hyperstack-org / hyperstack-org/hyperstack
QueuedMessages from unsent/expired Connections are getting left in the database
Open
performance
- Dominant language
- JavaScript
- Stars
- 538
- Forks
- 41
- PR merge metrics
- No merged PRs in 30d
Description
Lately I've noticed poor performance in our production database for the Hyperstack tables, and noticed we had about 400K QueuedMessages in the table. Upon looking in the code, I found this:
```ruby
# hyper-operation/lib/hyper-operation/transport/connection.rb:111
expired.delete_all
# hyper-operation/lib/hyper-operation/transport/connection.rb:173
inactive.delete_all
```
ActiveRecord `delete_all` does not run callbacks, so the dependent QueuedMessages are not being destroyed when the Connection is. This should be changed to `destroy_all` so that the QueuedMessages don't get left in the database and accumulate.
Contributor guide
Assessment
This issue has not been assessed yet.