Workspace starting fails with "Plugins installation process timed out" error

[SkorikSergey]

Describe the bug

Workspace starting fails with "Plugins installation process timed out" error. There is a problem with che-plugin-metadata-broker-v3-2-0 container starting - "flag provided but not defined: -cadir"(see logs below).

Che version

Eclipse Che 7.18.0-SNAPSHOT

Runtime

• kubernetes (include output of kubectl version)
• Openshift (include output of oc version)
• minikube (include output of minikube version and kubectl version)
• minishift (include output of minishift version and oc version)
• docker-desktop + K8S (include output of docker version and kubectl version)
• other: (please specify)

Screenshots

Installation method

• chectl
  chectl/0.0.20200812-next.0314dc6 linux-x64 node-v10.22.0
• OperatorHub
• I don't know

Environment

• my computer
  • Windows
  • Linux
  • macOS
• Cloud
  • Amazon
  • Azure
  • GCE
  • other (please specify)
• other: please specify

Eclipse Che Logs

che-plugin-metadata-broker-v3-2-0 container logs

flag provided but not defined: -cadir
Usage of /metadata-broker:
  -cacert string
    	Path to Certificate that should be used while connection establishing
  -disable-push false
    	Whether pushing of data and logs to endpoint should be disabled. false by default. Needed for testing and debugging purposes
  -enable-auth false
    	Whether authenticate requests on workspace master before allowing them to proceed.By default the value from 'CHE_AUTH_ENABLED' environment variable is used or false if it is missing (default true)
  -metas string
    	Path to configuration file on filesystem (default "//config.json")
  -print-events-only false
    	Output events that are usually sent Che master instead of regular logs to imitate what a user can see.false by default. Needed for testing and debugging purposes
  -push-endpoint string
    	WebSocket endpoint where to push statuses
  -registry-address string
    	Default address of registry from which to retrieve meta.yaml's when plugin FQNs do not specify a registry
  -runtime-id string
    	The identifier of the runtime in format 'workspace:environment:ownerId'
  -use-localhost-in-plugin-urls localhost
    	This configures the broker to use the localhost name instead of the Kubernetes service name to build Theia or VSCode plugin endpoint URL.`true` by default since until now all remote VS Code or Theia plugin containers are started on the same POD as the Theia IDE container (default true)```

[mmorhun]

Should be fixed by #17671

[mmorhun]

Fixed.