Predefined Environmental Variables for OpenPBS qsub

The following environment variable reflect the environment when the user run qsub

1. PBS_O_HOST – The host where you ran the qsub command.
2. PBS_O_LOGNAME – Your user ID where you ran qsub
3. PBS_O_HOME – Your home directory where you ran qsub
4. PBS_O_WORKDIR – The working directory where you ran qsub

The following reflect the environment where the job is executing

1. PBS_ENVIRONMENT – Set to PBS_BATCH to indicate the job is a batch job, or # to PBS_INTERACTIVE to indicate the job is a PBS interactive job
2. PBS_O_QUEUE – The original queue you submitted to
3. PBS_QUEUE – The queue the job is executing from
4. PBS_JOBNAME – The job’s name
5. PBS_NODEFILE - The name of the file containing the list of nodes assigned to the job
   

iWARP, RDMA and TOE

I have captured some basic information on iWARP, RDMA, TOE and RDMA communication....

Remote Direct Access Memory Access (RDMA) allows data to be transferred over a network from the memory of one computer to the memory of another computer without CPU intervention. There are 2 types of RDMA hardware: Infiniband and RDMA over IP (iWARP). OpenFabrics Enterprise Distribution (OFED) stack provides common interface to both types of RDMA hardware.

For more information: iWARP, RDMA and TOE by Linux Cluster

I/O and filled disk error when running Molpro 2010

I encountered this error when running molpro 2010 binary on a compute node.

ERROR WRITING        32768 WORDS AT OFFSET   20630927. TO FILE 1  IMPLEMENTATION=d
f   FILE HANDLE=  1018  IERR=******
 ? Error 
 ? I/O error
 ? The problem occurs in writew
Write error in iow_direct_write; fd=12, l=32768, p=20630927; write returns -1
This may indicate a filled disk, or that a disk quota has been exceeded
1:1:fehler:: 21556614
(rank:1 hostname:node-c00.cluster.spms.ntu.edu.sg pid:3742):ARMCI DASSERT fail. 
src/armci.c:ARMCI_Error():276 cond:0  1: ARMCI aborting 21556614 (0x148ed86).
Write error in iow_direct_write; fd=12, l=32768, p=20630927; write returns -1
This may indicate a filled disk, or that a disk quota has been exceeded
3:3:fehler:: 21556614
(rank:3 hostname:node-c00.cluster.spms.ntu.edu.sg pid:3745):ARMCI DASSERT fail. 
src/armci.c:ARMCI_Error():276 cond:0  3: ARMCI aborting 21556614 (0x148ed86).
0:0:fehler:: 21556614
(rank:0 hostname:node-c00.cluster.spms.ntu.edu.sg pid:3741):ARMCI DASSERT fail. src/armci.c:ARMCI_Error():276 cond:0
  0: ARMCI aborting 21556614 (0x148ed86).
Write error in iow_direct_write; fd=12, l=32768, p=20630927; write returns -1
This may indicate a filled disk, or that a disk quota has been exceeded
2:2:fehler:: 21556614(rank:2 hostname:node-c00.cluster.spms.ntu.edu.sg pid:3744):ARMCI DASSERT fail. src/armci.c:ARMCI_Error():276 cond:0
  2: ARMCI aborting 21556614 (0x148ed86).
Write error in iow_direct_write; fd=12, l=32768, p=20590741; write returns -1
This may indicate a filled disk, or that a disk quota has been exceeded
5:5:fehler:: 21556614
(rank:5 hostname:node-c00.cluster.spms.ntu.edu.sg pid:3747):ARMCI DASSERT fail. src/armci.c:ARMCI_Error():276 cond:0
  5: ARMCI aborting 21556614 (0x148ed86).

As the error message suggest, there is a filled disk / partition that is used by molpro 2010. Looks for your molprop

• Scratch file directories, 
• /tmp
• quota set by administrators.

All these will cause errors above

Goodbye to VSphere vRAM licensing

With the upcoming release of VSphere 5.1, Vmware is removing the vRAM licensing requirements and returning to previously CPU-based licensing model. You may want to read this interesting article on what is coming up beside removing vTax as MS coined it.

Information:

1. For a good summary of the new features in VSphere 5.1, do look at VMware releases vSphere 5.1
2.  Wave good-bye to VMware's unloved vSphere vRAM 'vTax'

Encountering PBS chdir /home/user1 failed. No such file or directory

If you encounter an error after you have qsub on Torque

PBS: chdir to /home/user1 failed: No such file or directory.

If you set OpenPBS to mail, you may see the issues more clearly
Post job file processing error; job 17676.headnode-h00 on host node-c05/7+node-c05/6+node-c05/5+node-c05/4+node-c05/3+node-c05/2+node-c05/1+node-c05/0Unknown resource type  REJHOST=node-c05 MSG=invalid home directory '/home/user1' specified, errno=2 (No such file or directory)

Apparently, the node 5 has the /home directory unmounted and  the job sent to them will be lost. The solution is very simple, remount the /home directory back again