Ruiyang Wang
Ruiyang Wang
### What happened + What you expected to happen `numba` is a JIT lib to compile python code to accelerate. It tries to type all functions and if it can't...
This PR is a proof of concept that, we *can* do bindings for Async C++ APIs. Specifically, we can wrap a C++ `ray::gcs::GcsClient` Async APIs (by callbacks) into Python Async...
Creates a direct Cython binding for `ray::gcs::GcsClient` and replaces the existing `PythonGcsClient` binding. The new binding is enabled by default; one can switch back with `RAY_USE_OLD_GCS_CLIENT=1`. The new binding is...
We added structured logging `RAY_LOG(INFO).WithField(kLogKeyJobID, job_id)`. However this is still a bit cumbersome and error prone, since one can write kLogKeyJobID with a NodeID and so on. This PR introduces...
Critical Dashboard API `GET /api/jobs` sends RPC to `JobInfoGcsService.GetAllJobInfo`, where the GCS sends RPC to each to `CoreWorkerService.NumPendingTasks` for the info of "how many running tasks do ya have rn?"....
### What happened + What you expected to happen When you have a task returning a Ray Object, and the task is in "waiting dependencies" status; and you do `ray.cancel(ref)`...
Some Ray IDs have those fields and some don't. Unify them with a list of methods and test. Note ObjectID is not in the list.
Like actor_head.py, we now update DataSource.nodes on delta. It first queries all node infos, then subscribes node deltas. Each delta updates: 1. DataSource.nodes[node_id] 2. DataSource.agents[node_id] 3. a warning generated after...
When starting a worker for a task, we have: - early bind: runtime env, is_gpu, is_actor - late bind: job_id, root_detached_id Now, move the latter 3 fields from "early bind"...
We already have this filter in GCS for state_manager.py. Also exposes it to node_head.py. Smoke tested each arg manually.