# What is the develop progress of connect multiple clusters or task level workdir?

**URL:** <https://discuss.ray.io/t/what-is-the-develop-progress-of-connect-multiple-clusters-or-task-level-workdir/8954>\
**Category:** Ray Core\
**Created:** [January 13, 2023, 7:39am UTC](https://discuss.ray.io/t/what-is-the-develop-progress-of-connect-multiple-clusters-or-task-level-workdir/8954 "2023-01-13T07:39:50Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![tarjintor](https://sea2.discourse-cdn.com/flex020/user_avatar/discuss.ray.io/tarjintor/32/3552_2.png) [@tarjintor](https://discuss.ray.io/u/tarjintor)\
**Post date:** [January 13, 2023, 7:39am UTC](https://discuss.ray.io/t/what-is-the-develop-progress-of-connect-multiple-clusters-or-task-level-workdir/8954/1 "2023-01-13T07:39:50Z")

</div>

Hello, I want to connect multiple ray clusters and I wish I can set workdir per-actor  
I read the doc [Ray Client: Interactive Development — Ray 2.2.0](https://docs.ray.io/en/latest/cluster/running-applications/job-submission/ray-client.html#connect-to-multiple-ray-clusters-experimental)  
It say this is still experimental。  
I will describe my use case below,and some doubts I have:  
1.I am design a framework for ml, there is a component ‘job\_consumer’ can consume task in json format from some queue,in these json,have information about which ray cluster this task should run on, for the purpose like resource isolate,but now I guess I can’t use this feature?  
2.Or even two task submitted at same cluster,these task have different workdir,but as I read from the doc,we can only set workdir when the first time we call ray.init

I don’t know how to deal with scenario 1,but for scenario 2, I use a lot of copying move files to same fixed workdir with renaming, and it works

Thanks all the contributors of ray project, and I benifit alot  
I don’t mean to push you finish the features  
I just need some work around, would please me give some advice, thank you again

---

<div class="post-metadata">

**Author:** ![Stephanie\_Wang](https://sea2.discourse-cdn.com/flex020/user_avatar/discuss.ray.io/stephanie_wang/32/18_2.png) [@Stephanie\_Wang](https://discuss.ray.io/u/Stephanie_Wang)\
**Post date:** [January 17, 2023, 3:58pm UTC](https://discuss.ray.io/t/what-is-the-develop-progress-of-connect-multiple-clusters-or-task-level-workdir/8954/2 "2023-01-17T15:58:27Z")

</div>

1 - There is no native way to connect multiple Ray clusters, so you’ll have to build out some type of router yourself. For production settings, it’s recommended to use [Kubernetes for Ray deployment](https://docs.ray.io/en/latest/cluster/kubernetes/index.html), and this should make it easier to manage and route to multiple Ray clusters.

2 - For managing different working directories (and really any other kind of environment dependency), you can use [Ray jobs](https://docs.ray.io/en/latest/cluster/running-applications/job-submission/quickstart.html#jobs-quickstart).

Please note that Ray Client is meant for interactive development, similar to an ipython experience, so it’s probably not what you’re looking for.