Changing the number of CPUs in a zone’s processor set (pooladm) Solaris 11.1

This post is mainly related to SuperCluster configuration, but can be applied to other solaris based systems.

In SuperCluster you can run the Oracle RDBMS in zones, and the zones are CPU capped. You may want to change the number of CPUs assigned to your zone for a couple of reasons

1) You have changed the number of CPUs that are available in the LDOM supporting this zone using a tool such as setcoremem and want to resize the zone to take this into account.

2) You need to change the zone size because you need more / less capacity.

Determine the number of processors that need to be reserved for the global zone (and all of your other zones!)

For SuperCluster you should reserve a miniumum of 2 cores per IB HCA in domains with a single HCA, a minimum of 4 cores for domains with 2 or more HCA. You also need to take into account other zones running on the system.

Find out how many CPUs are in the global zone

# psrinfo -pv

The physical processor has 16 cores and 128 virtual processors (0-127)

(... snipped output)
The core has 8 virtual processors (488-495)
The core has 8 virtual processors (496-503)
The core has 8 virtual processors (504-511)
SPARC-T5 (chipid 3, clock 3600 MHz)

So in my case, there are 512 virtual CPU, of which I need to reserve at least 32 (4 cores x 8 threads) for my global zone, and as I will also have an app zone on there, possibly a lot more. So lets say I’m going to keep 16 cores in global that would leave 384 for my dbzone

Get the name of the pool

The SuperCluster db zone creation tools create the pools with logical names based on the zonename. However, the way to be sure what pool is in use is to look at the zone’s definition

# zonecfg -z sc5acn01-d5 export |grep pool

If this doesn’t return the name of the pool, your zone is not using resource pools and may be using one of the other methods of capping CPU usage such as allocating dedicated cpu resources (ncpus=X). If so.. stop reading here as this is not the blog post you are looking for.

Find the processor set associated with your pool by either looking in /etc/pooladm.conf or by running pooladm with no parameters to get the current running config. Checking in pooladm is WAY more readable so that is my preferred method.

# pooladm |more

system default
string system.comment
int system.version 1
boolean system.bind-default true
string system.poold.objectives wt-load

int pool.sys_id 1
boolean true
boolean pool.default false
string pool.scheduler TS
int pool.importance 1
string pool.comment

Here’s an extract of my pooladm.conf

<pool name="" active="true" default="false" importance="1" comment="" res="pset_1" ref_id="pool_1">
<property name="pool.sys_id" type="int">1</property>
<property name="pool.scheduler" type="string">TS</property>
<pool name="pool_default" active="true" default="true" importance="1" comment="" res="pset_-1" ref_id="pool_0">
<property name="pool.sys_id" type="int">0</property>
<property name="pool.scheduler" type="string">TS</property>
<res_comp type="pset" sys_id="1" name="" default="false" min="8" max="8" units="population" comment="" ref_id="pset_1">

It looks to me that the res=”pset_1″ in the pool definition points to the ref_id=”pset_1″ in the pset definition.

OK, so now I know my pset is called and it currently has 8 cpus. I also know that my running config and my persistent config are synchronised.

Shutdown the database zone.

This may not be necessary, but since Oracle can make assumptions based on CPU count at startup, I think it is safest.

# zoneadm -z sc5acn01-d5 shutdown

Change the pset configuration

I’m going to do this by changing the config file to make it persistent, as there’s nothing more embarassing than making a change that is lost by a reboot. I set the processor set to have a minimum of 384 CPU and a maximum of 384 CPU.

# poolcfg -c 'modify pset ( uint pset.min=384 ; uint pset.max=384 )' /etc/pooladm.conf

Check that it has applied to your config file
# grep pset_sc5acn01-d5 /etc/pooladm.conf

Force it to re-read the file and use the new configuration

# pooladm -c
# pooladm -s

Now you can run pooladm without any arguments and get the running config. If it all looks ok, go ahead and boot your zone

# pooladm

(snipped output)

int pset.sys_id 1
boolean pset.default false
uint pset.min 384
uint pset.max 384
string pset.units population
uint pset.load 723
uint pset.size 384
string pset.comment


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s