Azure HDP cluster disappeared from Cloudbreak GUI and after reappearing won't sync

[#1]

I installed a HDP 2.5 cluster on Azure using Cloudbreak 1.6.3. Everything was running fine, until Microsoft stopped both the cluster and the cloudbreak VM because of a supposed 'service violation'. After that issue was resolved I restarted the cloudbreak vm and was told to upgrade to version 1.14.0 and regenerate the docker-compose.yml by the cdb doctor command. So I did. After starting cloudbreak and logging in, all seemed fine, only the reference to my cluster was missing. I also noticed that the GUI takes a long time loading (the page is greyed out and a 'Loading...' sign is displayed. It might be timing out on something...

• Starting the nodes manually resulted in Cloudbreak showing the cluster again. But subsequently syncing it from within cloudbreak results in an error 'Cluster can't be synchronized. [ stack status: STOPPED, cluster status: AVAILABLE ]'.
• Refreshing the Cloudbreak GUI makes the stop button appear, but clicking it results in 'cluster not available' error.
• Stopping all the nodes manually and syncing results in an error 'Cluster and it's stack sync failed please try again: Stack 11 has flows under operation, request not allowed'. The error stays the same, even after retry after some time.

How can I get the link to my (currently stopped) cluster back in the cloudbreak GUI and sync it?

[#2]

I could get a step further by

• manually updating the cluster status in the status table to 'STOPPED'
• manually starting all node vm's, logging in and halting them
• restarting cloudbreak

Then after some time cloudbreak syncs with the cluster and can startup the cluster. However it says that the infrastructure is started and that a cluster startup is ongoing. In Azure however the VM's still have a 'stopped' status.

When I look in the Cloudbreak startup log, I see an error:

Failed to convert from type [java.util.LinkedHashSet<?>] to type [com.sequenceiq.cloudbreak.api.model.AutoscaleStackResponse] for value 'com.sequenceiq.cloudbreak.domain.Stack@25194c0b'

Even after restarting Cloudbreak, these errors reoccur and the GUI still hangs in the 'update in progress: Starting Ambari cluster' fase...