All Khoros Community Forums Support (Not for Databricks Product Questions) posts

Re: post being marked as spam

Advika — Mon, 31 Aug 2026 10:16:02 GMT

Apologies for the inconvenience. Due to a system glitch, your post was not published and remained invisible on the forum. This has now been resolved, and your article is now live on the Community!

Thanks for submitting your project and sharing it with the Community.🙂

Re: Post marked at SPAM incorrectly

Advika — Mon, 31 Aug 2026 08:57:05 GMT

Hello @saketsuman,

Apologies for the inconvenience. Due to a system glitch, your post was not published and remained invisible on the forum. This has now been resolved, and your article is now live on the Community!

Thanks for submitting your project and sharing it with the Community.🙂

Post marked at SPAM incorrectly

saketsuman — Sun, 30 Aug 2026 19:52:29 GMT

Hi

Building a Databricks Solutions Architect Genie Ag... - Databricks Community - 166825 Please check why this is being marked as SPAM..its a Genie contest entry
Saket

post being marked as spam

VivekKumar — Sun, 30 Aug 2026 16:11:55 GMT

Hi,

My databricks app challenege is being tagged as SPAM. Please help.

https://community.databricks.com/t5/community-articles/genie-powered-app-challege-bushfire-exposure-across-victoria-s/m-p/166804#M1481

Re: Request to Merge Two Databricks Community Accounts

AOJ — Mon, 17 Aug 2026 12:16:27 GMT

Appreciate the guidance and the edit @Advika. Rookie mistake!

Re: Request to Merge Two Databricks Community Accounts

Advika — Mon, 17 Aug 2026 11:58:48 GMT

Hello @AOJ,

Regarding this request, could you please send an email to community@databricks.com from your corporate email address and CC your personal email address? We’ll review the request and get back to you with what’s possible.
Thank you!

Request to Merge Two Databricks Community Accounts

AOJ — Mon, 17 Aug 2026 11:50:42 GMT

Hello Databricks Community Support,

I currently have two Databricks Practitioner Community accounts registered under different email addresses and would like to consolidate them into one account.

Account A: Personal email

I would like this account to remain as my primary/surviving Community account.

Account B: Company email

Account B contains a sizeable amount of my existing Community activity, including posts, replies, kudos, badges and other profile history.

I would like to request that Account B be merged into Account A, with as much of the Community history, reputation, badges, posts and other activity as possible retained under Account A.

The reason for the migration is that Account B uses my employer-provided email address, and I would like my long-term Databricks Community identity to be associated with my personal email so that I do not lose access if my employment account changes in the future.

I also currently use Account A's email address for my Databricks Free Edition account/workspace. My request is specifically for the Databricks Practitioner Community accounts, and I do not want any changes made to my Free Edition workspace or platform account.

Could you please confirm:

Whether the two Community accounts can be merged
Whether the posts, replies, kudos, badges, rank and other Community history from Account B can be retained under Account A
Whether there is anything else you need from me to verify ownership of both accounts

Thank you.
[Personal Details Removed By MOD]

Re: Reducing Time to Help in the Databricks Community

MandyR — Wed, 05 Aug 2026 18:27:03 GMT

HI @Brahmareddy ! Apologies for the delay in reply.

We love this idea and have lots of plans for updating the Community in the coming year, including agentic support. Can't say much yet until it's official, but stay tuned for more news in this area!

Thanks again for your support and community love!

Re: Reducing Time to Help in the Databricks Community

vedhagiri — Fri, 31 Jul 2026 05:12:06 GMT

This is a really great idea. Using a Databricks Genie–style assistant for our Community makes total sense and would bring so much value. I especially loved your point that this should support human interaction, not replace real discussions. To build on your post, here are a few potential challenges we might face with an AI assistant and how we can overcome them:

Risk of Outdated Answers (RAG Freshness)

Challenge: Databricks features and syntaxes update very quickly across different DBR (Databricks Runtime) versions. An older accepted solution might not work on modern Lakehouse /Unity Catalog setups.

Solution: The assistant should prioritise content tagged with recent DBR versions and clearly mention the publication/updated date alongside the suggested solution.

