[Mauiusers] laod incorrectly distributed

Daniel Bourque dbourque at weatherdata.com
Wed Apr 16 08:10:02 MDT 2008


Hi,

    I got a 2 node cluster I'm testing to get familiar to torque/maui 
since we will soon be installing a 100+ cpu cluster. labc01n01 ( 
headnode/scheduler/worker ) and labc01n02 ( worker )


I cheated torque and said that each node has 20 CPUS in order to 
timeshare. My current maui node allocation policy is CPULOAD.

When I do a bunch of "sleep 300 | qsub"  , 1 job goes to labc01n01 and 
the rest goes to
labc01n02.  I ran other programs on labc01n02 to get the load higher 
than labc01n01 but new jobs still all goes to labc01n02...

here is a checknode -v output


checking node labc01n01

State:   Running  (in current state for 00:00:00)
Expected State:     Idle   SyncDeadline: Wed Apr 16 09:06:08
Configured Resources: PROCS: 20  MEM: 1002M  SWAP: 2916M  DISK: 1M
Utilized   Resources: [NONE]
Dedicated  Resources: PROCS: 1
Opsys:         linux  Arch:      [NONE]
Speed:      1.00  Load:       0.000
Location:   Partition: DEFAULT  Frame/Slot:  1/1
Network:    [DEFAULT]
Features:   [NONE]
Attributes: [Batch]
Classes:    [batch 0:1]

Total Time: 5:14:20:43  Up: 5:04:58:01 (93.02%)  Active: 1:31:16 (1.13%)

Reservations:
  Job '76'(x1)  -00:03:15 -> 00:56:45 (1:00:00)
JobList:  76
ALERT:  node has 1 procs dedicated but load is low (0.000)





[root at labc01n01 ~]# checknode -v labc01n02


checking node labc01n02

State:   Running  (in current state for 00:00:00)
Expected State:     Idle   SyncDeadline: Sat Oct 24 07:26:40
Configured Resources: PROCS: 20  MEM: 2018M  SWAP: 3950M  DISK: 1M
Utilized   Resources: [NONE]
Dedicated  Resources: PROCS: 3
Opsys:         linux  Arch:      [NONE]
Speed:      1.00  Load:       0.130
Location:   Partition: DEFAULT  Frame/Slot:  1/1
Network:    [DEFAULT]
Features:   [NONE]
Attributes: [Batch]
Classes:    [batch 17:50]

Total Time: 5:14:20:43  Up: 4:07:40:18 (77.17%)  Active: 1:36:26 (1.20%)

Reservations:
  Job '77'(x1)  -00:02:01 -> 00:57:59 (1:00:00)
  Job '78'(x1)  -00:01:52 -> 00:58:08 (1:00:00)
  Job '79'(x1)  -00:01:52 -> 00:58:08 (1:00:00)
JobList:  77,78,79
ALERT:  node has 3 procs dedicated but load is low (0.130)



Any insight would be appreciated.

Thanks


-- 
Daniel Bourque
Sr. Systems Engineer
WeatherData Service Inc
An Accuweather Company




More information about the mauiusers mailing list