Opened 4 months ago

Last modified 3 months ago

#3401 pending help

issue with JULES with suite.rc.CEDA-JASMIN

Reported by: aejigu Owned by: jules_support
Component: JULES Keywords: JULES, JASMIN, FLUXNET
Cc: tquaife Platform: JASMIN
UM Version:

Description

Hi Patrick,

I am at testing stage to run jules with rose but I get the attached error message. Can you point out the error. I am suspecting if I delete few texts in the suite.rc.CEDA-JASMIN when I edit it.

Best regards,
Amsale

Attachments (1)

IMG_7601.jpeg (1.8 MB) - added by pmcguire 4 months ago.
screenshot

Download all attachments as: .zip

Change History (37)

Changed 4 months ago by pmcguire

screenshot

comment:1 Changed 4 months ago by pmcguire

Hi Amsale:
What do you think you deleted in the suite.rc.CEDA-JASMIN?

Can you see the differences from your local edited copy of u-al752 from the version that is in MOSRS?
You can do this by fcm diff or fcm diff -r HEAD on cylc1.jasmin from the ~\roses\u-al752 directory.

Can you also give me permission to view your roses and cylc-run directories? You might need to also give me permission to view your home directory in order to do that; but you might want to remove my read/view privilege from other files first if they are private or confidential.
Patrick

comment:2 Changed 4 months ago by pmcguire

  • Reporter changed from pmcguire to aejigu
  • Summary changed from aejigu to issue with JULES with suite.rc.CEDA-JASMIN

comment:3 in reply to: ↑ description Changed 4 months ago by aejigu

Replying to aejigu:

Hi Patrick,

I am at testing stage to run jules with rose but I get the attached error message. Can you point out the error. I am suspecting if I delete few texts in the suite.rc.CEDA-JASMIN when I edit it.

Best regards,
Amsale

Hi Patrick,
Using the fcm idff, I can see only the recommended changes.

Last edited 4 months ago by aejigu (previous) (diff)

comment:4 follow-up: Changed 4 months ago by pmcguire

Hi Amsale:
I wonder what could be wrong then, if you have done the recommended changes?
Can you copy & paste here what fcm diff says on cylcl1.jasmin for your u-al752?

Also, I still can't access your directories:

[pmcguire@cylc1 ~]$ cd ~amsale/
-bash: cd: /home/users/amsale/: Permission denied
(base) [pmcguire@cylc1 ~]$ cd ~amsale/roses
-bash: cd: /home/users/amsale/roses: Permission denied

Can you fix that?
Patrick

comment:5 in reply to: ↑ 4 Changed 4 months ago by aejigu

Replying to pmcguire:

Hi Amsale:
I wonder what could be wrong then, if you have done the recommended changes?
Can you copy & paste here what fcm diff says on cylcl1.jasmin for your u-al752?

Also, I still can't access your directories:

[pmcguire@cylc1 ~]$ cd ~amsale/
-bash: cd: /home/users/amsale/: Permission denied
(base) [pmcguire@cylc1 ~]$ cd ~amsale/roses
-bash: cd: /home/users/amsale/roses: Permission denied

Can you fix that?
Patrick

Hi Patrick,

Sorry, can you try now?

comment:6 Changed 4 months ago by pmcguire

Hi Amsale:
Yes. That works! Thanks for giving me the permissions.

I did an diff -r ~pmcguire/roses/u-al752 ~amsale/roses/u-al752, and I do note two typos in your suite:

1) You misspelled suite in:
SUITE_DATA='/gws/nopw/j04/jules/pmcguire/fluxnet/kwilliam/suite_data'
It should be spelled with the e. And if you didn't know, suite is pronounced like sweet.

2) You misspelled CEDA_JASMIN as CEDA-JASMIN for the LOCATION.

There are also other differences, which are because I haven't checked in the changes I made (after the SLURM upgrade) into MOSRS. I will check those changes soon and I will update the documentation soon. But for now, there are some extra steps in the tutorial that I am writing here for you now:
1) Add these lines to app/fcm_make/file/fcm-make.cfg:

build.prop{fc.include-paths} =  /home/users/siwilson/netcdf.openmpi/include
build.prop{fc.lib-paths} =  /home/users/siwilson/netcdf.openmpi/lib /gws/nopw/j04/jules/admin/curl/curl-lotus-parallel-intel/lib

2) Change ~/roses/u-al752/site/suite.rc.CEDA_JASMIN:

