-
Notifications
You must be signed in to change notification settings - Fork 0
/
README
44 lines (31 loc) · 1.7 KB
/
README
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
Automated simulation running tools!
Stop living your life beholden to your simulations. Let them take
care of themselves. This set of scripts will requeue your
simulations, sending you emails updating you on your job's progress or
when your simulations run into trouble. In your job script, replace
the call to running Enzo with a call to simrun.pl. For example, if
your job script says the following:
mpirun -np 16 ./enzo.exe -d simulation.enzo >& estd.out
Change it to:
./simrun.pl -mpi "mpirun -np 16" -pf simulation.enzo
For simulation restarts, simrun.pl will check the OutputLog file to
pick up the last dataset written and restart from there. For
additional options, type:
./simrun.pl -h
For best results, set the following parameter in your parameter file.
NumberOfOutputsBeforeExit = 1
This will tell Enzo to exit cleanly after writing a single dataset.
After shutting down, simrun.pl will determine if there is enough time
remaining to reach the point where Enzo will write another dataset.
If not, simrun.pl will requeue the job and end the current job.
GETTING EMAIL UPDATES ON SIMULATION PROGRESS
Most supercomputers do not allow the compute nodes to see the
internet. In order for emails to be sent on job progress, a separate
process must be run from a login node. The postman.pl script will
monitor directories provided on the command line for messages left by
simrun.pl. The best way to do this is to run postman.pl from a screen
session on a login node. Simply list all the simulation directories
which are to be watched. For example,
./postman.pl simulation_dir_1 simulation_dir_2 ... &
The code will leave a log indicating how many directories it is
following and when it finds and sends messages.