Kevin O'Keefe
03/27/2026, 6:29 PMkapa.ai
03/27/2026, 6:29 PMDocStoreAccessException: Unable to write output for 22f6c74f-..._check
Caused by: java.io.IOException: HTTP error: 500 Internal Server Error
This occurs when the connector-sidecar tries to write its check output to the Airbyte workload API server via WorkloadOutputApi.writeWorkloadOutput, but receives an HTTP 500 error after exhausting all 5 retry attempts. [HTTP 500 issue]
This is a known issue in Airbyte OSS, particularly on Kubernetes/Helm deployments. It is related to the connector-sidecar not being able to successfully communicate with the workload API server. [Connector-Sidecar Auth Issue]
To help narrow down the root cause, could you provide:
1. Airbyte version you are running
2. Deployment method (Docker Compose, Kubernetes/Helm, etc.)
3. Did this work before? If so, did you recently upgrade or change any configuration?
If you are on Kubernetes/Helm, this may be related to a known issue where WORKLOAD_API_AUTH_HEADER and WORKLOAD_API_BEARER_TOKEN environment variables are not properly propagated to connector-sidecar containers. There is currently no straightforward Helm value to inject these into dynamically created sidecar containers. [Connector-Sidecar Auth Issue]
Note that switching from GCS staging to Batched Standard Inserts would not resolve this error, since the problem is at the platform communication layer, not the connector configuration level. š 0 button š 0 button ā¹ļø Tip: Mention @kapa.ai in the thread for follow-ups or DM the app for 1-to-1 help.Kevin O'Keefe
03/27/2026, 6:31 PMKevin O'Keefe
03/27/2026, 6:32 PMkapa.ai
03/27/2026, 6:32 PMcredentialsJsonPath when building job pod volume mounts, hardcoding the secret mount path to /secrets/gcp-creds/ instead. If your GCS credentials path doesn't match that hardcoded path, the connector-sidecar crashes immediately. [GCS credentialsJsonPath bug]
Workaround: Remove the credentialsJsonPath from your values.yaml and rely solely on credentialsJsonSecretKey, which forces the system to use the hardcoded /secrets/gcp-creds/ path:
storage:
type: gcs
secretName: airbyte-mounted-secrets
gcs:
projectId: my-gcp-project
credentialsJsonSecretKey: airbyte-gcp-json
# Remove credentialsJsonPath entirely 2. Connector-Sidecar Auth Token Propagation Issue Additionally, in v2.0.1, there is a known issue where WORKLOAD_API_AUTH_HEADER and WORKLOAD_API_BEARER_TOKEN are not propagated to connector-sidecar containers, causing HTTP 500 errors when writing output. [Connector-Sidecar Auth Issue]
Given that your setup worked before the bucket change, the GCS credentialsJsonPath bug (#1) is the most likely cause of your current failure. I'd recommend trying the workaround above first. š 0 button š 0 button ā¹ļø Tip: Mention @kapa.ai in the thread for follow-ups or DM the app for 1-to-1 help.Kevin O'Keefe
03/27/2026, 6:34 PMkapa.ai
03/27/2026, 6:34 PMcredentialsJsonPath bug (issue #73739) has been escalated to the internal platform team for investigation and a fix, but there is no mention of a specific chart version that resolves it. [GCS credentialsJsonPath bug]
The knowledge sources do not explicitly confirm that any currently released chart version fully fixes these issues. The latest chart version listed is 2.0.18 (App Version 2.0.1), which is the same version where these bugs are reported.
What I can confirm from the knowledge sources:
⢠The GCS credentialsJsonPath bug is tracked internally but no fix release is mentioned. [GCS credentialsJsonPath bug]
⢠The connector-sidecar auth issue is also tracked internally. [Connector-Sidecar Auth Issue]
In the meantime, the recommended workaround is to remove credentialsJsonPath from your values.yaml and use only credentialsJsonSecretKey with lowercase gcs as the storage type:
global:
storage:
type: "gcs"
secretName: your-secret-name
gcs:
projectId: your-gcp-project
credentialsJsonSecretKey: "gcp.json"
I'd recommend checking the Airbyte ArtifactHub and release notes directly for the latest chart versions and whether a fix has been released since these issues were filed. [Upgrading Airbyte] š 0 button š 0 button ā¹ļø Tip: Mention @kapa.ai in the thread for follow-ups or DM the app for 1-to-1 help.Kevin O'Keefe
03/27/2026, 6:37 PMkapa.ai
03/27/2026, 6:37 PMcredentialsJsonPath Helm bug is not your issue.
Your error ā the connector-sidecar failing with DocStoreAccessException: Unable to write output and HTTP 500 ā is then more likely related to the connector-sidecar authentication issue in v2.0.1, where the sidecar containers don't receive the necessary tokens to communicate with the workload API server. [Connector-Sidecar Auth Issue]
A few things to check:
1. Verify API_AUTHORIZATION_ENABLED: Ensure API_AUTHORIZATION_ENABLED=true is set in your airbyte-airbyte-env ConfigMap. Without this, the /api/v1/workload/{id}/output endpoint may not be reachable. [Connector-Sidecar Auth Issue]
2. Check if the issue is specific to the connection test: Since this started when you ran a test after changing the GCS bucket, it's worth verifying whether existing syncs (not tests) are also failing, or if it's isolated to the check operation.
3. HMAC key permissions: Ensure the service account linked to your HMAC key has the required GCS permissions: [GCS Destination Docs]storage.multipartUploads.abort
storage.multipartUploads.create
storage.objects.create
storage.objects.delete
storage.objects.get
storage.objects.listOne community report found that granting Storage Object Admin to the service account resolved a similar "failed status but file present" issue. [GCS Failed Status]
The knowledge sources don't provide a definitive fix for the HTTP 500 workload API issue in v2.0.1 beyond what's already been mentioned. I'd recommend opening or tracking issue #70451 for updates. š 0 button š 0 button ā¹ļø Tip: Mention @kapa.ai in the thread for follow-ups or DM the app for 1-to-1 help.Kevin O'Keefe
03/27/2026, 6:38 PMkapa.ai
03/27/2026, 6:38 PMWORKLOAD_API_BEARER_TOKEN. For example: [source-postgres 502 issue]
kubectl -n airbyte-abctl rollout restart deployment airbyte-abctl-worker
kubectl -n airbyte-abctl rollout restart deployment airbyte-abctl-workload-api-server
However, the knowledge sources do not explicitly state that all components need to be restarted ā only the worker and workload API server are mentioned in the workarounds. Some users also restarted the server deployment:
kubectl -n airbyte-abctl rollout restart deploy/airbyte-abctl-workload-api-server airbyte-abctl-worker airbyte-abctl-server
[source-postgres 502 issue]
That said, since your setup was working before the bucket change and you haven't made token/auth changes, a restart alone may not resolve the underlying HTTP 500 issue. The knowledge sources don't confirm that a restart without any configuration change would fix the problem in your specific case. š 0 button š 0 button ā¹ļø Tip: Mention @kapa.ai in the thread for follow-ups or DM the app for 1-to-1 help.