Wrap possible autoloaded models in interlock loading #730
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Since we don't eager load the models that could be loaded here, we need to wrap these calls in the interlock loading to ensure rails' interlock knows we're loading code in this block, so two threads don't try to autoload at the same time. Fixes timing issues resulting in errors of the variety: ``` LoadError (Unable to autoload constant ManageIQ::Providers::Openstack::StorageManager::CinderManager, expected /home/mmarosi/.rbenv/versions/2.5.3/lib/ruby/gems/2.5.0/bundler/gems/manageiq-providers-openstack-4481f21d2ebe/app/models/manageiq/providers/openstack/storage_manager/cinder_manager.rb to define it) ``` Fixes ManageIQ/manageiq#19772 In that issue, we had 3 requests happen in 3 threads at the same time: ``` [----] I, [2020-01-28T16:39:43.358817 #6310:3fd8bc363880] INFO -- : Started GET "/ems_cloud/ems_cloud_form_fields/new" for ::1 at 2020-01-28 16:39:43 -0500 [----] I, [2020-01-28T16:39:43.394484 #6310:3fd8bc363754] INFO -- : Started OPTIONS "/api/providers" for ::1 at 2020-01-28 16:39:43 -0500 [----] I, [2020-01-28T16:39:43.690661 #6310:3fd8bc3633a8] INFO -- : Started OPTIONS "/api/providers" for ::1 at 2020-01-28 16:39:43 -0500 ``` We were able to reduce that issue to this recreation script run locally against `rails server`: ```ruby threads = [] threads << Thread.new do `curl -k -L -u admin:smartvm -i -X GET http://localhost:3000/ems_cloud/ems_cloud_form_fields/new` end 2.times do threads << Thread.new do `curl -k -L -u admin:smartvm -i -X OPTIONS http://localhost:3000/api/providers` end end threads.join ```
|
Note, I don't know how to test this. The method is already tested and testing race conditions leaves me in the pits. 🏎 |
|
Checked commit jrafanie@6f602dc with ruby 2.5.5, rubocop 0.69.0, haml-lint 0.20.0, and yamllint 1.10.0 |
|
It feels like we could be playing whack-a-mole with any code that could possibly load a model and could possibly be run from a thread :/ I guess that is relatively rare but I'd be surprised if this didn't also happen somewhere in ui-classic? |
agrare
approved these changes
Jan 30, 2020
agrare
added a commit
that referenced
this pull request
Jan 30, 2020
…_loading Wrap possible autoloaded models in interlock loading
|
@miq-bot add-label constant problem |
|
@agrare Cannot apply the following label because they are not recognized: constant problem |
|
@agrare These constant problems will probably go away with Rails 6 and zeitwerk |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Since we don't eager load the models that could be loaded here, we need to wrap
these calls in the interlock loading to ensure rails' interlock knows we're
loading code in this block, so two threads don't try to autoload at the same
time.
Fixes timing issues resulting in errors of the variety:
Fixes ManageIQ/manageiq#19772
In that issue, we had 3 requests happen in 3 threads at the same time:
We were able to reduce that issue to this recreation script run locally against
rails server: