Using geo-hpcc cluster¶
Connecting from Windows¶
To access the command line:
- Open PuTTY
- Host name:
dh2-seismo1203.seismo.helsinki.fi
- Click Open
- Type in the given username (
proxyuser
) and password - Give command
ssh clusteruser@geo-hpfe.seismo.helsinki.fi
, whereclusteruser
is the provided username
To access the files:
- Open WinSCP
- Click New Site, fill in Host name:
geo-hpfe.seismo.helsinki.fi
, User name (clusteruser
) and Password will be provided. - Click Advanced…
- Choose Connection, Tunnel, and check Connect through SSH tunnel
- Fill in values: Host name:
dh2-seismo1203.seismo.helsinki.fi
, User name (proxyuser
) and Password - Click OK
- Click Save
- Click Login
WinSCP session will be saved, so that next time you can just double-click the hostname on the left, and the connection will be opened.
Connecting from Linux¶
To access the command line:
ssh proxyuser@dh2-seismo1203.seismo.helsinki.fi
ssh clusteruser@geo-hpfe.seismo.helsinki.fi
Note, usernames need to be replaced with the provided individual usernames.
To copy files from DOUAR output to the current directory on loca machine:
scp -oProxyCommand="ssh -W %h:%p proxyuser@dh2-seismo1203.seismo.helsinki.fi" clusteruser@geo-hpfe.seismo.helsinki.fi:/globalscratch/clusteruser/douar/modelname/OUT/*.vtk .
Running DOUAR models¶
- Modify the model input file, e.g.,
nano ~/douar/inputs/rift.txt
- Submit the job:
~/bin/submitdouar.sh -i ~/douar/inputs/rift.txt -n 16
- Monitor the job status:
squeue
- Post-process the output:
cd /globalscratch/clusteruser/douar/rift_20180514100500/OUT
~/bin/process_outbin.sh
- Note that the directory name of the output is formed from the model name (
rift
) and the date and time of the job submission
- Copy VTK files to your local machine (see instructions above) to be opened in ParaView
You can cancel your job with scancel jobid
where jobid is the numerical ID of the job, and can be found using squeue