cancel
Turn on suggestions
Showing results for 
Search instead for 
Did you mean: 
Help
Community Platform Discussions
Connect with fellow community members to discuss general topics related to the Databricks platform, industry trends, and best practices. Share experiences, ask questions, and foster collaboration within the community.
cancel
Turn on suggestions
Showing results for 
Search instead for 
Did you mean: 

Forum Posts

Start a conversation
SB93
by • New Contributor II
  • 194 Views
  • 1 replies
  • 0 kudos

Help Needed: Executor Lost Error in Multi-Node Distributed Training with PyTorch

Hi everyone,I'm currently working on distributed training of a PyTorch model, following the example provided here. The training runs perfectly on a single node with a single GPU. However, when I attempt multi-node training using the following configu...

Community Platform Discussions
Reply
  • 194 Views
  • 1 replies
  • 0 kudos
View Replies
Latest Reply
cgrant
Databricks Employee
  • 0 kudos

We do not recommend using spot instances with distributed ML training workloads that use barrier mode, like TorchDistributor as these workloads are extremely sensitive to executor loss. Please disable spot/pre-emption and try again.

  • 0 kudos
manoj_2355ca
by • New Contributor III
  • 3523 Views
  • 2 replies
  • 0 kudos

cannot create external location: invalid Databricks Workspace configuration

HI AllI am trying to create databricks storage credentials , external location and catalog with terraform.cloud : AzureMy storage credentials code is working correctly . But the external location code is throwing below error when executing the Terraf...

Community Platform Discussions
azuredatabricks
Reply
  • 3523 Views
  • 2 replies
  • 0 kudos
View Replies
Latest Reply
badari_narayan
New Contributor II
  • 0 kudos

Hi @manoj_2355ca , I am also facing the same error, did you get the solution for it?

  • 0 kudos
1 More Replies
vigneshkannan12
by • New Contributor
  • 3590 Views
  • 5 replies
  • 0 kudos

typing extensions import match error

I am trying to install the stanza library and try to create a udf function to create NER tags for my chunk_text in the dataframe.Cluster Config: DBR 14.3 LTS SPARK 3.5.0 SCALA 2.12below code:def extract_entities(text    import stanza    nlp = stanza....

Community Platform Discussions
Reply
  • 3590 Views
  • 5 replies
  • 0 kudos
View Replies
Latest Reply
Optimusprime
New Contributor II
  • 0 kudos

@SaadhikaB Hi, when I run  dbutils.library.restartPython(), I get the following error 

  • 0 kudos
4 More Replies
ramisinghl01
by • New Contributor
  • 90 Views
  • 0 replies
  • 0 kudos

PYTEST: Module not found error

Hi,Apologies, as I am trying to use Pytest first time. I know this question has been raised but I went through previous answers but the issue still exists.I am following DAtabricks and other articles using pytest. My structure is simple as -tests--co...

Community Platform Discussions
Reply
  • 90 Views
  • 0 replies
  • 0 kudos
ismaelhenzel
by • Contributor
  • 2813 Views
  • 4 replies
  • 3 kudos

Failure when deploying a custom serving endpoint LLLM

I'm currently experimenting with vector search using Databricks. Everything runs smoothly when I load the model deployed in Unity Catalog into a notebook session and ask questions using Python. However, when I attempt to serve it, I encounter a gener...

ismaelhenzel_0-1715091515103.png
Community Platform Discussions
Reply
  • 2813 Views
  • 4 replies
  • 3 kudos
View Replies
Latest Reply
Usmanr000
New Contributor II
  • 3 kudos

"Deploying a custom serving endpoint for LLMs can be challenging, especially when handling model dependencies and scaling issues. Has anyone found a reliable workaround for deployment failures? Also, for those looking for updates on government assist...

  • 3 kudos
3 More Replies
unj1m
by • New Contributor III
  • 3577 Views
  • 4 replies
  • 0 kudos

Resolved! What version of Python is used for the 16.1 runtime

I'm trying to create a spark udf for a registered model and getting:Exception: Python versions in the Spark Connect client and server are different. To execute user-defined functions, client and server should have the same minor Python version. Pleas...

Community Platform Discussions
Reply
  • 3577 Views
  • 4 replies
  • 0 kudos
View Replies
Latest Reply
AndriusVitkausk
New Contributor III
  • 0 kudos

Does this mean that:1. A new dbx runtime comes out2. Serverless compute automatically switches to the new runtime + new python version3. Any external environments that use serverless ie, local VScode / CICD environments also need to upgrade their pyt...

  • 0 kudos
3 More Replies
nikhil_2212
by • New Contributor
  • 192 Views
  • 1 replies
  • 0 kudos

Lakehouse monitoring metrices tables not created automatically.

Hello,I have an external table created in databricks unity catalog workspace and trying to "Create a monitor" for the same from quality tab.While creating the same the dashboard is getting created however the two metrices tables "profile" & "drift" a...

