Link claimed_executions to processes via process_name
#601
Conversation
This file contains hidden or 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
rosa
commented
Jul 21, 2025
| unless connection.column_exists?(:solid_queue_claimed_executions, :process_name) | ||
| add_column :solid_queue_claimed_executions, :process_name, :string | ||
| add_index :solid_queue_claimed_executions, :process_name | ||
| end |
There was a problem hiding this comment.
This migration is idempotent because it is already included in the setup schema, for people who are installing Solid Queue for the first time. If there's a new update in the future with a new migration, people should be able to run all of them without any issues.
rosa
force-pushed
the
claimed-executions-by-process-name
branch
from
bc62970 to
b18e146
Compare
And include it in the host application's deprecators.
It's been a long time already since version 1.0 went out.
To be used as `bin/rails solid_queue:update` to copy new migrations. The new migration links claimed executions to processes with `process_name`, and replaces the unique index on processes on `supervisor_id, name` with just `name`, as we'll want the name to uniquely identify a process because we'll rely on that to release claimed executions, independently from the supervisor. Even though it was really unlikely to have a collision on name because these are set randomly with `SecureRandom.hex(10)`, in this way we guarantee it. fix
People installing Solid Queue for the first time will get the final schema. Migrations are idempotent so they can later update and run migrations without any problems.
We can only use this if new migrations have been run. If not, we just emit a deprecation warning and continue as before.
In the deprecation warning and upgrade instructions.
rosa
force-pushed
the
claimed-executions-by-process-name
branch
from
b18e146 to
a06ef6b
Compare
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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.
That is, instead of linking them via the standard
process_id(claimed_execution.process_idandprocess.id).The reason for this change is that we're exploring the use of Solid Queue in a multi-tenanted setup, where we'll have an individual database per tenant, but a shared supervisor, dispatcher, scheduler, and workers for multiple tenants. We'll keep a supervisor-local database separated from the tenanted DBs, with just the
solid_queue_processestable. The supervisor will control the processes stored in that table in that local DB, whereas each tenant will use its own database with the remaining tables, which will contain the jobs and the job metadata.The consequence of this is that we can no longer rely on finding a process's in-flight jobs via its
id, because theidcan be (and most likely will be) duplicated across supervisors' local databases. The name, however, is randomly generated viaSecureRandom.hex(10), which, with a very high likelihood, guarantees unique names across supervisors' local DBs. In this way, if a process is killed, leaving in-flight jobs (claimed executions) behind, and we fail over a different supervisor with a different local DB, that supervisor will be able to know which claimed executions are orphaned.This requires a new migration, which is added here, but it's not enforced. The code still works without running the new migration, it just emits a deprecation warning like this:
This also includes a new
solid_queue:updatecommand to facilitate installing this migration and future migrations.cc @flavorjones