Chrome/Edge high memory usage for Databricks tabs.

[enhancederroruk]

Is it normal for Databricks tabs to be using such high memory?
The Chrome example I just got a screenshot of was this (rounded up/down)...

3 x Databricks tabs for one user, sized at
6gb, 4.5gb, and 2gb

Total = 12.5gb

I know it gets higher than this too, I've just not caught it in the act yet.

It gets to the point that Chrome/Edge uses up all the memory and eventually crashes.

Is Databricks known for consuming all the memory in browsers? ie: memory leaking.

[enhancederroruk]

Anyone?

[AlexG]

Having the issue for a while now. Open Databricks notebook tabs routinely take up 3GB of memory each.

[OliverD]

Boosting. I have the same issue and it's getting out of hand.

[datadrivenangel]

This is an issue for me on Edge on windows 11 for Azure Databricks. Gets in the way of work when having a few long open databricks tabs each consume gigabytes of ram.

The funniest part is that the delta tables getting processed for this particular notebook are ~10MB, so the browser memory to data ratio is ~100x!

[YT]

We are considering moving our data science development to Google Colab because of this. 3GB ram usage on Databaricks vs a 300MB for a similar notebook on Colab. It makes developing anything other than something pretty basic impossible.

[OliverD]

Have you beat with over 4!

[MateusPCardoso]

Lately, I've noticed that Databricks is consuming a lot of memory (from my local machine) in the Chrome tab. I see memory spikes especially when I'm using the SQL editor extensively — at some point, there's even a noticeable delay between typing and the response in the browser.

[StijnV]

Has there been any response on this at all? The RAM usage on my local machine, even for a single Databricks tab in Chrome, fully saturates 16gb of system RAM and steadily continues to grow up to 43gb incl. page file.

At this point the continuous page file usage makes the entire machine practically inoperable, with UI elements no longer working and simple Copy-Paste actions taking multiple seconds to complete. 

I think this can safely be classified as a memory leak. 

[szymon_dybczak]

Hi @StijnV ,

Yes, check below thread. Databricks is working on fix this issue:

Solved: Why Does Azure Databricks Consume So Much Memory W... - Databricks Community - 131322

[StijnV]

@szymon_dybczak ah, that's good to know. 

Thanks for the quick response, appreciate it! 

[szymon_dybczak]

No problem @StijnV . I hope that they will release fix soon because I also noticed high RAM usage 🙂

[datacorgo24]

+1 on SQL Editor specifically causing terrible RAM issues... Reached over 14GB this morning just leaving SQL Editor open... my other databricks tab with notebook open only consumed 300MB

[szymon_dybczak]

Wow, that's crazy RAM consumption 😯

[OliverD]

I mean this ticket has been open for two years. I'm just glad the problem has gotten bad enough for them to hopefully take it seriously. Because it has only gotten worse and worse.