# Uncategorized **URL:** https://discuss.ray.io/c/uncategorized/1.md?page=1 [Latest](https://discuss.ray.io/latest.md) · [Categories](https://discuss.ray.io/categories.md) · [Tags](https://discuss.ray.io/tags.md) **Page:** 2 --- ## [Multiple deployments on Ray service gets stuck on pending placement group](https://discuss.ray.io/t/multiple-deployments-on-ray-service-gets-stuck-on-pending-placement-group/23290)
**Author:** [@Antonio\_B](https://discuss.ray.io/u/Antonio_B)\ **Replies:** 1\ **Last updated:** [October 31, 2025, 6:30am UTC](https://discuss.ray.io/t/multiple-deployments-on-ray-service-gets-stuck-on-pending-placement-group/23290 "2025-10-31T06:30:57Z")
Hello, I’m trying to deploy one ray service on an AKS with multiple deployments deployed through LLMConfig, but I always get the worker pending, because ray tries to schedule more than 1 GPU. {'CPU': 2.0, 'GPU': 0.1}: 1… --- ## [Ray multiplexing for higher concurrency](https://discuss.ray.io/t/ray-multiplexing-for-higher-concurrency/23273)
**Author:** [@manickavela29](https://discuss.ray.io/u/manickavela29)\ **Replies:** 1\ **Last updated:** [October 27, 2025, 6:43am UTC](https://discuss.ray.io/t/ray-multiplexing-for-higher-concurrency/23273 "2025-10-27T06:43:44Z")
Hi Ray Team, I have deployed a Ray Service with model mulitplexing. Avg per model latency : 30ms Avg per model loadtime : 2sec The system is stable when i have till 200 concurrent request, but latency slowly con… --- ## [KeyError: 'advantages' in PPO MARL](https://discuss.ray.io/t/keyerror-advantages-in-ppo-marl/23257)
**Author:** [@tangjintong](https://discuss.ray.io/u/tangjintong)\ **Replies:** 1\ **Last updated:** [October 21, 2025, 8:09am UTC](https://discuss.ray.io/t/keyerror-advantages-in-ppo-marl/23257 "2025-10-21T08:09:06Z")
I use ray 2.50.1 to implement a MARL model using PPO. However, I meet the following problem: 'advantages' KeyError: 'advantages' During handling of the above exception, another exception occurred: File "/home/tangj… --- ## [Ray normal DAG vs Compiled DAG](https://discuss.ray.io/t/ray-normal-dag-vs-compiled-dag/23247)
**Author:** [@Monil](https://discuss.ray.io/u/Monil)\ **Replies:** 41\ **Last updated:** [October 18, 2025, 5:52am UTC](https://discuss.ray.io/t/ray-normal-dag-vs-compiled-dag/23247 "2025-10-18T05:52:49Z")
Hello Community, I am from offline inference platform team where we trigger standard Ray Batch Inference solution as suggested. Typically what we do is we create a ray dataset with out custom datasource We invoke map\_… --- ## [Ray Compiled Graphs](https://discuss.ray.io/t/ray-compiled-graphs/23245)
**Author:** [@Monil](https://discuss.ray.io/u/Monil)\ **Replies:** 2\ **Last updated:** [October 14, 2025, 11:05pm UTC](https://discuss.ray.io/t/ray-compiled-graphs/23245 "2025-10-14T23:05:31Z")
Hello Community, I wanted to understand the Ray Compile Graph and its execution details mentioned here Lazy Computation Graphs with the Ray DAG API — Ray 2.50.0 I am looking at Ray Execution DAG for Functions. So does … --- ## [\[kuberay\]\[rayVersion\]Is the rayVersion field in kuberay no longer useful?](https://discuss.ray.io/t/kuberay-rayversion-is-the-rayversion-field-in-kuberay-no-longer-useful/23241)
