Missing archi_submit file?

[nfreychet]

Hello,

When trying to start a job on ARCHER, after the reconfiguration step I get this message and the model stop running:

     Run started from UMUI
cp xnbrc: run on ARCHER, fix Aerosols
This job is using UM directory /work/y07/y07/umshared,
cp: cannot stat `/home/n02/n02/nfreyche/umui_runs/xnbrd-318102146/archi_submit': No such file or directory
Getfile failed. Return code  1
Trying to get  /home/n02/n02/nfreyche/umui_runs/xnbrd-318102146/archi_submit
Stopping
4911956.sdb

The job name is xnbrd

I'm not sure what this archi_submit is supposed to be?

Cheers,
Nicolas

[willie]

Hi Nicolas,

Could you set your permissions so we can read your files please

chmod -R g+rX /home/n02/n02/<your-username>
chmod -R g+rX /work/n02/n02/<your-username>

I suspect it is something to do with archiving. I notice that you are archiving to /work: this will not help you save space. It is better to archive to /nerc.

Regards
Willie

[nfreychet]

Hi Willie,

OK I opened the permissions.
I will try to change the archiving path to /nerc and see what it does.

Nicolas

[willie]

Hi Nicolas,

The file archi_submit is required by Simon Tett's archive code. Is there a missing hand edit to set this up?

Willie

[nfreychet]

Hi Willie,

Ha yes that's why it is missing. I just implemented Simon's code as a test but didn't realized I needed to copy some files too. I will be look at that in more details. Thanks a lot!

Nicolas

[nfreychet]

Hi Willie,

So I removed the archiving fcm of Simon, and tried to run the model again, but I still have the same error. Doing a "grep archi_submit" in the building directory I got these output:

umscripts/bin/UMScr_TopLevel:export ARCHIVE_SUBMIT=$UM_TMPDIR/archi_submit
umscripts/bin/UMScr_TopLevel: . getfile archi_submit $JOBDIR $ARCHIVE_SUBMIT $PRINT_INPUT

UPDATE: I just saw that I was missing the FCM archiving branch (as mentioned in http://cms.ncas.ac.uk/wiki/Archer/NercArchiving). I will add this branch and try again, to see if it solves the problem.

Nico

[willie]

Hi Nico,

For ARCHER, you do need an archiving branch however. Follow the instructions at http://cms.ncas.ac.uk/wiki/Archer/NercArchiving.

Regards
Willie

[nfreychet]

I think I edited my previous post in the same time you replied.
I will add the branch and see how it goes.

Thanks!
Nico

[nfreychet]

Hi Willie,

So I added the appropriate FCM archiving branch, but I still have the same error.

Nico

[willie]

Hi Nico,

In "Unified Model Automated Output Processing System" (UM documentation paper Y1) section 3.5, the files archi_submit and archi_script are mentioned. These are part of the UM8.6 python drivers for the MOOSE archiving system. Since you're not using UM8.6 or the MOOSE system, this code should not be invoked. Something has gone badly wrong.

I am guessing you're trying to port a job from another computer to ARCHER. It would be helpful to know what the original job was and what computer it ran on.

Regards,
Willie

[nfreychet]

Hi Willie,

It may be due to the MOOSE archiving system then.
The job I'm trying to run on ARCHER is a copy from a job (xnbrc) I was running on MONSOON, using MOOSE archiving. However, in my new job (xnbrd) I have switched to the HECToR archiving, so I don't understand why it would still call MONSOON system.

Cheers,
Nico

[willie]

Hi Nico,

I compared one of Simon Tett's jobs with yours. The salient difference is that he uses,

fcm:um_br/dev/um/vn8.5_ncas_local/src

where you used

fcm:um/branches/pkg/Config/vn8.5_ncas/src

So, if you switch off Jeff's archiving branch, switch on Simon Tett's and make the above change it will eliminate this error.

Regards,
Willie

[nfreychet]

Hi Willie,

Yes changing the FCM branch solved this problem.

Thanks a lot!
Nico