[torqueusers] compute nodes appear down

vitor at ceset.unicamp.br vitor at ceset.unicamp.br
Wed Jul 2 14:21:13 MDT 2008


Hi,

I have a cluster with 5 machines (1 headnode and 4 compute nodes). I
installed torque-2.3.0 in the headnode. Using make packages I installed
pbs_mom in the compute nodes.
When I tried to look for the nodes I got (pbsnodes -a)
-----------------
holmes
     state = free
     np = 4
     ntype = cluster
     status = opsys=linux,uname=Linux holmes 2.6.25.6-55.fc9.x86_64 #1 SMP
Tue Jun 10 16:05:21 EDT 2008
x86_64,sessions=10200,nsessions=1,nusers=1,idletime=1,totmem=24551324kb,availmem=23979164kb,physmem=8173068kb,ncpus=?
15201,loadave=0.10,netload=27379307,state=free,jobs=,varattr=,rectime=1215027438

a1
     state = down
     np = 4
     ntype = cluster

a2
     state = down
     np = 4
     ntype = cluster

a3
     state = down
     np = 4
     ntype = cluster

a4
     state = down
     np = 4
     ntype = cluster
-----------
Using /usr/local/sbin/pbs_mom -D on the headnode I got no problem but in
the compute nodes I got
MOM is up
do_rpp: cannot get protocol Premature end of message

Any suggestions of what is causing that?

Thanks,
Vitor



More information about the torqueusers mailing list