   [[JASMIN]]
        env-script = """
                 module add parallel-netcdf/intel
                 module list 2>&1
                 env | grep LD_LIBRARY_PATH
                 export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$HDF5_LIBDIR
                 env | grep LD_LIBRARY_PATH
                 """

to:

   [[JASMIN]]
        env-script = """
                 export PATH=/apps/jasmin/metomi/bin:$PATH
                 module load contrib/gnu/gcc/8.2.0
                 module load eb/OpenMPI/intel/3.1.1
                 module load jaspy/2.7
                 module list 2>&1
                 env | grep LD_LIBRARY_PATH
                 export NETCDF_FORTRAN_ROOT=/home/users/siwilson/netcdf.openmpi/
                 export NETCDF_ROOT=/home/users/siwilson/netcdf.openmpi/
                 export HDF5_LIBDIR=/home/users/siwilson/netcdf.openmpi/lib
                 module load intel/19.0.0
                 export I_MPI_PMI_LIBRARY=/usr/lib64/libpmi.so
                 export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$HDF5_LIBDIR
                 env | grep LD_LIBRARY_PATH
                 """

3) Change ~/roses/u-al752/site/suite.rc.CEDA_JASMIN:

    [[FCM_MAKE_CEDA_JASMIN]]
        inherit = None, JASMIN_BACKGROUND

        [[[environment]]]
            JULES_BUILD=normal
            JULES_OMP=noomp
            JULES_PLATFORM=jasmin-lotus-intel

to:

    [[FCM_MAKE_CEDA_JASMIN]]
        inherit = None, JASMIN_BACKGROUND

        [[[environment]]]
            JULES_BUILD=normal
            JULES_OMP=noomp
            NETCDF_FORTRAN_ROOT=/home/users/siwilson/netcdf.openmpi/
            NETCDF_ROOT=/home/users/siwilson/netcdf.openmpi/
            JULES_PLATFORM=jasmin-lotus-intel

Can you let me know if making these 5 sets of changes helps you to get the suite running?
Patrick

comment:7 follow-up: Changed 4 months ago by pmcguire

Hi Amsale:
My 2nd step number 2 above wasn't working. So this is the new 2nd step number 2:

Change ~/roses/u-al752/site/suite.rc.CEDA_JASMIN:

   [[JASMIN]]
        env-script = """
                 module add parallel-netcdf/intel
                 module list 2>&1
                 env | grep LD_LIBRARY_PATH
                 export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$HDF5_LIBDIR
                 env | grep LD_LIBRARY_PATH
                 """

to:

   [[JASMIN]]
        env-script = """
                eval $(rose task-env)
                export PATH=/apps/jasmin/metomi/bin:$PATH
                module load intel/19.0.0
                module load contrib/gnu/gcc/7.3.0
                module load eb/OpenMPI/intel/3.1.1
                module list 2>&1
                env | grep LD_LIBRARY_PATH
                export NETCDF_FORTRAN_ROOT=/home/users/siwilson/netcdf_par/3.1.1/intel.19.0.0/
                export NETCDF_ROOT=/home/users/siwilson/netcdf_par/3.1.1/intel.19.0.0/
                export HDF5_LIBDIR=/home/users/siwilson/netcdf_par/3.1.1/intel.19.0.0/lib
                export I_MPI_PMI_LIBRARY=/usr/lib64/libpmi.so
                export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$HDF5_LIBDIR
                env | grep LD_LIBRARY_PATH
                """

Patrick

comment:8 Changed 4 months ago by pmcguire

Hi Amsale:
One more thing:

I don't know if the test queue will work well now with this suite with SLURM.
And I don't know if you're still considered a new user or now, so I don't know if you are restricted from using the short-serial queue.

One thing to note is that the queueing time for the short-serial queue is rather long currently at times.

If the test queue doesn't work for you and if you're allowed to use the short-serial queue, and if the short-serial queue's queueing time is too long, you might consider switching from either test or short-serial to short-serial-4hr.

The queueing time for short-serial-4hr is currently rather short.
Patrick

comment:9 in reply to: ↑ 7 Changed 4 months ago by aejigu

Hi Patrick,

Thanks!
I incorporate your suggested corrections and I still get error message but a different one :).

###
[FAIL] file:bin/fluxnet_evaluation.py=source=fcm:jules.x_br/pkg/karinawilliams/r6715_python_packages/share/fluxnet_evaluation.py@18106: bad or missing value
###

Also, in the /work/scratch folder I can't see the scratch-nopw but scratch-pw, hence I changed the jules output directory accordingly and created the folder. Will that be another problem.