**Author:** [@Jackie\_Chen](https://discuss.ray.io/u/Jackie_Chen)\ **Replies:** 1\ **Last updated:** [October 13, 2025, 3:44pm UTC](https://discuss.ray.io/t/kuberay-rayversion-is-the-rayversion-field-in-kuberay-no-longer-useful/23241 "2025-10-13T15:44:46Z")
I tried it, and when starting RayJob or RayService through kuberay, not setting the rayVersion field has no impact on the program running. So, what is the role of this rayVersion field? --- ## [AttributeError: 'SingleAgentEnvRunner' object has no attribute 'get\_policy'](https://discuss.ray.io/t/attributeerror-singleagentenvrunner-object-has-no-attribute-get-policy/22317)
**Author:** [@Hasnain\_Fareed](https://discuss.ray.io/u/Hasnain_Fareed)\ **Replies:** 6\ **Last updated:** [October 2, 2025, 7:40am UTC](https://discuss.ray.io/t/attributeerror-singleagentenvrunner-object-has-no-attribute-get-policy/22317 "2025-10-02T07:40:44Z")
I am gettng this error, its been on me from two days, and I am not able to make progress, somebody, kindly help me ‘AttributeError: ‘SingleAgentEnvRunner’ object has no attribute ‘get\_policy’’ this is the whole code ’ L… --- ## [How can I use TD3 in latest rlib version 2.49?](https://discuss.ray.io/t/how-can-i-use-td3-in-latest-rlib-version-2-49/23213)
**Author:** [@das\_dasa](https://discuss.ray.io/u/das_dasa)\ **Replies:** 1\ **Last updated:** [September 30, 2025, 8:24pm UTC](https://discuss.ray.io/t/how-can-i-use-td3-in-latest-rlib-version-2-49/23213 "2025-09-30T20:24:26Z")
I found that TD3 could be used in rllib versions before 2.7. In later versions, it was moved to rllib\_contrib, but rllib\_contrib was removed on November 6, 2024. So, where can I find TD3 now? Do I have to use an older ve… --- ## [Find specific wheel](https://discuss.ray.io/t/find-specific-wheel/23164)
**Author:** [@jjgriffin2](https://discuss.ray.io/u/jjgriffin2)\ **Replies:** 2\ **Last updated:** [September 25, 2025, 7:49pm UTC](https://discuss.ray.io/t/find-specific-wheel/23164 "2025-09-25T19:49:03Z")
I’m trying to find the following wheel: commit = “f6c819feebb8385816097919610d8d426e78da35” version = “3.0.0.dev0” to reinstall this specific version on a Mac OSx installation, but the following fails: pip install ht… --- ## [Understanding what is Ray with RunLLM](https://discuss.ray.io/t/understanding-what-is-ray-with-runllm/23141)
**Author:** [@tajshaik](https://discuss.ray.io/u/tajshaik)\ **Replies:** 0\ **Last updated:** [September 24, 2025, 6:06am UTC](https://discuss.ray.io/t/understanding-what-is-ray-with-runllm/23141 "2025-09-24T06:06:44Z")
@RunLLM what is ray? --- ## [Unstable actors on GPU](https://discuss.ray.io/t/unstable-actors-on-gpu/15939)
**Author:** [@Zhanibek\_Datbayev](https://discuss.ray.io/u/Zhanibek_Datbayev)\ **Replies:** 5\ **Last updated:** [September 19, 2025, 2:35pm UTC](https://discuss.ray.io/t/unstable-actors-on-gpu/15939 "2025-09-19T14:35:12Z")
Hey folks, I’ve been trying to solve a simple batch inference problem using ray actors: Create 8 actors each owning separate GPU Get a list of files Submit a ray task that accepts batch of files, loads them and then u… --- ## [Latest Ray C++ API example failed to run](https://discuss.ray.io/t/latest-ray-c-api-example-failed-to-run/23105)
**Author:** [@Kafka](https://discuss.ray.io/u/Kafka)\ **Replies:** 0\ **Last updated:** [September 13, 2025, 8:06am UTC](https://discuss.ray.io/t/latest-ray-c-api-example-failed-to-run/23105 "2025-09-13T08:06:19Z")
Hi，I’m a fresher using Ray Cpp API. I followed some blog steps like: pip install -U ray\[cpp\] mkdir ray-template && ray cpp --generate-bazel-project-template-to ray-template cd ray-template && sh run.sh and failed in … --- ## [Ray Cluster is restarting after 5-10 second](https://discuss.ray.io/t/ray-cluster-is-restarting-after-5-10-second/23096)
