Hello @Arun Kumar
Yesterday We upgraded but LCM did not put the host into maintenance mode!?
before upgrading put the host into maintenance mode, after that all VM’s should migrate to another host except the Controller VM you have to shut down it manually using this command:
cvm_shutdown -P now
after the upgrade complete exit maintenance mode, and turn on controller VM.
I know this is not complete answer to your question, hopefully someone will answer all your points.
@Arun Kumar Here are the answers to your questions about the phoenix/foundation and LCM based upgrades:
1. Once the phoenix iso is created and mounted to the host along with the upgrade bundle, Can phoenix run independently and complete the upgrades or needs help from foundation for performing some tasks like host reboot?
--- Phoenix process cannot run independently during firmware upgrades. It still needs help from the foundation service which runs on the LCM leader CVM that will control all the other tasks like rebooting node to phoenix and then after upgrade, rebooting the node back to phoenix, etc.
2. During LCM upgrade, since the node undergoing upgrades, will have all VMs powered off including the CVM, will it take help from a remote CVM foundation to orchestrate the upgrades on this host?
--- During LCM upgrade, the host goes into maintenance mode. So, all the user VMs (except the pinned VMs) migrate off to another hosts in the cluster and the CVM which resides on this host is powered of. LCM leader CVM (another CVM in cluster which has all services in UP and running state) is responsible to control this whole upgrade process which includes:
a) performing the pre-upgrade steps like putting host and CVM in Maintenance mode, shutting down CVM and booting node in phoenix state
b) Initiate the upgrade process on the phoenix node (LCM leader CVM sends out a foundation based API call to the phoenix node to perform this action)
c) performing the post-upgrade steps like removing node from phoenix state and powering it on, removing host and CVM in Maintenance mode, stopping foundation service on itself etc.
3. Putting host in maintenance mode is taken care by LCM or foundation service?
--- The task is requested by LCM wherein LCM asks either acropolis service (for AHV clusters) or vCenter (for ESXi clusters) to put the host in Maintenance mode and then waits for a response from them to confirm whether host entered maintenance mode successfully or not. Foundation service is typically used during the actual upgrade process when the node has booted into phoenix mode.
4. Can you please give a high level working of the phoenix mechanism, since the hypervisor installer itself is a separate iso, which needs to be mounted, just wondering how phoenix handles multiple iso, does it mount simultaneously with the upgrade iso?
--- Re-imaging the node manually using phoenix ISO is different that upgrading the cluster components (AHV/firmware) using LCM. When you are re-imaging the node manually using a phoenix ISO, we have ways to create a single ISO consisting of (hypervisor + phoenix) OR (hypervisor + AOS + phoenix) OR (AOS + phoenix) iso. In case if that is not done, then you can always first use the ISO to deploy just the hypervisor and once the installation completes, unplug this ISO and then plug in the (AOS + phoenix) ISO to deploy the CVM on it and configure the hypervisor or if CVM is already present, then just use phoenix ISO to configure the hypervisor. Refer to KB-3523 for more details on how to create one.
--- When an upgrade happens with LCM, it uses a single ISO file to do that update at one time. So, it doesn't need to mount multiple ISOs simultaneously at one time
Kindly let us know if it helped to clear your doubts about this one.
@Nupur.Sakhalkar Thank you very much for the detailed explanation, it was clear and to the point, I had spent hours trying to figure out this, but I was not able to find any Article that gives in depth working of phoenix or foundation. This really helped me, Thanks
.
@Arun Kumar Glad to help you out on this one!
