memory issues - databricks

pavanb
New Contributor II

Hi All, All of a sudden in our Databricks dev environment, we are getting exceptions related to memory such as out of memory , result too large etc.

Also, the error message is not helping to identify the issue.

Can someone please guide on what would be the starting point to look into it.

I am getting this issue while reading a json file and dumping it into a dataframe.

Hubert-Dudek
Databricks MVP

OOM error is quite common, it means that probably your partitions are too large to fit into memory. Please analyze your SPARK UI - look for data spills, and skews and try to use smaller partitions or/and more shuffle partitions.


My blog: https://databrickster.medium.com/

View solution in original post

pavanb
New Contributor II

Thanks for the response @Hubert Dudek​ .

if i run the same code in test environment , its getting successfully completed and in dev its giving out of memory issue. Also the configuration of test nand dev environment is exactly same.

jose_gonzalez
Databricks Employee
Databricks Employee

Hi @Pavan Bangad​ ,

Just a friendly follow-up. Did @Hubert Dudek​ 's response help you to resolve your issue? if yes, please select it as best answer. If not, please let us know, so we can continue helping you to find a solution.