Project

General

Profile

Introduction to Moab for users » History » Version 4

Miguel Dias Costa, 29/02/2012 15:50

1 1 Miguel Dias Costa
h1. Introduction to Moab for users
2 1 Miguel Dias Costa
3 2 Miguel Dias Costa
Moab is the workload management system that is used in FEUP's high performance computing clusters.
4 2 Miguel Dias Costa
5 2 Miguel Dias Costa
h2. Getting started
6 2 Miguel Dias Costa
7 2 Miguel Dias Costa
After gaining ssh accesh to a moab frontend, a user can see what resources are available and check the status of the queue by running
8 2 Miguel Dias Costa
9 2 Miguel Dias Costa
<pre>
10 2 Miguel Dias Costa
mshow
11 2 Miguel Dias Costa
</pre>
12 2 Miguel Dias Costa
13 2 Miguel Dias Costa
In order to submit jobs, one needs to create a submit script that defines the requirements and the command to be executed. A simple example to get started would be
14 2 Miguel Dias Costa
15 2 Miguel Dias Costa
h3. Example serial job submission script helloworld.submit
16 2 Miguel Dias Costa
17 2 Miguel Dias Costa
<pre>
18 2 Miguel Dias Costa
#!/bin/bash                                                
19 2 Miguel Dias Costa
#MOAB -l nodes=1:ppn=1                                    
20 2 Miguel Dias Costa
#MOAB -l partition=torque1                                 
21 2 Miguel Dias Costa
#MOAB -j oe                                          
22 2 Miguel Dias Costa
cd $PBS_O_WORKDIR                                    
23 2 Miguel Dias Costa
#!/bin/bash
24 4 Miguel Dias Costa
./command
25 2 Miguel Dias Costa
</pre>
26 2 Miguel Dias Costa
27 3 Miguel Dias Costa
h3. Example parallel job submission script helloworld.submit
28 3 Miguel Dias Costa
29 3 Miguel Dias Costa
<pre>
30 3 Miguel Dias Costa
#!/bin/bash                                                
31 3 Miguel Dias Costa
#MOAB -l nodes=2:ppn=2
32 3 Miguel Dias Costa
#MOAB -l partition=torque1                                 
33 3 Miguel Dias Costa
#MOAB -j oe                                          
34 3 Miguel Dias Costa
cd $PBS_O_WORKDIR                                    
35 3 Miguel Dias Costa
mpirun -np 4 -machinefile $PBS_NODEFILE ./command | tee $PBS_JOBID.output
36 3 Miguel Dias Costa
</pre>
37 3 Miguel Dias Costa
38 2 Miguel Dias Costa
After creating this file, the job would be submitted by running
39 2 Miguel Dias Costa
40 2 Miguel Dias Costa
<pre>
41 2 Miguel Dias Costa
msub helloworld.submit
42 2 Miguel Dias Costa
</pre>
43 2 Miguel Dias Costa
44 2 Miguel Dias Costa
h2. Checking status
45 2 Miguel Dias Costa
46 2 Miguel Dias Costa
To see details about the status of each job in the queue, one would run
47 2 Miguel Dias Costa
48 2 Miguel Dias Costa
<pre>
49 2 Miguel Dias Costa
mshow
50 2 Miguel Dias Costa
</pre>
51 2 Miguel Dias Costa
52 2 Miguel Dias Costa
To check the status of a specific job,
53 2 Miguel Dias Costa
54 2 Miguel Dias Costa
<pre>
55 2 Miguel Dias Costa
checkjob jobid
56 2 Miguel Dias Costa
</pre>
57 2 Miguel Dias Costa
58 2 Miguel Dias Costa
To check the status of a specific node,
59 2 Miguel Dias Costa
60 2 Miguel Dias Costa
<pre>
61 2 Miguel Dias Costa
checknode nodeid
62 2 Miguel Dias Costa
</pre>
63 2 Miguel Dias Costa
64 2 Miguel Dias Costa
To cancel a job
65 2 Miguel Dias Costa
66 2 Miguel Dias Costa
<pre>
67 2 Miguel Dias Costa
canceljob jobid
68 2 Miguel Dias Costa
</pre>
69 2 Miguel Dias Costa
70 2 Miguel Dias Costa
h2. More Information
71 2 Miguel Dias Costa
72 2 Miguel Dias Costa
http://www.adaptivecomputing.com/resources/docs/