问题
I have just installed ICp CE edition 2.1.0 on Ubuntu 16.04 (one cluster, one master, one worker node, very basic installation). When opening the 'catalog' page (https://.......... :8443/catalog/), I get the message 'Error loading Charts'.
In the 'admin>repositories' page I can see ibm-charts https//blablabla and local-charts https://blablabla/helm-repo/....
The 'admin>metering dasboard dispays an error 'E_DATA_QUERY_ERROR: The query for loginbootstrap failed with the response '500 Internal Server Error'
I have done very few modifications in the config.yaml (and hosts) files in the cluster directory (just configured the password authentication). Maybe some more custom configuration is required.
I'm discovering/learning about this product,maybe there is an obvious explication for such kind of behavior according to an expert.
Thanks
回答1:
Regarding the "error loading charts", check the following: Deployments > helm-api > {click the pod name at the bottom} > logs. Then in another tab open the Admin > Repositories page and click Sync Repositories and watch the log in other tab. Attempt to open the Catalog as well and watch the same log.
If you are seeing any cloudant related error, one possible way to resolve is to delete the helm-api pod and it will reinitialize with the view and the error should go away.
There was possibly an issue when connecting to cloudant when we setup the connection to it. So that helm-api pod needs a restart in order to add some files to cloudant now that it has been initialized.
My understanding is that a fix will be going in to help automate this recovery step in the next release.
As for the 'E_DATA_QUERY_ERROR: The query for loginbootstrap failed with the response '500 Internal Server Error' that was supposedly fixed in the GA release. Are you certain that you have installed the latest ICP from dockerhub for the CE release?
https://www.ibm.com/support/knowledgecenter/SSBS6K_2.1.0/installing/install_containers_CE.html
回答2:
The two problems, chart loading error and metering 'loginbootstap' error, are likely to have the same root cause: a problem communicating with the Cloudant database at the time of first startup when databases would be initialized. Restarting the helm-api pod would help the charts, and restarting the metering-server and then the metering-ui pods should resolve the Metering error.
回答3:
Today I have seen the same issue on ICp 2.1.0.1 EE when I try to navigate Catalog -> helm charts page. Page loading for a while then ended with "error loading charts". Weird thing is I didn't do anything, just leave it, after several hours re-visit again and it works.
Next time, I will first try sync repository Manage -> Helm Repositories -> Sync repositories, then check helm-api pod: (kubectl is on Windows)
kubectl -n kube-system get pods |findstr helm-api
then kill the pod if it is not running.
来源：https://stackoverflow.com/questions/46941797/ibm-cloud-private-ce-2-1-0-catalog-error-loading-charts