**Author:** [@vivek\_bharatia](https://discuss.ray.io/u/vivek_bharatia)\ **Replies:** 1\ **Last updated:** [September 11, 2025, 5:16pm UTC](https://discuss.ray.io/t/ray-cluster-is-restarting-after-5-10-second/23096 "2025-09-11T17:16:44Z")
We are facing a weird issue where the raycluster is killed and reprovision if the rayservice is not completely up. Th worker nodes are failing health probe. we are getting the below error Any help will be appreciated!… --- ## [Metrics Logging in offline inference](https://discuss.ray.io/t/metrics-logging-in-offline-inference/23092)
**Author:** [@Monil](https://discuss.ray.io/u/Monil)\ **Replies:** 0\ **Last updated:** [September 11, 2025, 2:58am UTC](https://discuss.ray.io/t/metrics-logging-in-offline-inference/23092 "2025-09-11T02:58:58Z")
Hello Community, I am using Ray offline inference solution as describe here End-to-end: Offline Batch Inference — Ray 2.49.1 . Wondering anyone has experimented some sort of metrics logging with offline inference. Is … --- ## [Ray + torch.distributed/DDP resource management](https://discuss.ray.io/t/ray-torch-distributed-ddp-resource-management/7621)
**Author:** [@Eisbaer](https://discuss.ray.io/u/Eisbaer)\ **Replies:** 2\ **Last updated:** [September 10, 2025, 8:30pm UTC](https://discuss.ray.io/t/ray-torch-distributed-ddp-resource-management/7621 "2025-09-10T20:30:38Z")
Hello, I have a PyTorch training script, which handles all the nodes/processes through torch.distributed and $torchrun. I want to integrate Ray for hyperparameter tuning, however while staying with torch.distributed to… --- ## [Running Ray Job as a Flyte Task](https://discuss.ray.io/t/running-ray-job-as-a-flyte-task/23070)
**Author:** [@Monil](https://discuss.ray.io/u/Monil)\ **Replies:** 1\ **Last updated:** [September 8, 2025, 7:23pm UTC](https://discuss.ray.io/t/running-ray-job-as-a-flyte-task/23070 "2025-09-08T19:23:41Z")
Hello, I am triggering typic Ray Batch inference flow as a flyte task. In the flyte task I mention a container image that has pytorch version 2.6 lib dependency. Ray job invokes a Ray Actor on each node in Ray Cluster … --- ## [Ray Offline Inference: Division by Zero error](https://discuss.ray.io/t/ray-offline-inference-division-by-zero-error/23076)
**Author:** [@Monil](https://discuss.ray.io/u/Monil)\ **Replies:** 1\ **Last updated:** [September 8, 2025, 7:18pm UTC](https://discuss.ray.io/t/ray-offline-inference-division-by-zero-error/23076 "2025-09-08T19:18:55Z")
Hello, I am trying to trigger offline inference flow on ray cluster and getting this error ZeroDivisionError: division by zero raise e.with\_traceback(None) from SystemException() ray/data/exceptions.py", line 89, in … --- ## [How can I debug the root cuase when StreamSplitDataIterator blocked waiting?](https://discuss.ray.io/t/how-can-i-debug-the-root-cuase-when-streamsplitdataiterator-blocked-waiting/15463)
**Author:** [@Li\_Bin](https://discuss.ray.io/u/Li_Bin)\ **Replies:** 2\ **Last updated:** [September 5, 2025, 1:54pm UTC](https://discuss.ray.io/t/how-can-i-debug-the-root-cuase-when-streamsplitdataiterator-blocked-waiting/15463 "2025-09-05T13:54:06Z")
I am using Ray data with Ray train. If only one worker, it works fine. But if more than 1, the system will hang there just like dead. The final messages output look like this : (SplitCoordinator pid=64863) StreamSplitD… --- ## [When I run a ray-cpp(v-2.31.0) project,I find the task just in the local node, it don't spred](https://discuss.ray.io/t/when-i-run-a-ray-cpp-v-2-31-0-project-i-find-the-task-just-in-the-local-node-it-dont-spred/23072)
