Using Power Enterprise Pool on HMC V8R8.5

Answer

Create the Power Enterprise Pool


1- The HMC you use to create a pool will be set as the Master HMC by default
2- Download Pool XML Configuration file from the CoD website to your workstation:
http://www-912.ibm.com/PowerEntPool/PepPoolIdServlet
3- Secure copy the file to the HMC home directory:
# scp <file> hscroot@<HMCIP>:/home/hscroot
4- Use command mkcodpool to upload the XML file and create a new pool:
# mkcodpool -p <pool-name> -f <XML file name> -v


Add Managing HMCs


1- HMC 850 is required on all HMCs
2- All servers in a pool are no longer required to be managed by the same HMC pair, but one master HMC for the Pool is still required
3- All managing HMCs must have a network connection to the master HMC, RMC is still used to communicate between the HMCs and not that the Managing HMCs do not communicate with each other
4- Any managing HMC can be set as a master HMC at any time
5- To add a Managing HMC, run the below command from the Master HMC:
# chcodpool -o add -p <pool-name> --mc <IP of the Managing HMC> -u <user-ID> --passwd <password>


Setting a New Master HMC for the Pool


1- The HMC used to create the pool is set as the master HMC for the pool
2- Set a new master HMC while the current master HMC is running if possible
3- Operation can be initiated on any managing HMC
4- If the current master HMC is down or not reachable, operation can only be initiated on the HMC to be the new master
5- Only a managing HMC for the pool can be set as the master HMC
6- The master HMC or managing HMCs do not have to manage any servers in the pool
7- To set a new Master HMC (in case the current Master HMC is down or unreachable), you must run this command on the HMC that you want to set as the new master:
# chcodpool -p <pool-name> -o setmaster --mc this --force
8- To set a new Master HMC (In case the Master HMC is up and running), you can run this command on the current Master HMC or any of the Managing HMCs:
# chcodpool -p <pool-name> -o setmaster --mc <HMC IP>


Add Mobile CoD resources to a system in a Pool:


1- This operation can be done on any of the Managing HMCs or the Master HMC
2- In order to be assigned Mobile CoD resources, a server must have unlicensed resources.
3- Server must be in Standby or Operating state
4- You can use the chcodpool command to do that:
#chcodpool -o add -r {mem | proc} -p <pool-name> -m <managed-system> -q <quantity>
5- Mobile CoD resources remain on the server until the user removes them
6- Mobile CoD resources that are added are used to satisfy unreturned Mobile CoD resources first
7- Mobile CoD resources cannot be added to servers in Initializing, Powering Off, Power Off, or Error state
8- Mobile CoD resources can be added to servers with unreturned pool resources in No Connection, Pending Authentication, Failed Authentication, Incomplete (if the power state is Standby or Operating), or Recovery (if the power state is Standby or Operating) state, or an unmanaged server
9- The number of resources to be added cannot exceed the number of unreturned resources


Removing Mobile CoD Resources from a Server in a Pool


1- Server should be in Standby or Operating state
2- Incomplete and Recovery states are also OK if the power state is Standby or Operating
3- Initializing, Powering Off, Power Off, or Error state are also OK, since the server will need to be IPL'ed
4- You can use the chcodpool command to do that:
#chcodpool -o remove -p <pool-name> -m <managed-system> -r {mem | proc} -q <quantity> --force
5- If the resources removed are not free on the server, the resources are allowed to be removed anyway and are considered unreturned
6- They are added back to the available Mobile CoD resources and can be added to another server
7- Required for active partition migration
8- A compliance grace period timer is started
9- The total number of unreturned resources in the pool cannot exceed the total number of resources in the pool.


Upgrading the HMCs Managing an Existing Pool to V850


1- If the existing pool has a backup master HMC configured, it will automatically be added as a managing HMC when the master HMC is upgraded
2- Both HMCs managing the existing pool must be upgraded to 850
3- Once one HMC is upgraded to 850, all servers the other HMC is also managing will transition to Version Mismatch state until the other HMC is upgraded
4- It does not matter if the master HMC or the backup master HMC is upgraded first
5- No user actions are required for the pool after upgrading to 850 (assuming both HMCs are still managing the pool servers when they are upgraded)




FAQs:


1- I just took an outage and shut down an LPAR but that didn't release cores for deactivation, what should I do?


Unfortunately, shutting down an LPAR doesn't change the fact that its cores are still accounted for in the "Assigned to Partitions" value in the managed server's properties. So you won't be able to deactivate those static cores you're required to give back in trade for your new mobile cores... unless you use the chhwres command to "free" those cores:

hscroot@hmc1:~> chhwres -r proc -m 750-1 -o r --id 8 --procunits 4.0

The command above (r)emoves 4.0 processing units, adding them to the "Available" count and thus enabling a deactivation of that many more cores.

More info here: https://www.ibm.com/developerworks/community/blogs/cgaix/entry/using_chhwres_to_free_up_partition_resources?lang=en



2- I’m getting timeout errors when try to add a new Managing HMC or move resources, what should I do?

Update you HMC to V8 R8.5.0 Service Pack 1, as it fixed a problem preventing HMC to HMC communication for managing a Power enterprise pool. This problem can cause the addition of a managing HMC to the pool to fail and cause time out errors to occur when trying to assign resources to servers in the pool. It can also cause the connection states for managing HMCs to flip flop between "Connected" and "Unknown". The issue will also cause IBM.LparCmdRMd to fault and log an error in /var/log/messages.