Troubleshooting

How failures are handled

Enrichment runs are monitored by Greenpixie: failed runs alert our team directly and are retried or reprocessed. Nothing is written to your bucket on failure, so the symptom you will see is output not appearing, rather than an error file.

If output is missing, work through the checklist below, then contact your Greenpixie representative.

Checklist

Directory structure

The most common cause. Data must exactly match the directory structure, including the input base directory and the <platform> directory, even when they feel redundant (single platform, dedicated input bucket). Partners must also include the <org_id> and <timestamp> directories.

Column names

Each platform has required column names. AWS and GCP names are case sensitive. Every partition in a directory must share the same column names and data types.

Permissions

Confirm the bucket policy or role assignment from your platform setup is still in place: access changes on your side (policy edits, credential rotation, expired client secrets on Azure) are a common cause of silently stopped enrichments.

File size and format

Partitions must be under 500MB each, in a supported format. If your data involves a very large number of files per month, mention it to your Greenpixie contact so the run can be sized accordingly.

Partner triggers

If you are a partner and an upload has not been processed, confirm the .upload_finished file was written into the <timestamp> directory after all partitions finished uploading, and that your S3 event notification targets the SNS ARN we provided; see Partner trigger.

Still stuck?

Contact your Greenpixie representative with the bucket/container name, the directory path you expected output for, and roughly when the data was uploaded; we can check the run logs from our side.