Hallucination & Misleading Suggestions

Challenge: If the LLM generates answers directly instead of just retrieving posts, it might occasionally hallucinate non-existent PySpark functions or configurations.

Solution: Keep it grounded. Instead of letting it generate unverified code, it should function purely as a smart retriever—citing verified Community accepted solutions, official documentation, and MVP blogs with direct links.

"Hallway" Blockades for New/Edge-Case Issues

Challenge: Users with genuine new bugs or rare edge cases might get annoyed if forced through an AI suggestion layer before they can post.

Solution: Always keep a clear, one-click "Post Question Anyway" option readily available so users don't feel blocked when their issue is unique.

Data Privacy & Code Leaks

Challenge: While asking for help, users might accidentally paste sensitive company data, proprietary SQL, or access keys into the prompt.

Solution: Build a simple client-side check that warns users if potential credentials, IP addresses, or sensitive tokens are detected in their input before submitting.

It looks like you’ve already given this a lot of thought! I’d love to see the POC ideas or workflows you mentioned. Hopefully, the Community Team takes a look at this and shares if something similar is on the roadmap!

Re: Reducing Time to Help in the Databricks Community

Brahmareddy — Tue, 28 Jul 2026 12:25:01 GMT

Thank you Advika for considering the idea and sharing it with the team. I believe better content discovery could help members find relevant answers faster and reduce repeated questions. I would be happy to contribute, share more ideas, or support any early exploration.

Re: Reducing Time to Help in the Databricks Community

Brahmareddy — Tue, 28 Jul 2026 12:18:10 GMT

Thank you, Surya. Great points.

Re: Reducing Time to Help in the Databricks Community

Advika — Tue, 28 Jul 2026 09:16:19 GMT

Thanks for sharing this thoughtful idea, @Brahmareddy 🙌! Really appreciate you taking the time to write it up and explain the potential benefits.

Improving how members discover existing Community content is definitely an important area, and we'll share your feedback with the team for review. Thanks again for contributing your ideas, and please keep the suggestions coming!✌️

Re: Reducing Time to Help in the Databricks Community

suryaprayaga — Sat, 25 Jul 2026 08:32:37 GMT

Yes totally agree with this. Posts can be sub-categorized, tags can be created and responses can be scored based on how effective or important they are (even though this is already a feature currently).

I did a similar exercise for Amazon earlier where reviews are only sorted based on the rating & date posted and not on any other feature. There are many other intelligent ways of sorting and rating posts, creating scores to them and show only those that are relevant by creating AI agents that can give what the user wants.

I can contribute too.

Reducing Time to Help in the Databricks Community

Brahmareddy — Fri, 24 Jul 2026 04:35:35 GMT

Hi @MandyR , @Advika , and team

I wanted to share one idea for the Databricks Community. Can we explore a Genie-like experience for databricks community?

The Community already has many useful answers, accepted solutions, blogs, MVP articles, and discussions. But sometimes members may not find the right answer quickly. They may not know the exact keywords to search, or they may post a new question even when a similar answer already exists.

A Genie-like community assistant can help members find the most relevant answers faster.

For example, if someone asks about Unity Catalog permissions, Delta tables, PySpark errors, Auto Loader, Databricks SQL, Lakeflow, MLflow, or any Databricks feature, the assistant can search the Community content and suggest the best matching answers, accepted solutions, related articles, and documentation links.

This can help in many ways:

It can reduce duplicate questions.

It can reduce the time community members spend answering repeated questions.

It can help users get support faster.

It can make existing accepted solutions easier to find.

It can guide users to ask better questions with the right title, tags, error message, code snippet, and context.

In my view, this should not replace community discussions. It should support them.

The assistant can first suggest matching answers. If the user still needs help, it can guide them to post a better question with all the needed details.

The goal is simple: help every member reach the right answer faster and make the Community knowledge base more useful over time.

I am happy to ideate more and share inputs from a user and contributor perspective, including possible workflows, use cases, and POC ideas.

Would this be something the Community team can explore?