Thanks.
Amsale

Last edited 4 months ago by pmcguire (previous) (diff)

comment:10 Changed 4 months ago by pmcguire

Hi Amsale:
That error message that you got suggests that you are not logged in to MOSRS. Did you set that MOSRS login (asking for a password) in your bash startup files? Sometimes, it forgets the password, and the easiest thing to do is log out and log back in, when it then asks for the MOSRS password again.

Yes, I saw the same thing yesterday, when I tried to run the suite, with output going to /work/scratch-nopw . That partition didn't exist yesterday, or at least it wasn't visible from cylc1.jasmin. It might not have been visible from SLURM/LOTUS either. I have raised a ticket with the JASMIN Helpdesk about that. You can do the same thing if things like that happen to you, by emailing to support AT jasmin.ceda.uk . I temporarily did the same thing you did of changing the /work/scratch-nopw to /work/scratch-pw.

I also got error messages when I ran it yesterday that the job didn't have enough wallclock time. This was due to a mistake wherein we used the MM:SS format for the time specification instead of the HH:MM:SS format. We were only asking for 2:00 minutes of wallclock instead of 2:00:00 hours. You can change the two --time entries in ~/roses/u-al752/site/suite.rc.CEDA_JASMIN to be in HH:MM:SS format by adding :00 to the end of each string to indicate the wallclock time properly.

Patrick

comment:11 follow-up: Changed 4 months ago by pmcguire

Hi Amsale:
The JASMIN folks fixed the problem. They say that "/work/scratch-nopw is now mounted on cylc1.jasmin.ac.uk".
Patrick

comment:12 in reply to: ↑ 11 ; follow-up: Changed 4 months ago by aejigu

Replying to pmcguire:

Hi Patrick,

Thanks for the update!
Amsale

Hi Amsale:
The JASMIN folks fixed the problem. They say that "/work/scratch-nopw is now mounted on cylc1.jasmin.ac.uk".
Patrick

comment:13 in reply to: ↑ 12 Changed 4 months ago by aejigu

Hi Patrick,

The rose suite-run now pops a gcylc window but the fcm_make fails. I investigated the log files:

~/cylc-run/u-al752/log/job/1/fcm_make/01/job.err
~/cylc-run/u-al752/log/job/1/fcm_make/01/job.out

but can't tell what is going wrong.
Could you have a look at the following files please?

Thanks,
Amsale

Replying to aejigu:

Replying to pmcguire:

Hi Patrick,

Thanks for the update!
Amsale

Hi Amsale:
The JASMIN folks fixed the problem. They say that "/work/scratch-nopw is now mounted on cylc1.jasmin.ac.uk".
Patrick

comment:14 Changed 4 months ago by pmcguire

Hi Patrick,

Quick question: I am assuming that I need to mkdir for the output file and plots. Or, is it something created automatically?

I still get the fcm_make file failing when I execute rose suite-run.

Best regards,
Amsale

comment:15 follow-up: Changed 4 months ago by pmcguire

Hi Amsalework
Yes.
If you do this:

cd ~\roses\u-al752
grep -r mkdir *

then you will see that mkdir is used with the -p flag in the suite.rc file to create the directories that you specify.

Yes, when I try to run a copy of your suite, I get the same error that you do:

[FAIL] https://code.metoffice.gov.uk/svn/jules/main/trunk/etc/fcm-make/platform/custom.cfg@18213:6: reference to undefined variable

[FAIL] include =

[FAIL] undef($JULES_REMOTE)

[FAIL] fcm make -f /work/scratch-pw/pmcguire/cylc-run/u-al752amsale1/work/1/fcm_make/fcm-make.cfg -C /home/users/pmcguire/cylc-run/u-al752amsale1/share/fcm_make -j 4 # return-code=255

2020-10-20T12:01:05+01:00 CRITICAL - failed/EXIT

But, when I make these changes to a copy of your suite (as summarized by diff -r), it seems to start and finish the fcm_make compiling process properly:

diff -r ~amsale/rosesu-al752/ ~pmcguire/roses/u-al752amsale1/

diff -r /home/users/amsale/rosesu-al752/app/fcm_make/file/fcm-make.cfg /home/users/pmcguire/roses/u-al752amsale1/app/fcm_make/file/fcm-make.cfg

