#1240 (comment).

Regression Issue

• Select this option if this issue appears to be a regression.

Expected Behavior

I expect the required actions become clear so I can migrate to the newer default without just turning it off

Current Behavior

The checksum does not match 100% and the mechanism behind it is not so clear. It is unfortunate that I cannot be so sure if it's really inevitable and I can only turn the protection off, or there are actually any actions that I can take and they are only hard to find

Reproduction Steps

Although I have not been able to create a reproduction, the script that I posted above can be a starting point.

Claiming I can observe the error while not being able to reproduce is simply my fault. There should be something somewhere on my side so let me take time to further investigate and hopefully pinpoint, while posting the error and situation as the starting point.

For example, I freshly created a bucket to try to reproduce the issue, while the bucket I can associate the issue with is more than 5 years old. Getting an older bucket to test against is perhaps worth trying.

Sorry for this being a vague assumption. Let me update if things become clearer

Possible Solution

No response

Additional Information/Context

These are dependencies I'm observing the issue with

• ruby 3.3.8
• activestorage 7.2.2.2
• aws-sdk-core 3.239.2
• aws-sdk-s3 1.205.0
• net-http 0.7.0

Gem name ('aws-sdk', 'aws-sdk-resources' or service gems like 'aws-sdk-s3') and its version

aws-sdk-s3 1.178.0 and higher

Environment details (Version of Ruby, OS environment)

Ruby 3.3.8

[jterapin]

Thanks for the report! We'll look into this and help with debugging or provide a fix.

How often does this happen with the latest versions? Interesting that your test script couldn't reproduce it.

[sato11]

Thanks for the comment!

Regarding reproduction, I have managed to make the error happen constantly. Here is the updated script:

Details

# frozen_string_literal: true

require "bundler/inline"

gemfile(true) do
  source "https://rubygems.org"

  gem "rails", "~> 7.2.0"
  # If you want to test against edge Rails replace the previous line with this:
  # gem "rails", github: "rails/rails", branch: "main"

  gem "sqlite3"
  gem "aws-sdk-s3", "1.178.0"
  gem "tempfile", "0.2.1"
end

require "active_record/railtie"
require "active_storage/engine"
require "minitest/autorun"

ENV["DATABASE_URL"] = "sqlite3::memory:"

class TestApp < Rails::Application
  config.load_defaults Rails::VERSION::STRING.to_f

  config.root = __dir__
  config.hosts << "example.org"
  config.eager_load = false
  config.session_store :cookie_store, key: "cookie_store_key"
  config.secret_key_base = "secret_key_base"

  config.logger = Logger.new($stdout)
  Rails.logger  = config.logger

  config.active_storage.service = :amazon
  config.active_storage.service_configurations = {
    amazon: {
      service: "S3",
      bucket: ENV["AWS_S3_BUCKET"],
      region: ENV["AWS_S3_REGION"],
      access_key_id: ENV["AWS_ACCESS_KEY_ID"],
      secret_access_key: ENV["AWS_SECRET_ACCESS_KEY"],
    }
  }

  config.active_job.queue_adapter = :inline
end
Rails.application.initialize!

require ActiveStorage::Engine.root.join("db/migrate/20170806125915_create_active_storage_tables.rb").to_s

ActiveRecord::Schema.define do
  CreateActiveStorageTables.new.change

  create_table :users, force: true
end

class User < ActiveRecord::Base
  has_one_attached :profile
end

class BugTest < ActiveSupport::TestCase
  def test_upload_and_download
    user = User.create!
    profile = ::Tempfile.open("profile", encoding: ::Encoding::SJIS) do |tempfile|
      data = "こんにちは".encode(::Encoding::SJIS, undef: :replace)
      tempfile.write(data)
      ::ActionDispatch::Http::UploadedFile.new(filename: "profile", tempfile: tempfile)
    end
    user.profile.attach(profile)

    assert_equal "こんにちは", user.profile.download
  end
end

And the error (verbose)

