Michael
03/19/2026, 1:57 PMabctl on a kind (Kubernetes in Docker) provider, and we're hitting a consistent issue that breaks the main user account during the upgrade process.
Environment:
- Provider: kind
- Upgrade method: abctl local install with version flag
Issue:
During the upgrade, the process fails when trying to update the email for authentication:
INFO Using Kubernetes provider:
Provider: kind
Kubeconfig: /root/.airbyte/abctl/abctl.kubeconfig
Context: kind-airbyte-abctl
INFO Updating email for authentication
ERROR Unable to update the email address
ERROR unable to udpate the email address: failed to update organization: unexpected status code: 401
This error breaks the main/admin user, making the instance unusable after the upgrade.
Additional symptoms observed post-upgrade:
- Syncs appear to start but produce no log output in real time
- No errors are surfaced in the UI or logs when a sync is triggered
- Connectors show entitlement warnings in the logs (though these may be unrelated):
Connector entitlement not available. organizationId=OrganizationId(value=00000000-0000-0000-0000-000000000000)
What we tried:
- Incremental upgrades (minor version by minor version up to v2) — earlier minor bumps work, but something breaks at the v2 boundary
- Upgrading connectors after the upgrade — no improvement
- Multiple clean reinstalls/migrations — same result
Has anyone else hit this 401 on the email update step? Is there a known workaround or a fix in the pipeline?
Important note:
The authentication corruption actually happens earlier than we initially thought — not at the app v2.0.0 upgrade step, but already during the helm chart upgrade from 1.7.0 → 2.0.5.
To be precise:
• Helm chart 2.0.5 still runs app version 1.7.0 (it only updates the installation method)
• But even at this first step, authentication data gets corrupted, before we even attempt the actual app upgrade to v2.0.0 (helm chart 2.0.18)
So the broken email/auth update (401) occurs as soon as the helm chart is bumped to 2.0.5, not when moving to app 2.0.0.
Other error that we got:
2026/03/19 14:21:14 [emerg] 1#1: host not found in upstream "airbyte-abctl-airbyte-connector-builder-server-svc.airbyte-abctl:80" in /etc/nginx/conf.d/default.conf:6
nginx: [emerg] host not found in upstream "airbyte-abctl-airbyte-connector-builder-server-svc.airbyte-abctl:80" in /etc/nginx/conf.d/default.conf:6
WARNING Back-off pulling image "airbyte/db:1.7.0"
WARNING Encountered an issue deploying Airbyte:
Pod: airbyte-db-0.189e44b17e6eeb4d
Reason: Failed
Message: Error: ImagePullBackOff
Count: 64
Thanks in advance!Ian Alton
03/27/2026, 7:16 PM--values)
• When you say version flag, do you mean --chart-version?
There are a couple of things that could be going wrong here but it depends on your answers.Michael
03/31/2026, 3:30 PMIan Alton
04/02/2026, 12:43 AM--chart-version or with no flag at all? abctl local install should default to the latest version. The --chart flag was really designed to pull a local chart from a file path. --helm-chart isn't a documented flag so I'd be shocked if that worked.
This is the only thing about what you're saying that looks off to me. The rest seems normal.
Another thing I'm curious about is if you can just do a fresh deployment without upgrading (in a separate space). That, at least, should work.