Diagnostics - MPI Ping-Pong Quick Check - Failed to Run
-
24/جمادى الأولى/1432 01:45 ص
Hello Everyone,
The diagnostic test mentioned in the subject line along with the other MPI Ping-Pong test fails with "Failed to Run" on a new installation with one head node (online), one compute node, and two workstation nodes. The exception message detailed in this error states that this test needs more than one node to run which caused the Pre chack to fail. However, I do have two nodes running, the head node and one compute node (not to mention two workstation nodes). All other tests pass (except for one about having deployment services running) Any thoughts on how to investigate and fix this?
Thanks in advance,
Bob
جميع الردود
-
27/جمادى الأولى/1432 04:12 ص
Update: I got a sample that does Asian option pricing using excel to work correctly for all the nodes, so clearly the nodes are able to communicate with the head node.
Bob
-
01/جمادى الثانية/1432 05:16 م
Hello Bob,
Can you provide more info as how you start the diagnostic testing?
You may try the following to see whether it works:
1. In the Node management console, in the list of nodes, select any nodes (at least two) you want to run diagnostic tests
2. Right click the mouse and select Run Diagnostics...
3. In the popup windows, Select the MPI Performance, MPI Pingpong Latency
4. Click Run
5. Go to Diagnostics panel
6. Check the test results
Please let me know whehter this works for you.
Thanks,
James
- تم وضع علامة كإجابة بواسطة MadScientistBob 01/جمادى الثانية/1432 09:10 م
-
01/جمادى الثانية/1432 09:10 م
It seems to be working now, and I was testing the way you describe previously. I have been going through the book Windows HPC Server step by step and something the book told me to do must have fixed the issue.
Bob