Bugzilla – Bug 3205
globus-scheduler-provider-fork script does not get the correct value for totalnodes
Last modified: 2005-04-21 11:12:43
You need to log in before you can comment on or make changes to this bug.
I found this issue on Linux on zSeries (s390) m/c that $GLOBUS_LOCATION/libexec/globus-scheduler-provider-fork comes up with wrong number for totalnodes O/P format of /proc/cpuinfo on RH and SLES 9.0 on s390 m/c which is clear that there is only 1 processor. cat /proc/cpuinfo vendor_id : IBM/S390 # processors : 1 bogomips per cpu: 914.22 processor 0: version = FF, identification = 029804, machine = 2064 O/P of globus-scheduler-provider-fork is as shown below which gives the value as 2 which is not expected. <scheduler xmlns="http://mds.globus.org/batchproviders/2004/09" > <Queue name="default"> <totalnodes>2</totalnodes> <freenodes>0</freenodes> <maxCount>1</maxCount> <maxtime>unknown</maxtime> <maxCPUtime>unknown</maxCPUtime> <maxReqNodes>unknown</maxReqNodes> <maxRunningJobs>unknown</maxRunningJobs> <maxJobsInQueue>unknown</maxJobsInQueue> <maxTotalMemory>unknown</maxTotalMemory> <maxSingleMemory>unknown</maxSingleMemory> <whenActive>unknown</whenActive> <dispatchType>immediate</dispatchType> <status>enabled</status> </Queue> </scheduler> Also will the remaining parameters whose value which says "unknown" be implemented in the future releases ?
Some of these do not make much sense to Fork, but do for a scheduler. Hopefully, we will be able to populate more of this information in future releases. Do you have a specific need from the unpopulated information? Why do you ask? Any requirements that we can gather to prioritize our efforts is helpful.
fixed in CVS!