Containers
User IDs
In order to run unprivileged containers we need to assign uid’s and gid’s to the users. To be able to use the users also across e.g. NFS shares the id’s need to be stable/unique. Services provided by OS packages (e.g. Apache) use their normal IDs for now. The following table is used for the uid’s and gid’s we define.
User-/Groupname |
ID |
|---|---|
operator |
2500000 |
keystone |
2500001 |
ceilometer |
2500002 |
cinder |
2500003 |
glance |
2500004 |
gnocchi |
2500005 |
horizon |
2500006 |
neutron |
2500007 |
nova |
2500008 |
libvirt |
2500009 |
mariadb (*) |
2500010 |
rabbitmq (*) |
2500011 |
barbican |
2500012 |
tempest |
2500014 |
ssl-terminator |
2500015 |
ovn |
2500016 |
placement |
2500017 |
designate |
2500018 |
(*): MariaDB and RabbitMQ might migrate to a random per-cluster assignment from a to-be-defined range in the future.
API-Container Naming
To ease the generation and parsing of logs we use a standardized naming scheme for all api-container hosting an openstack api (within the api-pod).
The syntax is <api-name>-api. Example for the Keystone-API is keystone-api.