Comments (16)
Just a note that we're going to start working on this effort soon @zjorgensenbits, and will update this issue when it's ready.
from determined.
I will check it out when we update to this version and let you know. But, based on the screenshot, this is exactly what I had in mind!
from determined.
@zjorgensenbits Thanks for writing this up! I agree that this would be a useful feature. We've started thinking about the best way to implement it -- will keep you posted if we have any updates on the timeline here.
from determined.
Any thoughts/progress on this?
from determined.
Update: We'll add the ability to pick a configuration template to the notebook launch dialog and post an update once this feature ships.
from determined.
That would be super helpful, thanks!
from determined.
Hi, @zjorgensenbits and thanks for sharing your feedback. We recently shipped a new and more configurable way of launching Jupyeter Notebooks that allows you to pick a template for the Notebook configuration you're running. Does this address what you've expressed?
from determined.
Hey @zjorgensenbits have you had a chance to upgrade and test 0.16.x out?
Any feedback?
And if it's what you and/or your team were after, may we please close out this issue?
from determined.
I'll close this for now but please feel free to reopen if there are any questions or updates.
from determined.
Finally had a chance to upgrade and test this and it works great -- exactly what I was hoping for.
However, I believe I discovered a bug. If I click launch, everything works fine, but if I first click "show full config" and then click launch (even if I don't edit the config), the notebook immediately terminates. The log shows the following error: [2021-07-23, 17:13:24] test-notebook was terminated: container failed with non-zero exit code: error creating container: Error response from daemon: Duplicate mount point: /user_data/.
In our case, all of our bind mounts are defined in the master config. So it appears that if you hit show full config, it causes the bind mounts defined in the master config to be included again in the config file which causes this error.
from determined.
Thanks for the report @zjorgensenbits, we'll take a look.
from determined.
@zjorgensenbits do you mind expanding on the click launch you mention a bit?
If I click launch, everything works fine,
Is it that all the form inputs are empty like https://user-images.githubusercontent.com/12127420/122833216-6f12c200-d2a1-11eb-9e5c-ff9302874148.png or is it that some of those are changed, either at the time of launch or from a saved previous usage.
from determined.
I can select a template and resource pool and then click the Launch button at the bottom and everything works as expected. However, if instead of clicking the Launch button I instead click the "Show Full Config" button and then subsequently click the Launch button on that modal, the notebook fails to start.
from determined.
that's helpful thank you. I suspect it'd fail with the same error if you leave the fields in the form empty and click launch. We'll look into this.
from determined.
Yes, I believe the first time I tried the new feature the fields were blank and hitting launch caused a similar failure. Now though, since I've already selected options it seems to remember those options in the future.
from determined.
steps to reproduce: with the following in master config. a single bind mount should be enough as well
task_container_defaults:
bind_mounts:
- host_path: /tmp/abc
container_path: /tmp/c-abc
- host_path: /tmp/xyz
container_path: /tmp/c-xyz
the first request below fails with the reported error whereas the second one does not. we use the preview parameter to resolve the presented config in edit full config
flow
curl-da.sh "${DET_MASTER}/api/v1/notebooks" \
--data-raw '{"config":{"bind_mounts":[{"container_path":"/tmp/c-abc","host_path":"/tmp/abc","propagation":"rprivate","read_only":false},{"container_path":"/tmp/c-xyz","host_path":"/tmp/xyz","propagation":"rprivate","read_only":false}],"description":"Notebook (sincerely-unified-moray)","entrypoint":null,"environment":{"add_capabilities":null,"drop_capabilities":null,"environment_variables":{},"force_pull_image":false,"image":{"cpu":"determinedai/environments:py-3.7-pytorch-1.9-lightning-1.3-tf-2.4-cpu-2409e48","gpu":"determinedai/environments:cuda-11.1-pytorch-1.9-lightning-1.3-tf-2.4-gpu-2409e48"},"pod_spec":null,"ports":null},"resources":{"agent_label":"","devices":null,"resource_pool":"default","slots":1,"weight":1}}}'
curl-da.sh "${DET_MASTER}/api/v1/notebooks" \
--data-raw '{"config":{"resources":{"slots":1}}}'
It looks like an issue with how we're merging the config posted through the request and existing defaults in the cluster when it's an array type
from determined.
Related Issues (20)
- 🐛[bug]
- 🤔[question] add resource_pools HOT 1
- 🐛[bug] show_ssh_command error on Windows CMD: module 'os' has no attribute 'uname' HOT 5
- 🐛[bug] det CLI tool errors on Python 3.12 because it relies on distutils which was deprecated in Python 3.10 HOT 3
- 🤔[question] LOGs HOT 4
- 🤔 model registry - inference with pytorch model HOT 1
- 🐛[bug] Error Starting Up Cluster using det deploy HOT 4
- 🐛[bug] Bad ref on requirements.rst in Docs HOT 1
- 🐛[bug] Resources failed with non-zero exit code: container failed with non-zero exit code: 80 HOT 5
- 🐛[bug] Master refuses to accept agents connection HOT 4
- 🤔[question] Changing the default config path for the determined-agent.service HOT 5
- 🤔[question] Updating the default Determined-Pytorch container to 2.1/2.2 HOT 1
- 🐛[bug] Running Mnist Tutorial distributed causes Runtime Errors and Hanging behavior HOT 12
- 🤔[question] dialing to http://172.22.0.1:32862: dial tcp 172.22.0.1:32862: connect: connection refused HOT 2
- 🐛[bug] Kernel status: pending HOT 11
- 🤔[question] Where can I find the source code of the CLI? HOT 1
- 🤔[question] Can not connect to master node HOT 6
- 🤔[question] Open to updates to EKS deployment? HOT 6
- 🤔[question] How to get pod address by experiment HOT 1
- Integrated with VTable HOT 1
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from determined.