Hi

I have instaled the torque 4.2.6.1 and maui 3.3.1.

I have nodes with GPU and nodes without GPU.

I generated two queue. une for gpu nodes and other for the others nodes.


create queue gpu
set queue gpu queue_type = Execution
set queue gpu acl_host_enable = False
set queue gpu acl_hosts = n002
set queue gpu acl_hosts += n001
set queue gpu resources_default.neednodes = gpu
set queue gpu resources_default.nodes = 1
set queue gpu resources_default.walltime = 1440:00:00
set queue gpu enabled = True
set queue gpu started = True

create queue batch
set queue batch queue_type = Execution
set queue batch acl_hosts = n006
set queue batch acl_hosts += n005
set queue batch acl_hosts += n004
set queue batch acl_hosts += n003
set queue batch resources_default.neednodes = batch
set queue batch resources_default.nodes = 1
set queue batch resources_default.walltime = 1440:00:00
set queue batch enabled = True
set queue batch started = True

the property in the nodes is asigner:
Ej:
one node with GPU

create node n001
set node n001 state = free
set node n001 np = 8
set node n001 properties = gpu
set node n001 ntype = cluster
set node n001 status = rectime=1396011140
set node n001 status += varattr=
set node n001 status += jobs=
set node n001 status += state=free
set node n001 status += netload=15685031
set node n001 status += gres=
set node n001 status += loadave=0.00
set node n001 status += ncpus=8
set node n001 status += physmem=32981792kb
set node n001 status += availmem=37490084kb
set node n001 status += totmem=38101784kb
set node n001 status += idletime=77551
set node n001 status += nusers=1
set node n001 status += nsessions=1
set node n001 status += sessions=2987
set node n001 status += uname=Linux n001 2.6.32-358.el6.x86_64 #1 SMP 
Fri Feb 22 00:31:26 UTC 2013 x86_64
set node n001 status += opsys=linux
set node n001 gpus = 1

one node without gpu

create node n003
set node n003 state = free
set node n003 np = 16
set node n003 properties = batch
set node n003 ntype = cluster
set node n003 status = rectime=1396011229
set node n003 status += varattr=
set node n003 status += jobs=
set node n003 status += state=free
set node n003 status += netload=8922119
set node n003 status += gres=
set node n003 status += loadave=0.00
set node n003 status += ncpus=16
set node n003 status += physmem=32905824kb
set node n003 status += availmem=37380212kb
set node n003 status += totmem=38025816kb
set node n003 status += idletime=81451
set node n003 status += nusers=1
set node n003 status += nsessions=1
set node n003 status += sessions=3327
set node n003 status += uname=Linux n003 2.6.32-358.el6.x86_64 #1 SMP 
Fri Feb 22 00:31:26 UTC 2013 x86_64
set node n003 status += opsys=linux

But the maui mapping the all the nodes in the all classes.

In the PBSPro the comand is

set node n001 queue = gpu
set node n002 queue = gpu
set node n003 queue = batch
set node n004 queue = batch
set node n005 queue = batch
set node n006 queue = batch