Details

Fetching gem metadata from https://rubygems.org/...........
Resolving dependencies...
-- create_table(:active_storage_blobs, {:id=>:primary_key})
D, [2025-12-01T10:28:33.802350 #95848] DEBUG -- :    (2.9ms)  CREATE TABLE "active_storage_blobs" ("id" integer PRIMARY KEY AUTOINCREMENT NOT NULL, "key" varchar NOT NULL, "filename" varchar NOT NULL, "content_type" varchar, "metadata" text, "service_name" varchar NOT NULL, "byte_size" bigint NOT NULL, "checksum" varchar, "created_at" datetime(6) NOT NULL)
D, [2025-12-01T10:28:33.802477 #95848] DEBUG -- :    (0.0ms)  CREATE UNIQUE INDEX "index_active_storage_blobs_on_key" ON "active_storage_blobs" ("key")
   -> 0.0036s
-- create_table(:active_storage_attachments, {:id=>:primary_key})
D, [2025-12-01T10:28:33.802690 #95848] DEBUG -- :    (0.0ms)  CREATE TABLE "active_storage_attachments" ("id" integer PRIMARY KEY AUTOINCREMENT NOT NULL, "name" varchar NOT NULL, "record_type" varchar NOT NULL, "record_id" bigint NOT NULL, "blob_id" bigint NOT NULL, "created_at" datetime(6) NOT NULL, CONSTRAINT "fk_rails_c3b3935057"
FOREIGN KEY ("blob_id")
  REFERENCES "active_storage_blobs" ("id")
)
D, [2025-12-01T10:28:33.802922 #95848] DEBUG -- :    (0.0ms)  CREATE INDEX "index_active_storage_attachments_on_blob_id" ON "active_storage_attachments" ("blob_id")
D, [2025-12-01T10:28:33.802985 #95848] DEBUG -- :    (0.0ms)  CREATE UNIQUE INDEX "index_active_storage_attachments_uniqueness" ON "active_storage_attachments" ("record_type", "record_id", "name", "blob_id")
   -> 0.0005s
-- create_table(:active_storage_variant_records, {:id=>:primary_key})
D, [2025-12-01T10:28:33.803121 #95848] DEBUG -- :    (0.0ms)  CREATE TABLE "active_storage_variant_records" ("id" integer PRIMARY KEY AUTOINCREMENT NOT NULL, "blob_id" bigint NOT NULL, "variation_digest" varchar NOT NULL, CONSTRAINT "fk_rails_993965df05"
FOREIGN KEY ("blob_id")
  REFERENCES "active_storage_blobs" ("id")
)
D, [2025-12-01T10:28:33.803180 #95848] DEBUG -- :    (0.0ms)  CREATE UNIQUE INDEX "index_active_storage_variant_records_uniqueness" ON "active_storage_variant_records" ("blob_id", "variation_digest")
   -> 0.0002s
-- create_table(:users, {:force=>true})
D, [2025-12-01T10:28:33.803256 #95848] DEBUG -- :    (0.0ms)  DROP TABLE IF EXISTS "users"
D, [2025-12-01T10:28:33.803309 #95848] DEBUG -- :    (0.0ms)  CREATE TABLE "users" ("id" integer PRIMARY KEY AUTOINCREMENT NOT NULL)
   -> 0.0001s
D, [2025-12-01T10:28:33.803815 #95848] DEBUG -- :    (0.0ms)  CREATE TABLE "schema_migrations" ("version" varchar NOT NULL PRIMARY KEY)
D, [2025-12-01T10:28:33.804477 #95848] DEBUG -- :    (0.1ms)  CREATE TABLE "ar_internal_metadata" ("key" varchar NOT NULL PRIMARY KEY, "value" varchar, "created_at" datetime(6) NOT NULL, "updated_at" datetime(6) NOT NULL)
D, [2025-12-01T10:28:33.815685 #95848] DEBUG -- :   ActiveRecord::InternalMetadata Load (0.1ms)  SELECT * FROM "ar_internal_metadata" WHERE "ar_internal_metadata"."key" = ? ORDER BY "ar_internal_metadata"."key" ASC LIMIT 1  [[nil, "environment"]]
D, [2025-12-01T10:28:33.815877 #95848] DEBUG -- :   ActiveRecord::InternalMetadata Create (0.0ms)  INSERT INTO "ar_internal_metadata" ("key", "value", "created_at", "updated_at") VALUES ('environment', 'development', '2025-12-01 01:28:33.815721', '2025-12-01 01:28:33.815723') RETURNING "key"
Run options: --seed 48675