Re: Inference Lakehouse Monitor , how to create a monitor on more than one prediction

Sureshkrishna — Fri, 26 Jun 2026 06:03:29 GMT

This is a known limitation of the Databricks Lakehouse Monitoring API. Here's what you need to know and the workarounds:

The Limitation:
"prediction_col" must be a double type (not a string, and not a list/array)
Reference : https://medium.com/marvelous-mlops/streamlining-ml-model-monitoring-with-databricks-lakehouse-and-inference-tables-f0a99ab03291

The API only accepts a single scalar column. There's no native support for multi-output predictions like quantile regression.
Workaround Options:
1. Create separate monitors per quantile (Recommended) : Unpivot your quantile predictions into separate tables, each with its own scalar "prediction_col", and create one monitor per quantile

e.g., for quantiles [0.1, 0.5, 0.9]
for quantile, col_name in [("q10", "pred_q10"), ("q50", "pred_q50"), ("q90", "pred_q90")]:
spark.sql(f"""
CREATE OR REPLACE TABLE {CATALOG}.{SCHEMA}.inference_log_{quantile}
AS SELECT *, {col_name} AS prediction FROM {TABLE_NAME}
""")
w.quality_monitors.create(
table_name=f"{CATALOG}.{SCHEMA}.inference_log_{quantile}",
inference_log=MonitorInferenceLog(
granularities=GRANULARITIES,
timestamp_col=TIMESTAMP_COL,
model_id_col=MODEL_ID_COL,
prediction_col="prediction", 
problem_type=MonitorInferenceLogProblemType.PROBLEM_TYPE_REGRESSION,
label_col=LABEL_COL,
),
output_schema_name=f"{CATALOG}.{SCHEMA}",
assets_dir=f"{ASSETS_DIR}/{quantile}",
)

2. Use "custom_metrics" for the extra quantiles

Keep one quantile as the primary "prediction_col" (e.g., median), and track the others via custom metrics:

from databricks.sdk.service.catalog import MonitorMetric, MonitorMetricType
custom_metrics = [
MonitorMetric(
name="mean_pred_q10",
input_columns=["pred_q10"],
definition="avg(:pred_q10)",
output_data_type="double",
type=MonitorMetricType.CUSTOM_METRIC_TYPE_AGGREGATE,
),
MonitorMetric(
name="mean_pred_q90",
input_columns=["pred_q90"],
definition="avg(:pred_q90)",
output_data_type="double",
type=MonitorMetricType.CUSTOM_METRIC_TYPE_AGGREGATE,
),
]

w.quality_monitors.create(
table_name=TABLE_NAME,
inference_log=MonitorInferenceLog(
prediction_col="pred_q50", ]
...
),
custom_metrics=custom_metrics,
...
)

3. Use Snapshot analysis instead

If you don't need the inference-specific accuracy metrics (MSE, etc.), switch to a Snapshot monitor. It will profile all your quantile columns as regular numeric columns and still compute drift/distribution stats across all of them no "prediction_col" constraint.

Re: Inference Lakehouse Monitor , how to create a monitor on more than one prediction

AustinZaccor — Tue, 05 May 2026 01:27:09 GMT

Unfortunately this can only take a single column. To monitor a list of predictions for quantile regression, you need to flatten the list into distinct columns before creating the monitor.

Re: Most appropriate way to re-highlight a post on Databricks Community?

ChristianRRL — Mon, 04 May 2026 16:34:42 GMT

Perfect, I'll keep this in mind next time. Thanks!

Re: Most appropriate way to re-highlight a post on Databricks Community?

Advika — Mon, 04 May 2026 16:30:41 GMT

Hello @ChristianRRL,

I’d suggest adding a quick comment on your post, it helps bring it back up on the homepage so it gets more visibility.

Also, this post would be a better fit for the Community Forums Support page, since the discussion is about the community itself. 🙂

Most appropriate way to re-highlight a post on Databricks Community?

ChristianRRL — Mon, 04 May 2026 16:10:56 GMT

Hi there, wasn't really sure what the best way/place to post this. Please feel free to give me feedback on how to best do this in the future!

I posted a topic on Friday, and I think since it was right before the weekend my post may have been buried. Normally what would be the best way to bring attention to a post that hasn't been reviewed yet?

Declarative Automation Bundle - Reusable job_clust... - Databricks Community - 155976

Re: What does 50% even mean?

Steve_Lyle_BPCS — Thu, 30 Apr 2026 19:15:07 GMT

❤️