Error / Incident
All the cloudcontrol nodes are down in the system.
Cloudcontrols are the nodes that host the control plane for Openstack, so this means that no new actions can be scheduled (starting VMs, creating disks, logging in, ...). This means that even though the VMs might still be running, the system is not usable.
Cloudcontrols host a lot of services, for example:
- Galera + mariadb for all openstack services
- Nova-api, nova-scheduler, nova-conductor (VM instance management)
- RabbitMQ for all openstack services
- HAProxy for all openstack services
- Trove-api, trove-conductor (DB as a Service)
- Keystone (authentication/authorization/accounting)
- Glance (VM image management service)
- Neutron API/rpc (software defined network service)
- Cinder (block device management service)
This might be the result of an ongoing intervention gone wrong, so first try to communicate with the rest of the team and look for any ongoing actions:
- Ask on irc (#wikimedia-cloud-admin) if someone is taking any action and forgot about it
- Check SAL (or here) for host being rebooted/taken down
- Check Netbox to make sure the host is not being decommed or network changes were not applied
- Check puppet for recent changes that might have affected the node
Then you can continue trying to figure out why the hosts are going down and not coming back up:
- Try to ssh to the host
- Try to ssh to the management port (see SRE/Dc-operations/Hardware_Troubleshooting_Runbook)
Add here any issues you encounter.
- SAL (or here).
- Puppet merged changes
Communication and support
Support and administration of the WMCS resources is provided by the Wikimedia Foundation Cloud Services team and Wikimedia Movement volunteers. Please reach out with questions and join the conversation:
- Chat in real time in the IRC channel #wikimedia-cloud connect, the bridged Telegram group, or the bridged Mattermost channel
- Discuss via email after you subscribed to the cloud@ mailing list
Provide a list of phabricator tasks of previous incidents to be able to more review previous instances of the issue.