翻译官网救援/取消救援标准流程
1、用户在节点上调用Nova rescue
2、Nova ComputeManager调用virt驱动程序的rescue()方法,传入rescue_password作为参数
3、Virt驱动程序调用node.set_provision_state(RESCUE),并以rescue_password作为参数
4、Virt驱动程序在等待provision_state更改时循环,并根据需要更新Nova状态
5、Ironic API接收set_provision_state调用,并执行do_node_rescue RPC调用(ACTIVE - > RESCUING)
6、Ironic conductor在instance_info中设置了救援密码并将通知给相应的驱动
7、驱动程序使用配置的启动驱动程序启动救援虚拟磁盘(RESCUING - > RESCUEWAIT)。作为这个过程的一部分,Ironic将把节点放到rescue_network上,如ironic.conf配置
8、ramdisk启动,执行查找(/v1/lookup in ironic-api,获取节点信息,并开始心跳(/v1/heartbeat in ironic-api)
9、在收到心跳信号后,conductor 使用配置驱动器和救援密码(RESCUEWAIT - > RESCUING)调用finalize_rescue(/v1/commands),并从instance_info中删除救援密码,因为它不再需要
10、agent设置密码,从配置驱动器中的信息配置网络,并停止代理服务
11、conductor 将节点放回租户网络的网络端口,并将状态设置为RESCUE。
Standard rescue process
1、用户在节点上调用Nova unrescue
2、Nova称为Ironic unrescue()virt驱动程序。
3、Virt驱动程序调用node.set_provision_state(ACTIVE)。
4、Virt驱动程序在等待provision_state更改时循环,并根据需要更新Nova状态。
5、Ironic API接收set_provision_state调用,并执行do_node_unrescue RPC调用。
6、ironic conductor通知合适的驱动
7、驱动程序执行正常启动节点所需的操作,并将提供状态设置为ACTIVE。
Standard Unrescue process
ACTIVE -> RESCUING (initiate rescue)
RESCUING -> RESCUE (rescue succeeds)
RESCUING -> RESCUEWAIT (optionally, wait on external callback)
RESCUING -> RESCUEFAIL (rescue fails)
RESCUEWAIT -> RESCUING (callback succeeds)
RESCUEWAIT -> RESCUEFAIL (callback fails or abort issued)
RESCUEWAIT -> DELETING (delete instance without waiting)
RESCUE -> RESCUING (re-rescue node)
RESCUE -> DELETING (delete rescued instance)
RESCUE -> UNRESCUING (unrescue node)
UNRESCUING -> UNRESCUEFAIL (unrescue fails)
UNRESCUING -> ACTIVE (unrescue succeeds)
UNRESCUEFAIL -> RESCUING (re-rescue node after failed unrescue)
UNRESCUEFAIL -> UNRESCUING (re-unrescue node after failed unrescue)
UNRESCUEFAIL -> DELETING (delete instance that failed unrescuing)
RESCUEFAIL -> RESCUING (re-rescue after rescue failed)
RESCUEFAIL -> UNRESCUING (unrescue after failed rescue)
RESCUEFAIL -> DELETING (delete instance after failed rescue)
state machine impact