RDB loader not loading into databricks

For engineers
1

I’m encountering an issue with our Databricks cluster setup that has affected our RDB Loader 5.6.0 deployment. The cluster, which usually spins up automatically, has stopped doing so over the past week. Notably, I can confirm that SQS messages are still available, and the associated EC2 instance for RDB Loader is running without any issues.

After reviewing the setup, I found that no logs are appearing in CloudWatch, which suggests that the cluster may not be initializing as expected. Here are some relevant details:

Databricks Table Properties:
delta.columnMapping.mode = ‘name’
delta.minReaderVersion = ‘2’
delta.minWriterVersion = ‘5’

Databricks Runtime Version: Recently upgraded from 14.x to 15.4 LTS

Given this information, I’m wondering if the recent upgrade to Databricks runtime 15.4 LTS may be affecting the RDB Loader’s automated cluster spin-up functionality. Could there be any compatibility issues between RDB Loader 5.6.0 and Databricks 15.4 LTS, or specific configuration requirements to address this change?

2

Hey @Harshit if you are not seeing any logs from RDB Loader it could be that the loader itself might have crashed or gotten stuck. Have you tried rebooting the RDB Loader instance to see if that brings it back to life?

3

I manually started the databricks cluster. Upon doing so, it successfully began processing the data queued in SQS. However, I’m noticing that even after rebooting the RDB Loader EC2 instance, no logs are being generated in CloudWatch.

4

Nevermind, now after manual intervention the logs starts coming up. I will monitor and let you know if the databricks cluster spinning up automatically or not

1 Like