The `schema_to_deployment()` function preserve unset fields with unexpected default argument types. This change excludes unset fields in that function and also changes the dictionaries' default values to empty dicts.
Buffering writes to AWS S3 is highly recommended to maximize throughput. Reducing the number of remote I/O requests can make spilling to remote storages as effective as spilling locally.
In a test where 512GB of objects were created and spilled, varying just the buffer size while spilling to a S3 bucket resulted in the following runtimes.
Buffer Size | Runtime (s)
-- | --
Default | 3221.865916
256KB | 1758.885839
1MB | 748.226089
10MB | 526.406466
100MB | 494.830513
Based on these results, a default buffer size of 1MB has been added. This is the minimum buffer size used by AWS Kinesis Firehose, a streaming service for S3. On systems with larger availability, it is good to configure a larger buffer size.
For processes that reach the throughput limits provided by S3, we can remove that bottleneck by supporting more prefixes/buckets. These impacts are less noticeable as the performance gains from using a large buffer prevent us from reaching a bottleneck. The following runtimes were achieved by spilling 512GB with a 1MB buffer and varying prefixes.
Prefixes | Runtime (s)
-- | --
1 | 748.226089
3 | 527.658646
10 | 516.010742
Together these changes enable faster large-scale object spilling.
Co-authored-by: Ubuntu <ubuntu@ip-172-31-54-240.us-west-2.compute.internal>
Mainly the following things:
- This PR deletes the proto cache on RuntimeEnv, ensuring that the user's modification of RuntimeEnv can take effect in the Proto message.
- validate whole runtime env when serialize runtime_env.
- overload method `__setitem__` to parse and validate field when it has to modify.
Separate out the conversion of pandas dataframe to torch tensor in a utility function so that the same logic can be used in other places in Ray ML (for example during inference).
As discussed,
- Removes ConvertibleToTrainable interface and makes as_trainable part of the Trainer interface
- Moves Trainer interface to ray.ml.trainer from ray.ml.train.trainer
The concept of a Serve Application, a data structure containing all information needed to deploy Serve on a Ray cluster, has surfaced during recent design discussions. This change introduces a formal Application data structure and refactors existing code to use it.
This PR exposes the new checkpoint interface, implemented in #22691, to end users. It does this by replacing the old external facing TrialCheckpoint class with a merged class that supports the old TrialCheckpoint API (upload, download, save) as well as the new Checkpoint API.
With this PR, users can use the new Checkpoint interface for downstream processing of their Ray Tune results. In a follow-up PR, the new Checkpoint interface will be used internally within Ray Tune and Train for bookkeeping, however, that is not required to unblock the Ray ML use case.
Horovod updated the attributes of DistributedTrainableCreator and args to create Horovod RayExecutor.
horovod/horovod@a729ba7
The major issue is Horovod deprecated "slot" concept, use "worker" instead, which is more consistent with Generic Ray worker. The issue is currently blocking Uber DL trainers to use raytune.
This commit updates the Horovod RayExecutor init args.
Co-authored-by: Kai Fricke <kai@anyscale.com>
This PR consists of the following clean-up items for KubeRay autoscaler integration:
Remove the docker/kuberay directory
Move the Python files formerly in docker/kuberay to the autoscaler directory.
Use a rayproject/ray image for the autoscaler.
Add an entry point for the kuberay autoscaler to scripts.py. Use the entry point in the example config.
Slightly simplify the code that starts the autoscaler.
Ray versions are updated to Ray 1.11.0, which will be officially released within the next couple of days.
By default, Ray >= 1.11.0 runs without Redis. References to Redis are removed from the example config.
Add the autoscaler configuration test to the CI.
Update development documentation to reflect the changes in this PR.
`test_deploy` has become [flakey](https://flakey-tests.ray.io/#) due to timeout. Since `test_deploy` is already a "large" test, this change splits it into two testing files instead of simply increasing the timeout.
This PR splits up the changes in #22393 and introduces an implementation of the ML Checkpoint interface used by Ray Tune.
This means, the TuneCheckpoint class implements the to/from_[bytes|dict|directory|object_ref|uri] conversion functions, as well as more high-level functions to transition between the different TuneCheckpoint classes. It also includes test cases for Tune's main conversion modes, i.e. dict - intermediate - dict and fs - intermediate - fs.
These changes will be the basis for refactoring the tune interface to use TuneCheckpoint objects instead of TrialCheckpoints (externally) and instead of paths/objects (internally).
* refactor resource data structure in gcs
* fix comment
* fix lint error
* fix
* DISABLED_TestRejectedRequestWorkerLeaseReply as it depends on the update of normal task
Co-authored-by: 黑驰 <senlin.zsl@antgroup.com>
Follow-up to #22748, enabling tests in CI.
Conditions: A new RAY_CI_ML_AFFECTED condition is added for this test suite. The package currently depends on Ray Data, and will be triggered accordingly.
Dependencies: Adding DATA_PROCESSING_TESTING dependencies (set for install-dependencies.sh) for now.
test_plasma_unlimited::test_task_unlimited is flaky because one of the assertions is race-y and can trigger after the condition is no longer true (see #22883). This fixes the flake by:
- adding an assertion in between two object allocations to force the object store queue to flush
- keeping one of the ObjectRefs in scope to make sure that the object is still fallback-allocated by the time we reach the failing assertion
