Hi team, after upgrading to v0.9.5 I cannot login ...
# troubleshoot
a
Hi team, after upgrading to v0.9.5 I cannot login anymore with the root Datahub user nor through OIDC. The frontend and gms logs are shown below after trying to login with the root Datahub user. Any hints?
✅ 1
b
Can you expand the last line on the first image (frontend) log?
Thanks, I can see that this is after the user was authenticated, at this stage, a call is made from the frontend to GMS to generate a token for the user. The authentication used here is based on settings that need to be consistent between the frontend and backend. For the frontend, they are here and will be set (other than default) using
DATAHUB_SYSTEM_CLIENT_ID
and
DATAHUB_SYSTEM_CLIENT_SECRET
. Can you compare the environment variables in the frontend pod vs GMS? The environment variables are the same the GMS ones are similarly configured here. Also I want to make sure that both the frontend and GMS are on the same version, both v0.9.5.
a
Thanks for the hint! I will be off for a few days, but will check when I get back.
Hi @brainy-tent-14503, I ended up redeploying Datahub completely and that solved the login issue. Unfortunately, I don't have
kubectl
access so I was not able to check the values for the environment variables you suggested to look at. I do have some evidence to suggest that my issue was indeed related to these variables but I cannot prove it.
Anyway, for now, issue resolved.
b
Interesting, the bearer token is not related to oidc authentication, however it may indicate a fallback state after the oidc failure. Secondly, cookies generated by an older version are incompatible with the new version, I am curious as to whether you tried clearing cookies and going through the oidc authentication flow prior to re-deploying? Any additional context would be useful if you have it. Otherwise, I am glad it was resolved eventually. Thanks!
a
Yes, I did clear my cookies and try to login again through OIDC
But I could not login either
I'll check if I can still find the logs for that
b
When you re-deployed did you end up re-entering the oidc configuration in the settings?
a
No, that stayed the same
b
The first line indicates an issue identifying the signing algorithm. This information is provided in the oidc discovery uri configuration. Did you re-deploy v0.9.5 or the older version?
a
v0.9.5
b
The second log looks like mismatched values between the frontend and gms components which is after the oidc flow. Interesting, this is after the redeploy or during?
a
this is before the redeploy when the login issue occured
b
Ok, thank you. This will be helpful in further debugging what happen. Thank you! 🙏