Hello Natalia,
The error says that it is trying to change to a directory named
/home/fmontenegro, but that such a directory does not exist. This is not a
GROMACS error, but a problem related to how the cluster is setup. Your
cluster administrator should be able to help with this problem.
Best
This error is just for confirming that NFS falls, that is not functioning.
But when run Gromacs, the connection with NFs is lost just few time of
compute. This not happen when run NAMD 2.9 alone.
And this is the command for the cluster:
nohup /opt/openmpi/bin/mpirun -hostfile myhostfile2
So, if NFS is not functioning, there is little we can do on this list to
help. My only guess about it is that maybe NFS is working but the home
directories have not been exported to the working nodes. Perhaps a failure
of NIS or LDAP configuration?
I do not have experience with NAMD, so I cannot
we known that gromacs not find the directories. When mount the directories
in NFS, that we dont known is why gromacs makes that NFS falls down.
Any help!
Best regards.
2014-07-03 12:36 GMT-04:00 Victor Rosas Garcia rosas.vic...@gmail.com:
So, if NFS is not functioning, there is little we can
Then you need to do some troubleshooting. Using GROMACS is a complex
process, so you have to be very systematic in narrowing down the source of
the problem.
First, see if you can run GROMACS locally, in a single machine (preferably
one of the nodes in the cluster) using only one thread and then
Hi,
GROMACS doesn't make NFS go down, but it might expose the problem that
exists anyway - which happens frequently on clusters that aren't set up
appropriately. The GROMACS checkpoint process requires that the file system
be present, so that the checkpoint process can compute and record hash
Thanks Mark and Victor. I will see the checkpoint files.
Best regards!!
2014-07-03 15:38 GMT-04:00 Mark Abraham mark.j.abra...@gmail.com:
Hi,
GROMACS doesn't make NFS go down, but it might expose the problem that
exists anyway - which happens frequently on clusters that aren't set up