but this command is not allowed in the torque

th log in the maui is

03/28 10:17:03 ServerProcessRequests()
03/28 10:17:03 INFO:     not rolling logs (1456817 < 10000000)
03/28 10:17:03 MResAdjust(NULL,0,0)
03/28 10:17:03 MStatInitializeActiveSysUsage()
03/28 10:17:03 MStatClearUsage([NONE],Active)
03/28 10:17:03 ServerUpdate()
03/28 10:17:03 MSysUpdateTime()
03/28 10:17:03 INFO:     starting iteration 1705
03/28 10:17:03 MRMGetInfo()
03/28 10:17:03 MClusterClearUsage()
03/28 10:17:03 MRMClusterQuery()
03/28 10:17:03 MClusterClearUsage()
03/28 10:17:03 MRMClusterQuery()
03/28 10:17:03 MPBSClusterQuery(BIRLIBIRLOKUS,RCount,SC)
03/28 10:17:03 __MPBSGetNodeState(Name,State,PNode)
03/28 10:17:03 INFO:     PBS node n001 set to state Idle (free)
03/28 10:17:03 MPBSNodeUpdate(n001,n001,Idle,BIRLIBIRLOKUS)
03/28 10:17:03 MPBSLoadQueueInfo(BIRLIBIRLOKUS,n001,SC)
03/28 10:17:03 INFO:     queue 'gpu' started state set to True
03/28 10:17:03 INFO:     class to node mapping enabled for queue 'gpu'
03/28 10:17:03 INFO:     queue 'batch' started state set to True
03/28 10:17:03 INFO:     class to node mapping enabled for queue 'batch'
03/28 10:17:03 __MPBSGetNodeState(Name,State,PNode)
03/28 10:17:03 INFO:     PBS node n002 set to state Idle (free)
03/28 10:17:03 __MPBSGetNodeState(Name,State,PNode)
03/28 10:17:03 INFO:     PBS node n002 set to state Idle (free)
03/28 10:17:03 MPBSNodeUpdate(n002,n002,Idle,BIRLIBIRLOKUS)
03/28 10:17:03 MPBSLoadQueueInfo(BIRLIBIRLOKUS,n002,SC)
03/28 10:17:03 INFO:     queue 'gpu' started state set to True
03/28 10:17:03 INFO:     class to node mapping enabled for queue 'gpu'
03/28 10:17:03 INFO:     queue 'batch' started state set to True
03/28 10:17:03 INFO:     queue 'batch' started state set to True
03/28 10:17:03 INFO:     class to node mapping enabled for queue 'batch'
03/28 10:17:03 __MPBSGetNodeState(Name,State,PNode)
03/28 10:17:03 __MPBSGetNodeState(Name,State,PNode)
03/28 10:17:03 INFO:     PBS node n003 set to state Idle (free)
03/28 10:17:03 MPBSNodeUpdate(n003,n003,Idle,BIRLIBIRLOKUS)
03/28 10:17:03 MPBSLoadQueueInfo(BIRLIBIRLOKUS,n003,SC)
03/28 10:17:03 INFO:     queue 'gpu' started state set to True
03/28 10:17:03 INFO:     class to node mapping enabled for queue 'gpu'
03/28 10:17:03 INFO:     class to node mapping enabled for queue 'gpu'
03/28 10:17:03 INFO:     queue 'batch' started state set to True
03/28 10:17:03 INFO:     queue 'batch' started state set to True
03/28 10:17:03 INFO:     class to node mapping enabled for queue 'batch'
03/28 10:17:03 __MPBSGetNodeState(Name,State,PNode)
03/28 10:17:03 INFO:     PBS node n004 set to state Idle (free)
03/28 10:17:03 INFO:     PBS node n004 set to state Idle (free)
03/28 10:17:03 MPBSNodeUpdate(n004,n004,Idle,BIRLIBIRLOKUS)
03/28 10:17:03 MPBSLoadQueueInfo(BIRLIBIRLOKUS,n004,SC)
03/28 10:17:03 INFO:     queue 'gpu' started state set to True
03/28 10:17:03 INFO:     class to node mapping enabled for queue 'gpu'
03/28 10:17:03 INFO:     class to node mapping enabled for queue 'gpu'
03/28 10:17:03 INFO:     queue 'batch' started state set to True
03/28 10:17:03 INFO:     class to node mapping enabled for queue 'batch'
03/28 10:17:03 INFO:     class to node mapping enabled for queue 'batch'
03/28 10:17:03 __MPBSGetNodeState(Name,State,PNode)
03/28 10:17:03 INFO:     PBS node n005 set to state Idle (free)
03/28 10:17:03 MPBSNodeUpdate(n005,n005,Idle,BIRLIBIRLOKUS)
03/28 10:17:03 MPBSLoadQueueInfo(BIRLIBIRLOKUS,n005,SC)
03/28 10:17:03 INFO:     queue 'gpu' started state set to True
03/28 10:17:03 INFO:     class to node mapping enabled for queue 'gpu'
03/28 10:17:03 INFO:     queue 'batch' started state set to True
03/28 10:17:03 INFO:     class to node mapping enabled for queue 'batch'
03/28 10:17:03 __MPBSGetNodeState(Name,State,PNode)
03/28 10:17:03 INFO:     PBS node n006 set to state Idle (free)
03/28 10:17:03 MPBSNodeUpdate(n006,n006,Idle,BIRLIBIRLOKUS)
03/28 10:17:03 MPBSLoadQueueInfo(BIRLIBIRLOKUS,n006,SC)
03/28 10:17:03 INFO:     queue 'gpu' started state set to True
03/28 10:17:03 INFO:     class to node mapping enabled for queue 'gpu'
03/28 10:17:03 INFO:     queue 'batch' started state set to True
03/28 10:17:03 INFO:     class to node mapping enabled for queue 'batch'
03/28 10:17:03 INFO:     6 PBS resources detected on RM BIRLIBIRLOKUS
03/28 10:17:03 INFO:     class to node mapping enabled for queue 'batch'
03/28 10:17:03 INFO:     6 PBS resources detected on RM BIRLIBIRLOKUS
03/28 10:17:03 INFO:     resources detected: 6
03/28 10:17:03 MRMWorkloadQuery()
03/28 10:17:03 MRMWorkloadQuery()
03/28 10:17:03 MPBSWorkloadQuery(BIRLIBIRLOKUS,JCount,SC)
03/28 10:17:03 INFO:     queue is empty
03/28 10:17:03 INFO:     0 PBS jobs detected on RM BIRLIBIRLOKUS
03/28 10:17:03 WARNING:  no workload detected
03/28 10:17:03 MStatClearUsage(node,Active)
03/28 10:17:03 MClusterUpdateNodeState()
03/28 10:17:03 MClusterUpdateNodeState()
03/28 10:17:03 MQueueSelectAllJobs(Q,HARD,ALL,JIList,DP,Msg)
03/28 10:17:03 MQueueSelectAllJobs(Q,SOFT,ALL,JIList,DP,Msg)
03/28 10:17:03 
MQueueSelectJobs(SrcQ,DstQ,HARD,5120,4096,2140000000,EVERY,FReason,FALSE)
03/28 10:17:03 
MQueueSelectJobs(SrcQ,DstQ,SOFT,5120,4096,2140000000,EVERY,FReason,TRUE)
03/28 10:17:03 
MQueueSelectJobs(SrcQ,DstQ,HARD,5120,4096,2140000000,EVERY,FReason,TRUE)
03/28 10:17:03 
MQueueSelectJobs(SrcQ,DstQ,SOFT,5120,4096,2140000000,EVERY,FReason,TRUE)
03/28 10:17:03 MSchedUpdateStats()
03/28 10:17:03 INFO:     iteration: 1705   scheduling time:  0.003 seconds
03/28 10:17:03 MResUpdateStats()
03/28 10:17:03 INFO:     current util[1705]:  0/6 (0.00%)  PH: 0.00%  
active jobs: 0 of 0 (completed: 3)
03/28 10:17:03 MQueueCheckStatus()
03/28 10:17:03 MNodeCheckStatus()
03/28 10:17:03 MUClearChild(PID)
03/28 10:17:03 INFO:     scheduling complete.  sleeping 30 seconds


Please help my
_______________________________________________
mauiusers mailing list
[email protected]
http://www.supercluster.org/mailman/listinfo/mauiusers

Reply via email to