Setting up UM environment variables on ARCHER (need permission)

[tcowan]

Hi,

I'm a new UM user and I'm trying to get environment variables set in my .profile file on ARCHER (as per ​http://puma.nerc.ac.uk/trac/UM_TUTORIAL/wiki/UmTutorial/SettingUp).

I'm getting a permission denied when trying to access 8.2. Could I please be added permission?

Regards,
Tim

[willie]

Hi Tim,

What error messages do you get when you login to ARCHER? When does the "permission denied" message appear?

Regards

Willie

[tcowan]

Hi Willie,

I don't actually get any error messages when logging onto to ARCHER. I was receiving the message when I was submitting my model run on PUMA. I assumed it had to do with permissions when trying to access /work/n02/n02/hum/vn8.2/ — But I realise now I'm using vn8.5 and I've set that as my $UMDIR. So perhaps that is not why my job isn't running on ARCHER??

Is there a .log file that can help me find where my job is falling over? I have checked $HOME/output/ — but nothing is there.

When I submitted my latest job it stated:

Your job directory on host login.archer.ac.uk is: /home/n02/n02/tcowan/umui_runs/xmvqa-222145829

Cray PrgEnv? already loaded
craype-network-aries
cce/8.3.7
cray-libsci/13.0.1
PrgEnv?-cray/5.2.56
craype-ivybridge
cray-mpich/7.1.1
gcc/5.1.0
Total PEs requested: 96
NOTE: The following has been selected for running on the CRAY XC30

96 MPI task(s)
8 node(s)
12 MPI task(s) per node
2 OpenMP thread(s) per task

How can tell if it's running? I tried a qstat -u tcowan — but nothing came up.

Cheers,
Tim

[willie]

Hi Tim,

The hand edit

/home/simon.tett/umui_jobs/hand_edits/vn8.5/l_solvar2

only has permission for Simon to execute. You can see this in the file umui_jobs/xmvqa/EXT_SCRIPT_LOG on PUMA.

You need to take a copy of this file and make it executable (chmod +x <file>) and change the UMUI entry to your version. Or maybe get Simon to chmod g+rx the file.

Generally errors are presented on screen somewhere. Logs can be found on PUMA in ext.out files for the job (find ~ -name ext.out), also in the umui_jobs directory and on ARCHER in the output directory. You have nothing there, but if you look in the umui_runs on ARCHER you can see each failed attempt.

Regards

Willie

[tcowan]

Hi Willie,

He's now changed the permissions, however I think the problem may lie with at the bottom of my ext.out file:

Permission denied (publickey,keyboard-interactive).
# Time taken: 1 s⇒ ssh -n -oBatchMode=yes login.archer.ac.uk mkdir -p /home/n02/n02/stett2/build/xmvqa/baserepos/UMATMOS/cfg
[FAIL] ssh -n -oBatchMode=yes login.archer.ac.uk mkdir -p /home/n02/n02/stett2/build/xmvqa/baserepos/UMATMOS/cfg failed (255) at /home/fcm/fcm-2016.02.0/bin/../lib/FCM1/Dest.pm line 755.

Do I have to change this to my directory?

Thanks for all your help so far. It's much appreciated.

Cheers,
Tim

[willie]

Hi Tim,

The job set up is fine. I think this is a problem with your SSH setup. Please make sure that you have followed the instructions at http://cms.ncas.ac.uk/wiki/ArcherSshAgent.

On PUMA you will need to

ssh-add

and enter your pass phrase before submitting jobs from the UMUI. What does

ssh-add -l

give you?

Regards

Willie

[tcowan]

Hi Willie,

ssh-add -l gives me:

2048 SHA256:rY/H4/4R3F0klnAUAq7/lU4wp45aGHWvBKCKbOpNtnM /home/tcowan/.ssh/id_rsa (RSA)

I followed the SSH setup yesterday, so I think that part is fine.

Tim

[tcowan]

Hi Willie,

I think I've fixed the ssh problem now so now my ext.out file shows that everything has been extracted to /$HOME/build/… — however the job is still not submitting to the queue on ARCHER.

Any suggestions?

Cheers,
Tim

[willie]

Hi Tim,

That's great. It seems that you have no budget in the account n02-TITAN. Can you let us know what account you'll be using?

Regards

Willie

[tcowan]

Hi Willie,

Okay. I think I'll be using n02-titan. I should have joined the n02-titan and n02-cssp groups. Does that not automatically happen when you join n02? How do I go about determining what my budget is?

Tim

[tcowan]

Hi Willie,

I was just emailing Simon Tett (I'm using his code) and he thinks you can add me to n02-titan. I tried to get Grenville Lister (from Reading) but I'm not if it's been done. If you need authorisation please contact Simon Tett (simon.tett@…).

Cheers,
Tim

[willie]

Hi Tim,

You should now have access to that account. Try your job again.

Regards

Willie

[tcowan]

Hi Willie,

Sorry for delayed response — had a meeting to attend to. It's now running fine. Thanks for all your help. Appreciate it.

Cheers,
Tim