cortex
cortex copied to clipboard
Production infrastructure for machine learning at scale
Users that attach metadata using the logger won't be able to view their metadata because the logs command only extracts `message` and `exc_info` from logs. Add a flag to allow...
### Description A field like `priority` could be added to the job submission spec that could prioritize certain jobs over the others. The values the `priority` field could take could...
### Description When running `cortex deploy` in a situation when a rolling update cannot be performed (based on `max_surge` / `max_unavailable`), respond with an error rather than reaching a deadlocked...
### Motivation Users may want different API configurations during different times of the day ### Workarounds Until this is supported, it should be straightforward to achieve it using tools like...
Example command: `cortex deploy ` ### Motivation It can be useful to revert your API to a previous state.
### Description The operator should timeout if a request is taking too long ### Notes * Probably want to use `context.Context`: create new `Context` for every request, and that context...
* Istio [rate limiting](https://istio.io/docs/tasks/policy-enforcement/rate-limiting/) * Can use elasticache? ([stack overflow](https://stackoverflow.com/questions/54769212/elasticache-redis-cluster-and-istio))
#### Description Part of the `cortex cluster up` process would be to upload the appropriate version of all Cortex images to the user's private ECR. It might be worth considering...
#### Description Sometimes the auto scaling group fails to spin up instances, e.g. if the user is at their instance limit (spot or on demand), or if AWS is out...