I am attempting to run the JULES u-dz272 suite on JASMIN. It successfully ran for me a month ago, but it is now unable to find the lotus platform during job submission. See the error message below.
Image (a) - the output from “~/cylc-run/u-dz272/run1/log/job/1/fcm_make/01/job-activity.log”. No other files are produced in “~/cylc-run/u-dz272/run1/log/job/” folder.
My login is zaracra. I may not show up on a list associated with that suite as I still hadn’t sorted all the general run errors when it was previously ‘working’. However I was able to progress much further with the run than I am now able to.
emrobi is the suite owner who I initially worked with to troubleshoot this problem.
In emrobi’s files, the file “/home/users/emrobi/cylc-run/u-dz272/run1/log/job/1/fcm_make/01/job-activity.log” gave the output
I can’t see your files on there, although you can give people permission with:
chmod -R a+rX /home/users/zaracra
if you wish.
I think that emrobis suite ran since you were last successful. You can either checkout a new version and run that, or use the fcm diff command, or regular unix diff against emrobi’s directory.
There will be a minor change that either you or that user have made, and the repo comparison or regular diff -r to compare directories should show it. That is what I would do if I had permissions for your files.
From the output of cylc config I’m guessing you have a global.cylc config file in your ~/.cylc/flow/8 directory? If so move it out of the way and try resubmitting the workflow. Looks like it has overridden the central one entirely (which is where lotus is defined) rather than only the [install] section and thus you’ve lost the configuration for the lotus platform.
The output from cylc config should have a section like this in it defining what lotus is:
I’m waiting for the appropriate GWS accessv on JASMIN but I’ve been able to run the fcm_make task fine which uses lotus.
@zcra could you try logging out of the cylc server on JASMIN, logging back in again and then (as the very first thing you do) re-type cylc config and paste what you get, if anything?
Until this is fixed we won’t be able to proceed, as detailed above.
I’ve logged back into the cylc server (using ssh -A zaracra@login.jasmin.ac.uk followed by ssh -AX cylc2.jasmin.ac.uk), then typed “cylc config” and nothing was output. The prompt returned after a couple of seconds.
What do you get for which cylc and which python? For example, I get…
$ which cylc
/apps/jasmin/metomi/bin/cylc
$ which python
/usr/bin/python
It looks like you might have tried to install cylc and some other related tools via pip or conda at some point (see your ~/.local/bin/ directory) which may be an issue.
