Zones-9 [rctls / resource controls]

ZONES-9 [RESOURCE CONTROLS]

Great,

We had learned to assign mem & cpu to NGZ, but how we will check the utilization by NGZ

Rcapadm is command from which we can find and it relies on rcapd daemon

First we should enable the rcapadm

root@sol-test-1:>/# svcadm enable rcap

Enable the resource capping daemon so that it will be started now and also be started each time the system is booted

root@sol-test-1:>/# rcapadm -E

root@sol-test-1:>/# rcapstat -z 2 3  [3 samples on interval of 2 secs]

    id zone            nproc    vm   rss   cap    at avgat    pg avgpg

    17 newzone             -  212M  184M  500M    0K    0K    0K    0K

    17 newzone             -  212M  184M  500M    0K    0K    0K    0K

    17 newzone             -  212M  184M  500M    0K    0K    0K    0K\

[Plz refer man pages for the details about rcapadm]

We also have prstat –Z to monitor the zone

root@sol-test-1:>/# prstat -Z

[…………………]

ZONEID    NPROC  SWAP   RSS MEMORY      TIME  CPU ZONE

     0       53  250M  166M    16%   0:05:26 0.4% global

     1       32  217M  127M    12%   0:02:10 0.1% tzone4

    13       32  209M  196M    19%   0:01:32 0.0% clonedZone

    17       34  212M  185M    18%   0:00:32 0.0% newzone

OK… back to zone resources,

Previously we learned about zone resource capping, which needs reboot to take the changes in effect,

But what if we can’t reboot our zone but we must have to provide resource…

YES… HOT ADD/REDUCE OF MEM/CPU IS POSSIBLE…

root@sol-test-1:>/# zonecfg -z newzone info

capped-memory:

        physical: 500M

        [swap: 256M]

        [locked: 150M]

root@sol-test-1:>/# rcapadm -z newzone -m 300m

Here I had changed the ram size to 300mb

root@sol-test-1:>/# zonecfg -z newzone info |grep physical

        physical: e1000g0

        physical: e1000g0

        physical: 500M

But in zonecfg, it is still 500mb

Let’s check in the zone

bash-3.2# hostname

newzone

bash-3.2# prtconf -v |grep -i mem

prtconf: devinfo facility not available

Memory size: 300 Megabytes

Well… here it is changed

What does it mean?

It means that it is not persistent, just a reboot and it will back to value listed in zonecfg

Now change the SWAP on the fly,

root@sol-test-1:>/# prctl -n zone.max-swap -v 100m -r -e deny -i zone newzone

root@sol-test-1:>/# prctl -n zone.max-swap -i zone newzone

zone: 18: newzone

NAME    PRIVILEGE       VALUE    FLAG   ACTION                       RECIPIENT

zone.max-swap

        privileged       100MB      -   deny                                 -

        system          16.0EB    max   deny                                 -

-n         specifies the rctl name

-r         replaces the current value with the new value

-v         specifies a new value

-e         enables the specified action

-i         id type, Valid idtypes are process, task, project, or zone

        

root@sol-test-1:>/# prctl $$ |grep -i zone

zone.max-swap

zone.max-locked-memory

zone.max-shm-memory

zone.max-shm-ids

zone.max-sem-ids

zone.max-msg-ids

zone.max-lwps

zone.cpu-cap

zone.cpu-shares

RESOURCE TYPES

admin           Define the user name and the authorizations for that user for a given zone dedicated-cpu ncpus,importance

capped-cpu      ncpus

capped-memory   physical,swap,locked

fs              dir,special,raw,type,options

dataset         name

net             address,allowed-addressphysical,defrouter

device          match

rctl            name,value

The following zone-wide resource controls / rctls are available.

zone.cpu-cap

zone.cpu-shares      (preferred: cpu-shares)

zone.max-locked-memory

zone.max-lofi

zone.max-lwps        (preferred: max-lwps)

zone.max-msg-ids     (preferred: max-msg-ids)

zone.max-processes   (preferred: max-processes

zone.max-sem-ids     (preferred: max-sem-ids)

zone.max-shm-ids     (preferred: max-shm-ids)

zone.max-shm-memory  (preferred: max-shm-memory)

zone.max-swap