skypilot-org / skypilot

SkyPilot: Run AI and batch jobs on any infra (Kubernetes or 12+ clouds). Get unified execution, cost savings, and high GPU availability via a simple interface.
https://skypilot.readthedocs.io
Apache License 2.0
6.77k stars 505 forks source link

[Serve] Print out the service yaml in the controller log for better error tracking #3056

Open Michaelvll opened 9 months ago

Michaelvll commented 9 months ago

Since we now support updating services, it would be nice to print out the service yaml in the controller log whenever a new update is recieved. This will help debugging unexpected issues.

dtran24 commented 8 months ago

How would printing out the service yaml be different than the logging here? https://github.com/skypilot-org/skypilot/blob/3765f032824bcc3ee7f27d37edbbeded377e460d/sky/serve/controller.py#L118-L119 E.g. do we want like specific indentations from the original yaml or more fields included like here? https://github.com/skypilot-org/skypilot/blob/5cc68d2c9be0870d23b5e73ded1484b2e851268e/sky/serve/service_spec.py#L67-L74 If more fields, should this new log string exist in its own method, instead of using __repr__? https://github.com/skypilot-org/skypilot/blob/5cc68d2c9be0870d23b5e73ded1484b2e851268e/sky/serve/service_spec.py#L207-L212 cc @cblmemo in case you have any context as well

cblmemo commented 8 months ago

How would printing out the service yaml be different than the logging here?

https://github.com/skypilot-org/skypilot/blob/3765f032824bcc3ee7f27d37edbbeded377e460d/sky/serve/controller.py#L118-L119

E.g. do we want like specific indentations from the original yaml or more fields included like here? https://github.com/skypilot-org/skypilot/blob/5cc68d2c9be0870d23b5e73ded1484b2e851268e/sky/serve/service_spec.py#L67-L74

If more fields, should this new log string exist in its own method, instead of using __repr__? https://github.com/skypilot-org/skypilot/blob/5cc68d2c9be0870d23b5e73ded1484b2e851268e/sky/serve/service_spec.py#L207-L212

cc @cblmemo in case you have any context as well

github-actions[bot] commented 3 weeks ago

This issue is stale because it has been open 120 days with no activity. Remove stale label or comment or this will be closed in 10 days.