# Running:

I, [2025-12-01T10:28:33.948176 #95848]  INFO -- : ---------------------------------
I, [2025-12-01T10:28:33.948195 #95848]  INFO -- : BugTest: test_upload_and_download
I, [2025-12-01T10:28:33.948200 #95848]  INFO -- : ---------------------------------
D, [2025-12-01T10:28:33.950533 #95848] DEBUG -- :   TRANSACTION (0.0ms)  begin transaction
D, [2025-12-01T10:28:33.950602 #95848] DEBUG -- :   User Create (0.1ms)  INSERT INTO "users" DEFAULT VALUES RETURNING "id"
D, [2025-12-01T10:28:33.950700 #95848] DEBUG -- :   TRANSACTION (0.0ms)  commit transaction
D, [2025-12-01T10:28:33.973453 #95848] DEBUG -- :   TRANSACTION (0.0ms)  begin transaction
D, [2025-12-01T10:28:33.975141 #95848] DEBUG -- :   ActiveStorage::Blob Load (0.2ms)  SELECT "active_storage_blobs".* FROM "active_storage_blobs" INNER JOIN "active_storage_attachments" ON "active_storage_blobs"."id" = "active_storage_attachments"."blob_id" WHERE "active_storage_attachments"."record_id" = ? AND "active_storage_attachments"."record_type" = ? AND "active_storage_attachments"."name" = ? LIMIT ?  [["record_id", 1], ["record_type", "User"], ["name", "profile"], ["LIMIT", 1]]
D, [2025-12-01T10:28:33.980796 #95848] DEBUG -- :   ActiveStorage::Attachment Load (0.1ms)  SELECT "active_storage_attachments".* FROM "active_storage_attachments" WHERE "active_storage_attachments"."record_id" = ? AND "active_storage_attachments"."record_type" = ? AND "active_storage_attachments"."name" = ? LIMIT ?  [["record_id", 1], ["record_type", "User"], ["name", "profile"], ["LIMIT", 1]]
D, [2025-12-01T10:28:33.982019 #95848] DEBUG -- :   ActiveStorage::Blob Create (0.2ms)  INSERT INTO "active_storage_blobs" ("key", "filename", "content_type", "metadata", "service_name", "byte_size", "checksum", "created_at") VALUES (?, ?, ?, ?, ?, ?, ?, ?) RETURNING "id"  [["key", "hh8pvxd5yjvxdcasat73e9ki3166"], ["filename", "profile"], ["content_type", "application/octet-stream"], ["metadata", "{\"identified\":true}"], ["service_name", "amazon"], ["byte_size", 10], ["checksum", "IxbDzNtwSdxLOLuEU+P1OA=="], ["created_at", "2025-12-01 01:28:33.981280"]]
D, [2025-12-01T10:28:33.982585 #95848] DEBUG -- :   ActiveStorage::Attachment Create (0.1ms)  INSERT INTO "active_storage_attachments" ("name", "record_type", "record_id", "blob_id", "created_at") VALUES (?, ?, ?, ?, ?) RETURNING "id"  [["name", "profile"], ["record_type", "User"], ["record_id", 1], ["blob_id", 1], ["created_at", "2025-12-01 01:28:33.982096"]]
D, [2025-12-01T10:28:33.982908 #95848] DEBUG -- :   TRANSACTION (0.0ms)  commit transaction
I, [2025-12-01T10:28:36.500827 #95848]  INFO -- :   S3 Storage (2517.5ms) Uploaded file to key: hh8pvxd5yjvxdcasat73e9ki3166 (checksum: IxbDzNtwSdxLOLuEU+P1OA==)
E

Finished in 2.558948s, 0.3908 runs/s, 0.0000 assertions/s.