3,8c3,4

\{{{

build.prop{fc.include-paths} = /home/users/siwilson/netcdf.openmpi/include

build.prop{fc.lib-paths} = /home/users/siwilson/netcdf.openmpi/lib /gws/nopw/j04/jules/admin/curl/curl-lotus-parallel-intel/lib

\}}}

—-

build.prop{fc.include-paths} = /home/users/siwilson/netcdf.openmpi/include

build.prop{fc.lib-paths} = /home/users/siwilson/netcdf.openmpi/lib /gws/nopw/j04/jules/admin/curl/curl-lotus-parallel-intel/lib

diff -r /home/users/amsale/rosesu-al752/site/suite.rc.CEDA_JASMIN /home/users/pmcguire/roses/u-al752amsale1/site/suite.rc.CEDA_JASMIN

11,16c11,13

export NETCDF_FORTRAN_ROOT=/home/users/siwilson/netcdf_par/3.1.1/intel.19.0.0/

export NETCDF_ROOT=/home/users/siwilson/netcdf_par/3.1.1/intel.19.0.0/

export HDF5_LIBDIR=/home/users/siwilson/netcdf_par/3.1.1/intel.19.0.0/lib

—-

export NETCDF_FORTRAN_ROOT=/home/users/siwilson/netcdf_par/3.1.1/intel.19.0.0/

export NETCDF_ROOT=/home/users/siwilson/netcdf_par/3.1.1/intel.19.0.0/

export HDF5_LIBDIR=/home/users/siwilson/netcdf_par/3.1.1/intel.19.0.0/lib

47,48c44

NETCDF_FORTRAN_ROOT=/home/users/siwilson/netcdf.openmoi/

—-

NETCDF_FORTRAN_ROOT=/home/users/siwilson/netcdf.openmpi/

50c46,47

NETCDF_ROOT = /home/users/siwilson/netcdf.openmpi/

JULES_PLATFORM=jasmin-lotus-intel

—-

NETCDF_ROOT=/home/users/siwilson/netcdf.openmpi/

JULES_PLATFORM=jasmin-lotus-intel

There were several typos, extra characters, extra spaces, missing lines, joined lines, and line splits.

Does this help?
Patrick

Last edited 4 months ago by pmcguire (previous) (diff)

comment:16 in reply to: ↑ 15 Changed 4 months ago by aejigu

Replying to pmcguire:

Ah, Thanks Patrick. Which means I need to delete the folders I created then?
Would be also great if you give me permission for me your copy of the rose suite to compare those changes myself.

Thanks.
Amsale

Last edited 4 months ago by pmcguire (previous) (diff)

comment:17 Changed 4 months ago by pmcguire

  • Cc tquaife added

Hi Amsalework:
If you already created the folders, then maybe you don't need to delete them.
But if you rerun the job, you might want to set new values for the names of the output folder in the script, so that it doesn't overwrite or partially overwrite the older folder.

I added tquaife to the recipient list of this ticket, so you don't necessarily need to resend the ticket comments to him by email.
But if there is something private or confidential that isn't appropriate for a ticket, email is fine.

You should already have permission to read the folders.
Patrick

Last edited 4 months ago by pmcguire (previous) (diff)

comment:18 Changed 4 months ago by aejigu

Thanks Patrick, that is helpful!

comment:19 Changed 4 months ago by aejigu

Hi Patrick,

Sorry, can you double check for me the permission, I can't access your files.

Thanks,
Amsale

-bash-4.2$ diff -r /home/users/pmcguire/roses/u-al752amsale1/site/suite.rc.CEDA_JASMIN /home/users/amsale/roses/u-al752/site/suite.rc.CEDA_JASMIN
diff: /home/users/pmcguire/roses/u-al752amsale1/site/suite.rc.CEDA_JASMIN: Permission denied

comment:20 Changed 4 months ago by pmcguire

Hi Amsale:
I just fixed that. Can you try again?
Thanks,
Patrick

comment:21 Changed 4 months ago by aejigu

Hi Patrick,

It works, thanks.
Amsale.

comment:22 Changed 4 months ago by pmcguire

Hi Amsale
So everything works now? Does it compile and run?
Patrick

comment:23 Changed 4 months ago by aejigu

Hi Patrick,

The fcm_make is completed but the jules still fails after submission.
The two files we compare previously look the same now.

Thanks,
Amsale

comment:24 Changed 4 months ago by pmcguire

Hi Amsale:
I am glad you were able to get it to compile!! That is good progress.

