Deployed the hive helm chart as is but liveness and readiness fails

[ChrisTomAlx]

Hi,
Thanks for this great helm chart, but when I installed the helm chart without changing any of the default values hive-server would keep crashing and restarting. I used the command helm install my-hive gradiant/hive --version 0.1.4

Things I have already tried

• I pinpointed the issue to liveness and readiness probes of the hive-server statefulset failing.
• I increased resources given to hive and hive-metastore as well
• I made sure the firewall wasn't blocking any ports but still the probes fail

Environment

• GKE
• Master version - 1.16.13-gke.401

What else can I try? Any help would be appreciated

[cgiraldo]

Hi,

This is a bug. We have change default user of hive-server from root to "hive", and permissions in hdfs /tmp folder were not updated accordingly.

[cgiraldo]

Try now with:

helm install my-hive gradiant/hive --version 0.1.5

Thanks for file the bug!

[ChrisTomAlx]

Hi @cgiraldo Thanks a lot for the quick response.

I tried installing 0.1.5 with helm install my-hive gradiant/hive --version 0.1.5 and now I am getting this error Error: secrets "my-hive-postgresql" already exists
Completely cleaned the cluster and retried as well, still ran into the same issue. Just to confirm it wasn't an issue with my cluster I installed 0.1.4 and it installed but with the same old liveness issue.

[cgiraldo]

I made a mistake with hive requirements since postgresql is already provided by hive-metastore.
I have revert hive-0.1.5 release and make a new one.
Now It should work.

Maybe you need to do a helm repo update.

Also all pods in running state can take several minutes.

[ChrisTomAlx]

It works now. I will play around with the helm chart and raise issues if I come across any.
Thanks a lot :)))