  1) Error:
BugTest#test_upload_and_download:
ActiveStorage::IntegrityError: ActiveStorage::IntegrityError
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activestorage-7.2.3/lib/active_storage/service/s3_service.rb:145:in `rescue in upload_with_single_part'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activestorage-7.2.3/lib/active_storage/service/s3_service.rb:142:in `upload_with_single_part'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activestorage-7.2.3/lib/active_storage/service/s3_service.rb:34:in `block in upload'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activesupport-7.2.3/lib/active_support/notifications.rb:210:in `block in instrument'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activesupport-7.2.3/lib/active_support/notifications/instrumenter.rb:58:in `instrument'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activesupport-7.2.3/lib/active_support/notifications.rb:210:in `instrument'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activestorage-7.2.3/lib/active_storage/service.rb:165:in `instrument'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activestorage-7.2.3/lib/active_storage/service/s3_service.rb:30:in `upload'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activestorage-7.2.3/app/models/active_storage/blob.rb:257:in `upload_without_unfurling'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activestorage-7.2.3/lib/active_storage/attached/changes/create_one.rb:26:in `upload'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activesupport-7.2.3/lib/active_support/core_ext/object/try.rb:15:in `public_send'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activesupport-7.2.3/lib/active_support/core_ext/object/try.rb:15:in `try'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activestorage-7.2.3/lib/active_storage/attached/model.rb:142:in `block in has_one_attached'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activesupport-7.2.3/lib/active_support/callbacks.rb:407:in `instance_exec'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activesupport-7.2.3/lib/active_support/callbacks.rb:407:in `block in make_lambda'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activesupport-7.2.3/lib/active_support/callbacks.rb:208:in `call'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activesupport-7.2.3/lib/active_support/callbacks.rb:564:in `block in invoke_after'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activesupport-7.2.3/lib/active_support/callbacks.rb:564:in `each'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activesupport-7.2.3/lib/active_support/callbacks.rb:564:in `invoke_after'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activesupport-7.2.3/lib/active_support/callbacks.rb:111:in `run_callbacks'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activesupport-7.2.3/lib/active_support/callbacks.rb:914:in `_run_commit_callbacks'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activerecord-7.2.3/lib/active_record/transactions.rb:385:in `committed!'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activerecord-7.2.3/lib/active_record/connection_adapters/abstract/transaction.rb:307:in `block in commit_records'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activerecord-7.2.3/lib/active_record/connection_adapters/abstract/transaction.rb:344:in `run_action_on_records'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activerecord-7.2.3/lib/active_record/connection_adapters/abstract/transaction.rb:306:in `commit_records'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activerecord-7.2.3/lib/active_record/connection_adapters/abstract/transaction.rb:596:in `block in commit_transaction'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activesupport-7.2.3/lib/active_support/concurrency/null_lock.rb:9:in `synchronize'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activerecord-7.2.3/lib/active_record/connection_adapters/abstract/transaction.rb:584:in `commit_transaction'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activerecord-7.2.3/lib/active_record/connection_adapters/abstract/transaction.rb:628:in `block in within_new_transaction'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activesupport-7.2.3/lib/active_support/concurrency/null_lock.rb:9:in `synchronize'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activerecord-7.2.3/lib/active_record/connection_adapters/abstract/transaction.rb:613:in `within_new_transaction'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activerecord-7.2.3/lib/active_record/connection_adapters/abstract/database_statements.rb:361:in `transaction'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activerecord-7.2.3/lib/active_record/transactions.rb:414:in `block in with_transaction_returning_status'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activerecord-7.2.3/lib/active_record/connection_adapters/abstract/connection_pool.rb:425:in `with_connection'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activerecord-7.2.3/lib/active_record/connection_handling.rb:298:in `with_connection'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activerecord-7.2.3/lib/active_record/transactions.rb:410:in `with_transaction_returning_status'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activerecord-7.2.3/lib/active_record/transactions.rb:362:in `save'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activerecord-7.2.3/lib/active_record/suppressor.rb:52:in `save'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activestorage-7.2.3/lib/active_storage/attached/one.rb:61:in `attach'
    main.rb:70:in `test_upload_and_download'

It is very suggestive that the combination of irregularly forced encoding (i.e. Encoding::SJIS) and multibyte characters as input (i.e. "こんにちは") might corrupt the output checksum.

With that said,

How often does this happen with the latest versions?

I experienced at least two features went down without fluctuation before reverting the upgrade. I can confirm multibyte characters participate in both of two features.

I hope the test script above reproduces the failure not only in my environment so that it can corroborate my hypothesis.

Will you take a look again? @jterapin

On the other hand, I'll be looking into why we need to force the unorthodox encoding which might have caused the failure, wondering if it is a tech debt in itself.

[jterapin]

Hi! I've been digging into this case and here's what I've got so far:

I ran your script and hit a similar error. Then, I wanted to see if I get the same error when I test the S3 gem directly (without ActiveStorage in the mix).

Based on the initial stacktrace, it looks like ActiveStorage is using either TransferManager or S3 Resource behind the scenes. I decided to run the below code:

tm = Aws::S3::TransferManager.new
tempfile = Tempfile.new('foo')
data = "こんにちは".encode(::Encoding::SJIS, undef: :replace)
tempfile.write(data)
tm.upload_file(tempfile, key: 'my-key', bucket: 'my-bucket')
# => [Aws::S3::Client 200 0.252548 0 retries] put_object(<omitted>)
# => true

Since there wasn't a visible error here, I'm plan to look into where the S3 calls are happening on the ActiveStorage side. I'll keep you updated here.

[jterapin]

In my investigations so far - I'm thinking it really comes down to encoding (like you suspected).

I'm going to assume the input below is similar to what you already have set up in prod (specific encoding, etc). I observed the following:

  def test_upload_and_download
    user = User.create!
    profile = ::Tempfile.open("profile", encoding: ::Encoding::SJIS) do |tempfile|
      data = "こんにちは".encode(::Encoding::SJIS, undef: :replace)
      puts "=== STEP 1: Original data ==="
      puts "Original SJIS encoding: #{data.encoding}" #  Windows-31J

      tempfile.write(data)
      tempfile.flush
      tempfile.rewind

      puts "=== STEP 2: After writing to tempfile ==="
      content = tempfile.read
      puts "Tempfile content encoding: #{content.encoding}" # UTF-8
      puts "Bytes match original? #{content.bytes == data.bytes}"  # false
      tempfile.rewind

      ::ActionDispatch::Http::UploadedFile.new(filename: "profile", tempfile: tempfile)
    end
    user.profile.attach(profile)
    assert_equal "こんにちは", user.profile.attach(profile).downloaded
  end

As you can see, there's definitely some mismatch happening. I made the following changes to the test:

def test_upload_and_download
  user = User.create!
  profile = ::Tempfile.open("profile") do |tempfile| 
    tempfile.binmode  
    data = "こんにちは".encode(::Encoding::SJIS, undef: :replace)
    tempfile.write(data)
    ::ActionDispatch::Http::UploadedFile.new(filename: "profile", tempfile: tempfile)
  end
  user.profile.attach(profile)

  # Decode the SJIS bytes back to UTF-8 for comparison
  downloaded = user.profile.download.force_encoding('Windows-31J').encode('UTF-8')
  assert_equal "こんにちは", downloaded
end

With the change above, I got the test to pass. Could you check if setting the tempfile to binmode fixes things on your end? I need more time to dig into whether this is something the SDK should consider or not. There's no timeline on when I'll have the final outcome, but hope this unblocks you for now.

[sato11]

Thank you @jterapin, this is a beautiful analysis full of learning!