According to the error message you got, in:
~amsale/cylc-run/u-al752/log/job/1/jules_no_adv_presc0/01/job.err
it says among other things:
"Please verify that both the operating system and the processor support Intel® X87, CMOV, MMX, FXSAVE, SSE, SSE2, SSE3, SSSE3, SSE4_1, SSE4_2 and POPCNT instructions."

This is the same error message as in this ticket:
http://cms.ncas.ac.uk/ticket/3390#comment:9
My current best advice for what to try for that ticket is here:
http://cms.ncas.ac.uk/ticket/3390#comment:22

You could also switch back to the test queue instead of the short-serial-4hr queue, since that worked for me earlier today.
Patrick

comment:25 Changed 4 months ago by aejigu

Hi Patrick.
Thanks

I tried with both test and short-serial-4hrs but still I get the issue, jules fails after submission. In the previous ticket you recommended adding NETCDF libraries: You also need to have the NETCDF libraries defined in your file u-an231_slurm/include/jasmin/suite.rc:

fcm_make?

inherit = None, linux
[job?]

batch system = background
execution time limit = PT10M

[directives?]

—time = 00:30
-n = 1

[environment?]

JULES_BUILD = normal
JULES_OMP = noomp
JULES_SOURCE = $JULES_FCM$AT_JULES_REVISION
JULES_PLATFORM = jasmin-lotus-intel
NETCDF_FORTRAN_ROOT=/home/users/siwilson/netcdf.openmpi/
NETCDF_ROOT=/home/users/siwilson/netcdf.openmpi/

But where is the /include/jasmin/suite.rc? Or, is it the suite.rc in roses, or cylc-run? Sorry, it is newbies question.

Thanks,
Amsale

Last edited 4 months ago by aejigu (previous) (diff)

comment:26 Changed 4 months ago by pmcguire

Hi Amsale:
The user for the other ticket is using a different suite that has a different structure, so that file include/jasmin/suite.rc doesn't exist for you. You don't need to add it. But there might be a few things in that file that you need to do in your site/suite.rc.CEDA_JASMIN file, for example.
Patrick

comment:27 Changed 4 months ago by aejigu

Hi Patrick,

I am wondering why the exact copy is running on your account but not on mine. Which bits in the mentioned ticket do you think I need to add?

Many thanks,
Amsale

comment:28 Changed 4 months ago by pmcguire

Hi Amsale:
It does indeed look like your ~amsale/roses/u-al752 suite is more or less identical to my modified version of your suite: ~pmcguire/roses/u-al752amsale1

One problem with that version of the suite is that it uses the test partition/queue for running the code, so it only can run 8 separate FLUXNET sites at the same time, so it might take a while to run.

You might consider changing to the short-serial partition/queue for running the code instead, so you can run more FLUXNET sites at the same time. But to do that, you will also need to make sure that you have the ivybridge128G intel processor type in the short-serial queue, and maybe compile with the same short-serial queue processor type at the same time.

I also looked at your error file:
~amsale/cylc-run/u-al752/log/job/1/jules_at_neu_presc0/01/job.err
It has the error:
[FATAL ERROR] file_ts_open: Given file '/gws/nopw/j04/jules/pmcguire/fluxnet/kwilliam/suite_data/vn1.3/fluxnet/AT_Neu-met.dat' does not exist

You can compare to my error file:
~pmcguire/cylc-run/u-al752amsale1/log/job/1/jules_at_neu_presc0/01/job.err
and my output file:
~pmcguire/cylc-run/u-al752amsale1/log/job/1/jules_at_neu_presc0/01/job.out

I just checked with:ls -lt /gws/nopw/j04/jules/pmcguire/fluxnet/kwilliam/suite_data/vn1.3/fluxnet/AT_Neu-met.dat
and that file does indeed exist.

Have you applied for and have you received access to the jules GWS? You can apply at the JASMIN accounts portal if you haven't applied for it yet. If you haven't received access yet, maybe you an contact the JASMIN Helpdesk available by email at support AT jasmin.ac.uk . That is one of the first steps in this tutorial. Once you have access, you should then be able to view files in the jules group workspace (GWS) in /gws/nopw/j04/jules .

Does that help?
Patrick

comment:29 follow-up: Changed 4 months ago by pmcguire

Hi Amsale:
Is everything working now?
Patrick

comment:30 in reply to: ↑ 29 Changed 4 months ago by aejigu

Replying to pmcguire:

Hi Patrick,

I realized that I needed to request access for the jules GWS following your comment and requested. I am waiting for the grant of access, will let you know the progress.

