Since Friday my and my colleagues’ GROMACS jobs running on baobab fail after a few minutes. Some fail without a clear message, which I think gets lost during the output writing process, but few of them die with the following error
Fatal error: XTC error. This indicates you are out of disk space, or a simulation with major instabilities resulting in coordinates that are NaN or too large to be represented in the XTC format.
suggesting that there is no space available (simulations are fine and run clean on local hardware). We are running on scratch, which is not full, so I wonder what might be the culprit?
We have updated a system parameter, and the storage issue should now be resolved.
Please let us know if you experience any further I/O performance or stability issues.