I did give it a try and this is what I happened to observe. Firstly, setting IO to binmode actually resolved the reproduction test case. Then, I incorporated the change and tested against the real product. The result was the same error remaining.

It has turned out that there is actually one more subtle complication that I have omitted in the test case. Excuse me, I thought it was irrelevant when the error used to pop up with or without this detail. But in reality it does seem more relevant now with the presence of binmode.

This is the key change that causes regression on my end:

diff --git i/main.rb w/main.rb
index 80c2bce..29847ac 100644
--- i/main.rb
+++ w/main.rb
@@ -62,7 +62,7 @@ end
 class BugTest < ActiveSupport::TestCase
   def test_upload_and_download
     user = User.create!
-    profile = ::Tempfile.open("profile") do |tempfile|
+    profile = ::Tempfile.open("profile", encoding: ::Encoding::SJIS) do |tempfile|
       tempfile.binmode
       data = "こんにちは".encode(::Encoding::SJIS, undef: :replace)
       puts "=== STEP 1: Original data ==="

The output suggests they are ok on the surface with encodings being Windows-31J and ASCII-8BIT (I believe they are siblings), as opposed to how it used to be Windows-31J and UTF-8.

=== STEP 1: Original data ===
Original SJIS encoding: Windows-31J
=== STEP 2: After writing to tempfile ===
Tempfile content encoding: ASCII-8BIT
Bytes match original? true

Despite that, ActiveStorage::IntegrityError recurs.

Details

Finished in 2.575497s, 0.3883 runs/s, 0.0000 assertions/s.

  1) Error:
BugTest#test_upload_and_download:
ActiveStorage::IntegrityError: ActiveStorage::IntegrityError
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activestorage-7.2.3/lib/active_storage/service/s3_service.rb:145:in `rescue in upload_with_single_part'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activestorage-7.2.3/lib/active_storage/service/s3_service.rb:142:in `upload_with_single_part'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activestorage-7.2.3/lib/active_storage/service/s3_service.rb:34:in `block in upload'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activesupport-7.2.3/lib/active_support/notifications.rb:210:in `block in instrument'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activesupport-7.2.3/lib/active_support/notifications/instrumenter.rb:58:in `instrument'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activesupport-7.2.3/lib/active_support/notifications.rb:210:in `instrument'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activestorage-7.2.3/lib/active_storage/service.rb:165:in `instrument'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activestorage-7.2.3/lib/active_storage/service/s3_service.rb:30:in `upload'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activestorage-7.2.3/app/models/active_storage/blob.rb:257:in `upload_without_unfurling'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activestorage-7.2.3/lib/active_storage/attached/changes/create_one.rb:26:in `upload'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activesupport-7.2.3/lib/active_support/core_ext/object/try.rb:15:in `public_send'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activesupport-7.2.3/lib/active_support/core_ext/object/try.rb:15:in `try'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activestorage-7.2.3/lib/active_storage/attached/model.rb:142:in `block in has_one_attached'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activesupport-7.2.3/lib/active_support/callbacks.rb:407:in `instance_exec'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activesupport-7.2.3/lib/active_support/callbacks.rb:407:in `block in make_lambda'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activesupport-7.2.3/lib/active_support/callbacks.rb:208:in `call'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activesupport-7.2.3/lib/active_support/callbacks.rb:564:in `block in invoke_after'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activesupport-7.2.3/lib/active_support/callbacks.rb:564:in `each'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activesupport-7.2.3/lib/active_support/callbacks.rb:564:in `invoke_after'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activesupport-7.2.3/lib/active_support/callbacks.rb:111:in `run_callbacks'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activesupport-7.2.3/lib/active_support/callbacks.rb:914:in `_run_commit_callbacks'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activerecord-7.2.3/lib/active_record/transactions.rb:385:in `committed!'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activerecord-7.2.3/lib/active_record/connection_adapters/abstract/transaction.rb:307:in `block in commit_records'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activerecord-7.2.3/lib/active_record/connection_adapters/abstract/transaction.rb:344:in `run_action_on_records'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activerecord-7.2.3/lib/active_record/connection_adapters/abstract/transaction.rb:306:in `commit_records'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activerecord-7.2.3/lib/active_record/connection_adapters/abstract/transaction.rb:596:in `block in commit_transaction'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activesupport-7.2.3/lib/active_support/concurrency/null_lock.rb:9:in `synchronize'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activerecord-7.2.3/lib/active_record/connection_adapters/abstract/transaction.rb:584:in `commit_transaction'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activerecord-7.2.3/lib/active_record/connection_adapters/abstract/transaction.rb:628:in `block in within_new_transaction'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activesupport-7.2.3/lib/active_support/concurrency/null_lock.rb:9:in `synchronize'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activerecord-7.2.3/lib/active_record/connection_adapters/abstract/transaction.rb:613:in `within_new_transaction'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activerecord-7.2.3/lib/active_record/connection_adapters/abstract/database_statements.rb:361:in `transaction'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activerecord-7.2.3/lib/active_record/transactions.rb:414:in `block in with_transaction_returning_status'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activerecord-7.2.3/lib/active_record/connection_adapters/abstract/connection_pool.rb:425:in `with_connection'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activerecord-7.2.3/lib/active_record/connection_handling.rb:298:in `with_connection'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activerecord-7.2.3/lib/active_record/transactions.rb:410:in `with_transaction_returning_status'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activerecord-7.2.3/lib/active_record/transactions.rb:362:in `save'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activerecord-7.2.3/lib/active_record/suppressor.rb:52:in `save'
    /Users/sato11/.asdf/installs/ruby/3.3.8/lib/ruby/gems/3.3.0/gems/activestorage-7.2.3/lib/active_storage/attached/one.rb:61:in `attach'
    main.rb:83:in `test_upload_and_download'

