Are there any known issues running Nastran (application and data) from a Lustre filesystem?

We have recently switched our main, shared storage platform to a Lustre filesystem-based disk array and, since doing so Nastran jobs appear to have increased in time to complete and occasionally hang entirely (only to work fine on a re-run).

Reports are that the processes go into a 'wait' state of "futex_wait_queue_me"

Is this something that has any precedent ? We have also queried the hardware vendor and area awaiting response, but at this point only Nastran seems to be impacted.

Thank you,

Neil

0 Sanjay Patel over 4 years ago

Hi Neil, We have not seen or received any such issues. It would be interesting to know what vendor has to say. Do these jobs run successfully on other filesystems? Looking at the f04 file from two successful jobs may help determine where job is taking more time. You may consider opening a support request with relevant info and nastran version info etc.
Cancel
Vote Up 0 Vote Down

Sign in to reply

Verify Answer

Cancel
0 neil.swallow2 over 4 years ago

Hello, Thanks for the reply. Yes they ran on the filesystem we replaced. The inconsistency is the thing that's making it tricky to track down (and lack of obvious hard errors) but it's good to know you haven't seen any specific issues with Lustre - that's at least one thing I can tick off the list. I'll see if that f04 file indicates anything and, if we get no further I will raise a support request.

thanks again,

Neil
Cancel
Vote Up 0 Vote Down

Sign in to reply

Verify Answer

Cancel