**Author:** [@youwu699](https://discuss.ray.io/u/youwu699)\ **Replies:** 1\ **Last updated:** [September 5, 2025, 6:45am UTC](https://discuss.ray.io/t/when-i-run-a-ray-cpp-v-2-31-0-project-i-find-the-task-just-in-the-local-node-it-dont-spred/23072 "2025-09-05T06:45:39Z")
\#include \ #include \ #include \ #include \ #include \ #include \ #include \ double integrate(double start, double end) { char hostname\[256\]; gethos… --- ## [Ray not autoscaling with Ray Data and submitted using jobs api](https://discuss.ray.io/t/ray-not-autoscaling-with-ray-data-and-submitted-using-jobs-api/23066)
**Author:** [@rahul-chandra-sanas](https://discuss.ray.io/u/rahul-chandra-sanas)\ **Replies:** 0\ **Last updated:** [September 4, 2025, 11:19am UTC](https://discuss.ray.io/t/ray-not-autoscaling-with-ray-data-and-submitted-using-jobs-api/23066 "2025-09-04T11:19:54Z")
The number of actors dont’t increase more than the head node. Earlier I was using ray client to submit jobs. Now I have changed it to use the ray jobs API as this seems to be the recommended way in production. While usin… --- ## [Deploying Ray Service on Local (MacOS) Kubernetes cluster using kind](https://discuss.ray.io/t/deploying-ray-service-on-local-macos-kubernetes-cluster-using-kind/23062)
**Author:** [@Yohanes\_Jhouma](https://discuss.ray.io/u/Yohanes_Jhouma)\ **Replies:** 0\ **Last updated:** [September 4, 2025, 7:40am UTC](https://discuss.ray.io/t/deploying-ray-service-on-local-macos-kubernetes-cluster-using-kind/23062 "2025-09-04T07:40:54Z")
Hi I am currently trying to explore the RayServe for inference use case. However when I tried to deploy the service on my local kubernetes cluster using kind the head pod flag as unhealthy and did not start the dashboard… --- ## [Windows & Python 3.13 Support](https://discuss.ray.io/t/windows-python-3-13-support/23061)
**Author:** [@hhmdb](https://discuss.ray.io/u/hhmdb)\ **Replies:** 0\ **Last updated:** [August 31, 2025, 10:18am UTC](https://discuss.ray.io/t/windows-python-3-13-support/23061 "2025-08-31T10:18:08Z")
I noticed there are no wheels for this combination. Is there support for Windows & Python 3.13? If not, is this planned and is there somewhere to get updates on progress? --- ## [ModuleNotFoundError: No module named 'src.utils'](https://discuss.ray.io/t/modulenotfounderror-no-module-named-src-utils/23059)
**Author:** [@lunasaw](https://discuss.ray.io/u/lunasaw)\ **Replies:** 0\ **Last updated:** [August 29, 2025, 5:23pm UTC](https://discuss.ray.io/t/modulenotfounderror-no-module-named-src-utils/23059 "2025-08-29T17:23:20Z")
Has anyone encountered this problem? When I was released by ray server, the serialization package was normal, but when the work was running, I couldn’t find this src.util. Is there any problem with setting parameters for… --- ## [Tuner restore do not loads the existing files in the trial directory correctly.. get\_trial\_dir() points to new location when restored](https://discuss.ray.io/t/tuner-restore-do-not-loads-the-existing-files-in-the-trial-directory-correctly-get-trial-dir-points-to-new-location-when-restored/23011)
**Author:** [@KKNakka](https://discuss.ray.io/u/KKNakka)\ **Replies:** 1\ **Last updated:** [August 18, 2025, 3:03pm UTC](https://discuss.ray.io/t/tuner-restore-do-not-loads-the-existing-files-in-the-trial-directory-correctly-get-trial-dir-points-to-new-location-when-restored/23011 "2025-08-18T15:03:22Z")
Hi, I am facing issues with ray tune with probably newer versions… lets say i start raytune with below mincode.. import os import time from ray import tune from ray.air import session, RunConfig, CheckpointConfig impor… --- ## [Manual Scale Ray Deployment](https://discuss.ray.io/t/manual-scale-ray-deployment/22968)