1 runs, 0 assertions, 0 failures, 1 errors, 0 skips

I find myself unable to make sense, but does this suggest that there is a case where the exactly matching bytes (though different encodings) calculate different checksums, given the error message ("The CRC32 you specified did not match the calculated checksum") is unchanged? This is not very intuitive a speculation and I'm eager to know what you would think.

To recap, the binmode solution does not seem to unblock me unfortunately. On the other hand, I'm more convinced that it is more effective to give up supporting this troublesome legacy encoding.

Again, thank you for the investigation!

[jterapin]

Hi! Thank you for your patience, @sato11 ~ I hope that the following will help you:

So the checksum mismatch is happening because of how Ruby handles encoding when reading.
See below for debugging code:

Details

    user = User.create!
    puts "== CASE 1: ORIGINAL CASE =="
    profile = ::Tempfile.open("profile", encoding: ::Encoding::SJIS) do |tempfile|
      puts "=== STEP 0: Tempfile Check ==="
      puts "Tempfile internal encoding: #{tempfile.internal_encoding}" # UTF-8 (default)
      puts "Tempfile external encoding: #{tempfile.external_encoding}" # Windows-31J

      data = "こんにちは".encode(::Encoding::SJIS, undef: :replace)
      puts "=== STEP 1: Original data ==="
      puts "Original SJIS encoding: #{data.encoding}" #  Windows-31J
      tempfile.write(data)

      puts "=== STEP 2: Read after writing to temp ==="
      content = tempfile.read
      puts "Data encoding on tempfile.read: #{content.encoding}" #  UTF-8
      puts "Bytes match original? #{content.bytes == data.bytes}" # false
      tempfile.rewind
      ::ActionDispatch::Http::UploadedFile.new(filename: "profile", tempfile: tempfile)
    end

    puts "=== STEP 3: Checking profile contents AFTER write/close ==="
    profile_content = profile.open.read
    puts "Data: #{profile_content.encoding}" # UTF-8
    puts "Tempfile internal encoding: #{profile.tempfile.internal_encoding}" # UTF-8
    puts "Tempfile external encoding: #{profile.tempfile.external_encoding}" # Windows-31J
    profile.rewind

    begin
      user.profile.attach(profile)
      puts "ORIGINAL CASE SUCCESS"
    rescue
      puts "ORIGINAL CASE FAILED"
    end