Community Platform Discussions
Reply
  • 192 Views
  • 1 replies
  • 0 kudos
View Replies
Latest Reply
Advika
Databricks Employee
  • 0 kudos

Hello @nikhil_2212! It looks like this post duplicates the one you recently posted. A response has already been provided to the Original post. I recommend continuing the discussion in that thread to keep the conversation focused and organised.

  • 0 kudos
VijayP
by • New Contributor
  • 180 Views
  • 1 replies
  • 0 kudos

Stream processing large number of JSON files and handling exception

application writes several JSON  (small)  files and the expected volumes of these files are high ( Estimate: 1 million during the peak season in a hourly window) . As per current design, these files are streamed through Spark Stream and we use autolo...

Community Platform Discussions
Reply
  • 180 Views
  • 1 replies
  • 0 kudos
View Replies
Latest Reply
cgrant
Databricks Employee
  • 0 kudos

We have customers that read millions of files per hour+ using Databricks Auto Loader. For high-volume use cases, we recommend enabling file notification mode, which, instead of continuously performing list operations on the filesystem, uses cloud nat...

  • 0 kudos
mrstevegross
by • Contributor
  • 631 Views
  • 4 replies
  • 1 kudos

Resolved! How best to measure the time-spent-waiting-for-an-instance?

I'm exploring using an instance pool. Can someone clarify for me which job event log tells me the time-spent-waiting-for-an-instance? I've found 2 candidates:1. The delta between "waitingForCluster" and "started" on the "run events" log, accessible v...

mrstevegross_0-1741800626468.png mrstevegross_1-1741800790749.png
Community Platform Discussions
Reply
  • 631 Views
  • 4 replies
  • 1 kudos
View Replies
Latest Reply
julieAnderson
New Contributor II
  • 1 kudos

 System Logs or Event Timings

  • 1 kudos
3 More Replies
Forssen
by • New Contributor II
  • 264 Views
  • 2 replies
  • 1 kudos

Resolved! When is it time to change from ETL in notebooks to whl/py?

Hi!I would like some input/tips from the community regarding when is it time to go from a working solution in notebooks to something more "stable", like whl/py-files?What are the pros/cons with notebooks compared to whl/py?The way i structured things...

Community Platform Discussions
Reply
  • 264 Views
  • 2 replies
  • 1 kudos
View Replies
Latest Reply
Isi
Contributor
  • 1 kudos

Hey @Forssen ,My advice:Using .py files and .whl packages is generally more secure and scalable, especially when working in a team. One of the key advantages is that code reviews and version control are much more efficient with .py files, as changes ...

  • 1 kudos
1 More Replies
LasseL
by • New Contributor III
  • 268 Views
  • 1 replies
  • 0 kudos

Resolved! Deduplication with rocksdb, should old state files be deleted manually (to manage storage size)?

Hi, I have following streaming setup:I want to remove duplicates in streaming.1) deduplication strategy is defined by two fields: extraction_timestamp and hash (row wise hash)2) watermark strategy: extraction_timestamp with "10 seconds" interval--> R...

Community Platform Discussions
Reply
  • 268 Views
  • 1 replies
  • 0 kudos
View Replies
Latest Reply
LasseL
New Contributor III
  • 0 kudos

Found solution. https://kb.databricks.com/streaming/how-to-efficiently-manage-state-store-files-in-apache-spark-streaming-applications <-- these two parameters.

  • 0 kudos
IGRACH
by • New Contributor II
  • 418 Views
  • 6 replies
  • 2 kudos

Disable exiting current cell when moving around with keyboard arrows

Is there any way do disable exiting current cell when I move cursor around with arrows. When I press up arrow or down arrow it will exit the current cell and go to another cell. Can that functionally be disabled so when I hold up or down arrow key, c...

Community Platform Discussions
Reply
  • 418 Views
  • 6 replies
  • 2 kudos
View Replies
Latest Reply
IGRACH
New Contributor II
  • 2 kudos

Is there any place where I can put this as a request.

  • 2 kudos
5 More Replies
robbdunlap
by • New Contributor III
  • 3646 Views
  • 8 replies
  • 22 kudos

Turn Off Auto-reveal of Navigation Sidebar

I work with the navigation sidebar closed and use the stacked hamburgers symbol in the upper left to reveal it when I want. Now, if you mouse over the left edge of the browser window too slowly it will auto-reveal the navigation sidebar. I do not wan...

Community Platform Discussions
Reply
  • 3646 Views
  • 8 replies
  • 22 kudos
View Replies
Latest Reply
Advika
Databricks Employee
  • 22 kudos

I've checked with the team, and there's no way to turn this off. However, they are making adjustments to improve the experience, and a fix to refine the sidebar functionality is on the way.

  • 22 kudos
7 More Replies

Join Us as a Local Community Builder!

Passionate about hosting events and connecting people? Help us grow a vibrant local community—sign up today to get started!

Sign Up Now
Top Kudoed Authors
View all
Top