Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

DM-45988: Support storing init-outputs in Prompt Processing central repo #4436

Merged
merged 7 commits into from
Apr 8, 2025

Conversation

kfindeisen
Copy link
Collaborator

This PR adds a Job and CronJob that create init-outputs and output collections for the Prompt Processing service. By design, these jobs share much of their config with the primary service and must be updated together in order to work correctly.

@kfindeisen kfindeisen requested a review from dspeck1 March 28, 2025 17:44
Copy link
Contributor

@dspeck1 dspeck1 left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM. There a couple comments.

The initializer is a small-scale process that creates all the pipeline
init-outputs that the main service might need, and commits them to the
central repo exactly once. As such, it doesn't need much memory and no
storage, but *must* run before the service is viable.

The partial template allows the job definition to be shared between a
Job and a CronJob, ensuring consistency.
The service does not currently use Sasquatch, but it could in principle
send performance metrics.
The init service constructs all tasks in all pipelines, and
PackageAlertsTask tries to connect to the Kafka server at
construction time.
Rerunning the initializer daily is necessary because the run collection
names include the day_obs (and this feature is very useful for anybody
trying to analyze or reproduce the data later).
Knative HSC-gpu, LSSTCam, and LSSTComCam, and Keda LSSTComCam are
currently unused.
The job needs to prepare version- and configuration-specific settings
for the main service. By default, Argo only runs a job if no existing
job object exists.
The service depends on the init job to correctly and self-consistently
initialize the run collections for the day. Without the block, the
service would receive visits but then fail during preprocessing.
@kfindeisen kfindeisen added this pull request to the merge queue Apr 8, 2025
Merged via the queue into main with commit fcdb88c Apr 8, 2025
6 checks passed
@kfindeisen kfindeisen deleted the tickets/DM-45988 branch April 8, 2025 18:49
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants