Process is deployed via desktop modeler successfully but same is NOT visible in the Operate

Camunda 8 Topics
,
1

Hi All,
I am using camunda self managed deployment, In this using operate 8.7.2 and zeebe also 8.7.2, with external Open-search

when deployed via desktop modeler, i’m not able to see the any data in operate. I verified the operate logs there is NO errors as such.

Tried solution: Changed camunda version from 8.7.2 to 8.7.4 and also 8.7.7 as well.

Can you please guide on the following task.
1. check data whether data is persisted in operate.
2. Why data is NOT pulled by exporters.

2

Hi there! :wave:

I can help you troubleshoot this issue. This is a common problem with Camunda self-managed deployments where processes deployed via Desktop Modeler aren’t visible in Operate. Let me address your specific questions:

Classification: Problem :wrench:

1. How to Check if Data is Persisted in Operate

To verify if data is being properly exported and imported:

Check OpenSearch Indices:

  • Look for zeebe-record-* indices in your OpenSearch cluster
  • Verify these indices contain data after deploying a process
  • Check if the indices are being created and populated by Zeebe’s exporter

Check Operate Logs:

  • Look for importer errors or warnings about missing data
  • Check for any index-related issues in the Operate startup logs
  • Verify the Operate importer is running successfully

2. Why Data is NOT Pulled by Exporters

Based on your setup (Zeebe 8.7.2/8.7.4/8.7.7 with OpenSearch), here are the most likely causes:

Common Root Causes:

  1. Index Template Issues :dart:

    • Incorrect mapping for fields like value.variables (should be mapped as keyword)
    • This prevents Operate from querying process instances even if data exists
    • Solution: Update your OpenSearch index templates with correct mappings

  2. Exporter Configuration Problems

    • Index naming mismatches between Zeebe exporter and Operate expectations
    • Incorrect OpenSearch endpoint configuration
    • Known Issue: There was a bug with index prefixes in Helm charts causing data to be written to wrong indices

  3. Version Compatibility

    • From Camunda 8.6.0+, Zeebe and Operate must run on exact same minor and patch versions
    • OpenSearch 2.9+ is required
    • Ensure you’re using the OpenSearch exporter (not Elasticsearch exporter)

Troubleshooting Steps:

  1. Verify Zeebe Export:

    # Check if zeebe-record indices exist and have data
curl -X GET "your-opensearch:9200/_cat/indices/zeebe-record-*?v"

  2. Check Index Templates:

    # Verify index templates are correct
curl -X GET "your-opensearch:9200/_index_template/zeebe-record-*"

  3. Restart Operate:

    • After fixing any index/template issues, restart Operate to ensure it picks up changes

  4. Test with Simple Process:

    • Deploy a basic process without complex BPMN features to isolate the issue

Immediate Actions:

  1. Check your Zeebe configuration for the OpenSearch exporter settings
  2. Verify all components (Zeebe, Operate, OpenSearch) are on compatible versions
  3. Look for any index prefix misconfigurations in your deployment
  4. Review Operate importer logs for specific error messages

References:

Could you share:

  • Your Zeebe and Operate configuration files
  • Any specific error messages from Operate logs
  • Output from checking your OpenSearch indices

This will help me provide more targeted assistance! :rocket:

3

Resolved - There was data inconsistency in opensearch DB. Cleared the all records under zeebe-record*, operate*, tasklist* and also data in PVC in k8 cluster.
post this restart the all the pods in k8.