search

ceph / ceph-ansible

Ansible playbooks to deploy Ceph, the distributed filesystem.
Apache License 2.0
1.69k stars 1.01k forks source link

login to authenticated registry failing #4636

Closed pcuzner closed 5 years ago

pcuzner commented 5 years ago

When using downstream authenticated registry vars, the images pulls still fail. When I looked at the playbook, the login process is in ceph-container-common, but for the dashboard this role is not included?

ceph-ansible-4.0.2-1.el8cp.noarch group_vars/all.yml

---
alertmanager_container_image: registry.redhat.io/openshift4/ose-prometheus-alertmanager:4.1
ceph_conf_overrides:
  global:
    osd_crush_chooseleaf_type: 0
    osd_pool_default_size: 1
ceph_docker_image: rhceph-beta/rhceph-4-rhel8
ceph_docker_registry: registry.redhat.io
ceph_docker_registry_auth: true
ceph_docker_registry_password: <my password>
ceph_docker_registry_username: <my user>
ceph_origin: repository
ceph_repository: rhcs
ceph_repository_type: iso
ceph_rhcs_iso_path: /usr/share/ansible-runner-service/iso/RHCEPH-4.0-RHEL-8-20191007.6-x86_64-dvd.iso
ceph_rhcs_version: 4
cluster_network: 10.90.90.0/24
containerized_deployment: false
dashboard_enabled: true
grafana_container_image: registry.redhat.io/rhceph/rhceph-3-dashboard-rhel7:3
ip_version: ipv4
monitor_address_block: 10.90.90.0/24
node_exporter_container_image: registry.redhat.io/openshift4/ose-prometheus-node-exporter:v4.1
prometheus_container_image: registry.redhat.io/openshift4/ose-prometheus:4.1
public_network: 10.90.90.0/24
radosgw_address_block: 10.90.90.0/24
guits commented 5 years ago

@pcuzner could you share a full log?

pcuzner commented 5 years ago

rerunning now with latest stable branch (instead of downstream rpm)

pcuzner commented 5 years ago

@guits The task that fails is the grafana startup. I've attached the stdout from the run and also the grafana-server systemd log that shows the issue trying to pull from the auth'd registry

grafana-startup.log play_stdout.log

dsavineau commented 5 years ago

That's a valid issue. The ceph-container-common role isn't executed when deploying the dashboard with packages. I'm only surprise that it's not failing on node-exporter or prometheus because both roles are executed before grafana. @pcuzner Could check both node-exporter and prometheus logs ? I think we should see the same issue than grafana