Many thanks,
Amsale


Hi Amsale:
Is everything working now?
Patrick

comment:31 Changed 4 months ago by pmcguire

Hi Amsale:
Thanks. If it takes too long for getting jules GWS access, you might contact support AT jasmin.ac.uk .
Patrick

comment:32 Changed 4 months ago by aejigu

Hi Patrick,

Now finally I am granted the access for jules gws and the jules runs successfully. However, make_plots failed with the job.err message: ImportError?: No module named matplotlib for /home/users/amsale/cylc-run/u-al752/bin/fluxnet_evaluation.py. Where do I need to module load?? In the tutorial #13 you mentioned "You might now be able to study (with ncinfo or python2.7, etc.; if you use python2.7, you might need to use sci1.jasmin or sci2.jasmin instead of cylc1.jasmin in order to get the proper python libraries working) the NETCDF output files in where you set them to be in your rose-suite.conf file …" But the plots are not been created.

Also, can I run the make_plots separately since the fcm_make and jules are successfully completed?

Thanks,
Amsale

comment:33 Changed 4 months ago by pmcguire

Hi Amsale:

I am glad that you now have jules GWS access and that JULES now runs on SLURM on LOTUS on JASMIN for you. I will update the documentation to put an extra check that a user actually has jules GWS access before they can proceed with the rest of the tutorial.

The JASMIN folks recently changed (November 3) the default jaspy module and Python version. Previously, the default jaspy module was jaspy/2.7 but now the default jaspy module is jaspy/3.7, which uses Python 3.7 (see https://help.jasmin.ac.uk/article/4729-jaspy-envs ). The FLUXNET/JULES u-al752 suite currently uses Python 2.7 for plotting, so we need to revert the u-al752 suite to jaspy/2.7.

One way to do this is to change the [[PLOTTING_CEDA_JASMIN]] section in ~/roses/u-al752/site/suite.rc.CEDA_JASMIN so that it now explicitly loads the jaspy/2.7 module with an extra env-script statement. It was relying on the pre-loaded jaspy module before:

    [[PLOTTING_CEDA_JASMIN]]
        inherit = None, JASMIN_LOTUS
        env-script = """
                 eval $(rose task-env)
                 export PATH=/apps/jasmin/metomi/bin:$PATH
                 module load jaspy/2.7
                 module list 2>&1
                 env | grep LD_LIBRARY_PATH
                 """
#        [[[remote]]]
#            host = sci3
        [[[directives]]]
            --time = 08:00:00
            --ntasks = 1

One you make and save those changes (with vi or emacs, etc.), then you can reload the suite's settings by typing rose suite-run --reload when you are in the ~/roses/u-al752 directory. You can have the cylc Xwindows GUI open beforehand, or you can open it afterwards with rose sgc. Since the u-al752 suite has already been partially run already, you can then retrigger the make_plots app in the cylc GUI by right clicking that make_plots app and selecting Trigger (run now). You should then see that the app gets submitted and that it starts running.

If you want to use python2.7 interactively on sci1.jasmin or sci2.jasmin to study the NETCDF files directly, you will need to type module load jaspy/2.7 at the command line of sci1.jasmin or sci2.jasmin before starting python2.7. Alternatively, you can use the default Python (v3.7) with iris by typing module load jaspy, followed by python, followed by import iris. But this doesn't help directly for the SLURM/LOTUS batch nodes.

Does this help?
Patrick

comment:34 Changed 4 months ago by aejigu

Hi Patrick,

Thanks for the follow up and all helpful replies.

Yes- it will save some time if you mention, specially for new jasmin users, the gws needed to get the jules tutorial running.

I implemented your suggestions yesterday but it is still on 'submitted' status. Hopefully it will work, I will keep you updated!

Best regards,
Amsale

comment:35 Changed 4 months ago by pmcguire

HI Amsale:

I am glad you agree that the tutorial needs to emphasize better that the jules GWS is needed. That was already written in the tutorial, but I will emphasize it even more.
I haven't changed the tutorial yet, since for some reason editing is not permitted at the moment.

You can switch from the short-serial queue to the short-serial-4hr queue in the suite (and reduce the plot running time from 8 hours to 4 hours), and reload the suite, and then retrigger the plotting app. And maybe then you don't need to wait so long in the queue for it to got from being just submitted to get it to start running.

Patrick

comment:36 Changed 3 months ago by grenville

  • Status changed from new to pending
Note: See TracTickets for help on using tickets.