-
Notifications
You must be signed in to change notification settings - Fork 151
Open
Milestone
Description
This issue will track progress and release notes for the 0.4 release.
Expected release date: March 19, 2026
Planned Changes
- RDMA
- Support for AWS EFA
- Potentially support for ROCm RDMA
- RDMA - Refactor the manager actor so it can support multiple backends (fallback to TCP, support EFA). #2571 Unified API to make it easier to port to other hardware or TCP backend
- Observability
- Add more structured live status data to distributed SQL telemetry #2448 Native per-job dashboard compatible with OSS, based on telemetry gathered from the whole monarch job
- Perfetto traces for OSS
- mesh admin API and reference tui/cli/etc. #2803 Admin Terminal UI
- Persist live SQL telemetry data after job ends for post-mortem debugging #2977
- Kubernetes
- Support for just-in-time provisioning of pods
- Add out-of-cluster support for KubernetesJob #2657 External gateway for an out-of-cluster client to drive a job
- Large scale barrier reproducible benchmark. #2800 Reproducible benchmark for large-scale barrier, along with improvements to the performance of this barrier
- Implement ProcSpawner RFC. #2548 New ProcSpawner API for supporting processes spawned by other mechanisms, such as processes spawned in a Docker container
Comments can be added for notable completed tasks / PRs that don't fall into the above categories
Reactions are currently unavailable
Metadata
Metadata
Assignees
Labels
No labels