Solved

LCM firmware udpate failed performing action reboot_from_phoenix

  • 25 May 2020
  • 1 reply
  • 1247 views

Hi, everyone

I’m trying to use LCM to update firmware of host machines on my cluster. It all went well until the post action phase. Error message said: ‘Operation failed. Reason: LCM failed performing action reboot_from_phoenix in phase PostActions on ip address xx’.

I searched the KB and found KB9177,but it’s about ‘Mixed Hypervisor cluster‘, my cluster uses solely AHV so it doesn’t applied.

Anyway I still tried to follow the KB9177’s suggestion and upgraded my cluster’s foundation to 4.5.3 and retried the LCM firmware update process on another host but still got the same ‘ LCM failed performing action reboot_from_phoenix’ error.

I used the workaround provided in that KB to make the affected two hosts’ CVM out of maintenance mode. It works and the cluster is back to normal. Then I logged on to the affected hosts’ IMM and found out that actually the primary IMM2 firmware has already been updated by the LCM( the backup IMM2 firmware is not upgraded), and when I go to the LCM section in PE, the hosts affected by the error is no longer listed in the firmware update page.

So I am confused now: is the ‘reboot_from_phoenix’ action a necessity or not ? Since the host’s primary IMM2 firmware has already been updated to the target version, does that mean the LCM firmware update process is in fact successful?

 

AOS 5.15

LCM 2.3.1

foundation 4.5.3

host machine: LENOVO HX5510

old firmware version: 3.40

target firmware version: 5.30

 

icon

Best answer by Nashma 8 June 2020, 20:38

Hi @Welsper 

Most firmware updates through LCM are run from a CentOS-based staging area called Phoenix. reboot_to_phoenix is an API used to enter this stage and invoke the upgrade activities in this stage. Similarly, reboot_to_host, is later used to exit this staging area and boot the upgraded node back into its hypervisor after firmware upgrades. 

In your situation, the API had issues in post upgrade phase and hence you see the firmwares reflect the latest versions despite the error thrown on the LCM activity. In such situations, we reboot the host into hypervisor and bring the host and CVM out of maintenance mode.

View original

This topic has been closed for comments

1 reply

Userlevel 1
Badge +1

Hi @Welsper 

Most firmware updates through LCM are run from a CentOS-based staging area called Phoenix. reboot_to_phoenix is an API used to enter this stage and invoke the upgrade activities in this stage. Similarly, reboot_to_host, is later used to exit this staging area and boot the upgraded node back into its hypervisor after firmware upgrades. 

In your situation, the API had issues in post upgrade phase and hence you see the firmwares reflect the latest versions despite the error thrown on the LCM activity. In such situations, we reboot the host into hypervisor and bring the host and CVM out of maintenance mode.