The above outputs:

== CASE 1: ORIGINAL CASE ==
=== STEP 0: Tempfile Check ===
Tempfile internal encoding: UTF-8
Tempfile external encoding: Windows-31J
=== STEP 1: Original data ===
Original SJIS encoding: Windows-31J
=== STEP 2: Read after writing to temp ===
Data encoding on tempfile.read: UTF-8
Bytes match original? false
=== STEP 3: Checking profile contents AFTER write/close ===
Data: UTF-8
Tempfile internal encoding: UTF-8
Tempfile external encoding: Windows-31J
ORIGINAL CASE FAILED

So when our SDK's Digest class does .read on the tempfile, it's getting UTF-8 bytes - hence why the checksum calculation failed.

To keep the legacy encoding around, I would suggest that you set both internal/external encoding to the same thing.

Details

    user = User.create!
    puts "== CASE 4 : set both internal/external encoding to same =="
    profile = ::Tempfile.open("profile", encoding: "Windows-31J:Windows-31J") do |tempfile|
      puts "=== STEP 0: Tempfile Check ==="
      puts "Tempfile internal encoding: #{tempfile.internal_encoding}" #
      puts "Tempfile external encoding: #{tempfile.external_encoding}" # Windows-31J
      data = "こんにちは".encode(::Encoding::SJIS, undef: :replace)
      puts "=== STEP 1: Original data ==="
      puts "Original SJIS encoding: #{data.encoding}" #  Windows-31J

      tempfile.write(data)
      tempfile.flush
      tempfile.rewind

      puts "=== STEP 2: Data after writing to temp ==="
      content = tempfile.read
      puts "Data encoding after temp write: #{content.encoding}" #  Windows-31J
      puts "Bytes match original? #{content.bytes == data.bytes}"
      tempfile.rewind
      ::ActionDispatch::Http::UploadedFile.new(filename: "profile", tempfile: tempfile)
    end

    puts "=== STEP 3: Checking profile contents AFTER write/close ==="
    profile_content = profile.open.read
    puts "Data encoding: #{profile_content.encoding}" # Windows-31J
    puts "Tempfile internal encoding: #{profile.tempfile.internal_encoding}"
    puts "Tempfile external encoding: #{profile.tempfile.external_encoding}" #Windows-31J
    profile.rewind

    begin
      user.profile.attach(profile)
      puts "CASE 4 SUCCESS"
    rescue
      puts "CASE 4 FAILED"
    end

    downloaded = user.profile.download.force_encoding('Windows-31J').encode('UTF-8')
    assert_equal "こんにちは", downloaded

This outputs:

== CASE 4 : set both internal/external encoding to same ==
=== STEP 0: Tempfile Check ===
Tempfile internal encoding: 
Tempfile external encoding: Windows-31J
=== STEP 1: Original data ===
Original SJIS encoding: Windows-31J
=== STEP 2: Data after writing to temp ===
Data encoding after temp write: Windows-31J
Bytes match original? true
=== STEP 3: Checking profile contents AFTER write/close ===
Data encoding: Windows-31J
Tempfile internal encoding: 
Tempfile external encoding: Windows-31J
CASE 4 SUCCESS

Rather than Windows-31J, you could use: BINARY:BINARY instead.

Please let me know if this works on your end!

[sato11]

Hey @jterapin, thanks for the follow-up!

This is really intriguing! I have been completely unaware that there is such a distinctive concept of internal / external encodings under the hood. I'm not sure if I have digested its subtleties in full, but let me report that it seems those functionalities that I have associated the error with are running successfully after I followed your suggestion to explicitly set the external encoding 👍

[jterapin]

That's great to hear! I will close this ticket now but feel free to re-open it if there's any questions.

[github-actions]

This issue is now closed. Comments on closed issues are hard for our team to see.
If you need more assistance, please open a new issue that references this one.