Based on the knowledge sources provided, intermittent SSL errors when running Elementary Data (

edr

) from a Kubernetes cluster—especially when connecting to Databricks—are not uncommon. The error you’re seeing,

SSLZeroReturnError(6, 'TLS/SSL connection has been closed (EOF) (_ssl.c:1147)')

, suggests that the TLS/SSL connection is being unexpectedly closed during communication. Here are some relevant findings and suggestions from the knowledge sources: 1. Underlying Cause: Similar SSL errors have been reported when using Databricks with dbt and Elementary. In one case, the error was traced back to the Databricks connection itself, not specifically to Elementary. The advice was to check the dbt profile and network/proxy settings, as well as to confirm that the same error occurs with plain dbt commands (like

dbt deps

), which would indicate a broader connectivity issue rather than an Elementary-specific bug. If the error happens with dbt alone, it’s likely a Databricks or network configuration issue, not Elementary itself [[GitHub Issue #1308](https://github.com/elementary-data/elementary/issues/1308)]. 2. Environment Differences: There are reports that running the same commands locally works fine, but issues arise in containerized or orchestrated environments like Kubernetes. This points to possible differences in network policies, SSL certificate handling, or resource constraints in the Kubernetes environment [[Slack thread](https://slack.com/archives/C02CTC89LAX/p1701136879967749)]. 3. SSL Certificate Handling: Several users have resolved SSL issues by updating the

certifi

package, ensuring the correct CA certificates are available, or by running scripts to install certificates in their environment. For macOS, there are specific steps involving

install_certifi.py

, but for Linux/Kubernetes, ensuring the container image has up-to-date CA certificates is key [[Slack thread](https://slack.com/archives/C02CTC89LAX/p1678807933973469)]. 4. Retry and Connection Handling: The error message you posted shows that urllib3 is retrying connections, which means the underlying library is attempting to recover from dropped connections. This could be due to network instability, aggressive timeouts, or resource limits in the Kubernetes cluster. Recommended Actions: - Check Network Policies: Ensure there are no network policies, firewalls, or proxies in your Kubernetes cluster that could be interrupting long-lived SSL connections to Databricks. - Update CA Certificates: Make sure your container image has the latest CA certificates installed (e.g., update

ca-certificates

package). - Test with dbt Alone: Try running the equivalent dbt commands in the same Kubernetes environment to see if the issue is specific to Elementary or a broader connectivity problem. - Resource Limits: Check if your pods have sufficient memory and CPU, as resource exhaustion can cause unexpected connection drops. - Elementary Version: Make sure you are using the latest version of Elementary, as newer releases may include fixes for SSL and connection handling [[Slack thread](https://slack.com/archives/C02CTC89LAX/p1701136879967749)]. If the problem persists after these steps, it may be necessary to involve your Databricks administrator or cloud infrastructure team to investigate possible network or SSL termination issues between your Kubernetes cluster and Databricks. If you have more specific logs or error messages, sharing them could help narrow down the root cause further.