The current version of the Zone Manager supports CPU processor
sets via the -p flag. Here is the usage:
-p resource|resource_arg
resource can be either cpu or ram.
resource_arg is either number of processors
or Mb of RAM depending on the resource specified.
Processor count enables you to specify the number
of processors that will be assigned to this zone.
(Not yet available) RAM count enables you to
specifiy the maximum amount of RAM in bytes that
this zone can use.
You can put a zone in a processor set both at creation via
the add action or via modify for an existing zone. For
example:
zonemgr -a add -n myzone -P pw -z /zones -p cpu|2
or
zonemgr -a modify -n myzone -p cpu|2
I am holding off on adding support for memory resource capping
until its method has solidified which may now be the case with
build 65.
Regards,
Brad
On Thu, 2007-02-22 at 17:50 -0500, Michael Barrett wrote:
You can attach your zones to static or dynamic pools that will control
the CPU utilization of them. You can weight zones via the share concept
within the same pool to determine which zone is more important within a
given pool. You can also control the inband and outband network traffic
of the zone. You can then further set projects up inside your zones to
control the memory and CPU utilization of applications or groups of
users that may be running within them. The layers look like this:
Raw host utilization
static/dynamic pool CPU utilization
zones: shares | IPQoS
projects: shares | mem limits
Solaris will soon add memory set constraints for zones.
It you want a tool to not only discover all of that graphically no
matter how you have created it across your enterprise, allow you to set
thresholds to generate SNMP traps and email notifications on resource
breeches or status conditions, and provision these conceptscheck out
Solaris Container Manager.
Thanks,
Mike
[EMAIL PROTECTED] wrote:
Hello,
Following a recent successfull deployment of 8 default sparse zones using
zonemgr.
I would like to add some simple resource caps to prevent one app stealing
all the resources (as actually happened - on a T2k). What would be the next
most logical step?
TIA
Tony
This message posted from opensolaris.org
___
zones-discuss mailing list
zones-discuss@opensolaris.org
___
zones-discuss mailing list
zones-discuss@opensolaris.org
___
zones-discuss mailing list
zones-discuss@opensolaris.org