This has happened twice now - for regularly scheduled Benchmarks

2005-06-15 19:38:25 [---] Running CPU benchmarks
2005-06-15 19:38:33 [---] Aborting CPU benchmarks, one or more active tasks are still running.
2005-06-15 19:38:33 [---] Aborting CPU benchmarks, one or more active tasks are still running.
2005-06-15 19:38:34 [---] Resuming computation and network activity
2005-06-15 19:38:34 [---] request_reschedule_cpus: Resuming activities
2005-06-15 19:38:36 [---] request_reschedule_cpus: process exited

It on Windows 2k with CP/PP/Einstein in an equal mix.

---

This has happened twice now - for regularly scheduled Benchmarks

2005-06-15 19:38:25 [---] Running CPU benchmarks
2005-06-15 19:38:33 [---] Aborting CPU benchmarks, one or more active tasks are still running.
2005-06-15 19:38:33 [---] Aborting CPU benchmarks, one or more active tasks are still running.
2005-06-15 19:38:34 [---] Resuming computation and network activity
2005-06-15 19:38:34 [---] request_reschedule_cpus: Resuming activities
2005-06-15 19:38:36 [---] request_reschedule_cpus: process exited

It on Windows 2k with CP/PP/Einstein in an equal mix.

If the currently running Science Application does not stop when told, this is the error you get. At this point, you are seeing the request made to start benchmarks, one or more science applications does not halt, the time passes, benchmarking is abandoned, just in time for the process to be exited requireing a new scheduling. Probably the time delay for allowing the processes to cleanly exit is not long enough in the benchmarking code. THis seems to be happening more frequently with the later editions, CPDN is also susceptible to this.

I suspect that the changes made for things like checkpoining and the like that increases the stabillity may be making the shut-down process longer than it was previously.

---

This has happened twice now - for regularly scheduled Benchmarks

2005-06-15 19:38:25 [---] Running CPU benchmarks
2005-06-15 19:38:33 [---] Aborting CPU benchmarks, one or more active tasks are still running.
2005-06-15 19:38:33 [---] Aborting CPU benchmarks, one or more active tasks are still running.
2005-06-15 19:38:34 [---] Resuming computation and network activity
2005-06-15 19:38:34 [---] request_reschedule_cpus: Resuming activities
2005-06-15 19:38:36 [---] request_reschedule_cpus: process exited

It on Windows 2k with CP/PP/Einstein in an equal mix.

If the currently running Science Application does not stop when told, this is the error you get. At this point, you are seeing the request made to start benchmarks, one or more science applications does not halt, the time passes, benchmarking is abandoned, just in time for the process to be exited requireing a new scheduling. Probably the time delay for allowing the processes to cleanly exit is not long enough in the benchmarking code. THis seems to be happening more frequently with the later editions, CPDN is also susceptible to this.

I suspect that the changes made for things like checkpoining and the like that increases the stabillity may be making the shut-down process longer than it was previously.

But it seems to be a bug, does it not? If the CC doesn't wait long enough for the cruncher to stop, maybe the CC needs to wait longer!

It's a bug..... Prior to the messages I posted, I did get one that CP was suspended and removed from memory.

---