**Author:** [@RR\_R](https://discuss.ray.io/u/RR_R)\ **Replies:** 0\ **Last updated:** [August 8, 2025, 7:15pm UTC](https://discuss.ray.io/t/manual-scale-ray-deployment/22968 "2025-08-08T19:15:37Z")
Due to several reasons, no kubernetes or any cloud based platform available. I want to achieve the followings: Given I have 3 nodes, I want all of nodes to run the same ray deployment (serving the same endpoint) to for… --- ## [\`dataset.write\_bigquery\` infers \`NULLABLE\` mode](https://discuss.ray.io/t/dataset-write-bigquery-infers-nullable-mode/22920)
**Author:** [@Orson\_Adams](https://discuss.ray.io/u/Orson_Adams)\ **Replies:** 0\ **Last updated:** [July 31, 2025, 8:09pm UTC](https://discuss.ray.io/t/dataset-write-bigquery-infers-nullable-mode/22920 "2025-07-31T20:09:28Z")
We’re attempting to use the write\_bigquery method to write our pipeline results to BQ. We have an existing table (with the write modes that make sense to use). How can we ensure that the write modes (NULLABLE, REQUIRED … --- ## [AttributeError: 'NoneType' object has no attribute 'outputs' on tqdm\_ray.\_\_del\_\_ in IPython/Jupyter](https://discuss.ray.io/t/attributeerror-nonetype-object-has-no-attribute-outputs-on-tqdm-ray-del-in-ipython-jupyter/22810)
**Author:** [@wxie2013](https://discuss.ray.io/u/wxie2013)\ **Replies:** 0\ **Last updated:** [July 8, 2025, 3:20pm UTC](https://discuss.ray.io/t/attributeerror-nonetype-object-has-no-attribute-outputs-on-tqdm-ray-del-in-ipython-jupyter/22810 "2025-07-08T15:20:16Z")
Description: When using tqdm\_ray.tqdm within an IPython or Jupyter environment, an AttributeError: 'NoneType' object has no attribute 'outputs' is consistently raised and ignored (as an Exception ignored in:) during prog… --- ## [Ray distributed debugger failed when accepting connection](https://discuss.ray.io/t/ray-distributed-debugger-failed-when-accepting-connection/22802)
**Author:** [@Tao\_Wang](https://discuss.ray.io/u/Tao_Wang)\ **Replies:** 0\ **Last updated:** [July 5, 2025, 3:42am UTC](https://discuss.ray.io/t/ray-distributed-debugger-failed-when-accepting-connection/22802 "2025-07-05T03:42:37Z")
I’m using ray debugger on vscode, I followed the official instructions to set breakpoint and run my code, the code stops at the breakpoint, then I use the ray distributed debugger to connect to that port by adding one cl… --- ## [Ray debug not attached](https://discuss.ray.io/t/ray-debug-not-attached/22801)
**Author:** [@Tao\_Wang](https://discuss.ray.io/u/Tao_Wang)\ **Replies:** 0\ **Last updated:** [July 5, 2025, 3:22am UTC](https://discuss.ray.io/t/ray-debug-not-attached/22801 "2025-07-05T03:22:25Z")
I’m using vscode with ray distributed debuger, I followed the instructions to set with my own project, but the process stuck at here and the ray debugger never get an attached process as expected. I have tried downgrade… --- ## [Uv + ray in example is not working](https://discuss.ray.io/t/uv-ray-in-example-is-not-working/22783)
**Author:** [@JasonZhu1313](https://discuss.ray.io/u/JasonZhu1313)\ **Replies:** 1\ **Last updated:** [July 2, 2025, 1:22am UTC](https://discuss.ray.io/t/uv-ray-in-example-is-not-working/22783 "2025-07-02T01:22:49Z")
[Previous page](https://discuss.ray.io/c/uncategorized/1.md) [Next page](https://discuss.ray.io/c/uncategorized/1.md?page=2)