Suite not creating astart file

Unified Model
1

Dear CMS,

I’m trying to run suite u-dq291, which is a UKESM1.1 AMIP suite and a copy of u-dp788. Both suites are identical, using mid-Holocene orbital and GHG forcing, except for the surface fraction ancillary:

u-dp788 → Surface fractions from a piControl experiment
u-dq291 → Idealised vegetation prescribing grasses and trees over the Sahara Desert

Both suites have l_triffid = true but l_veg_compete = false so that the surface fractions are held fixed.

u-dp788 is currently running with no problems, but u-dq291 fails on the first timestep with the error:

[1] ?  Error code: 3
[1] ?  Error from routine: GLUE_CONV_6A
[1] ?  Error message: Mid conv went to the top of the model at point           14 in seg on call  1
[1] ?  Error from processor: 504
[1] ?  Error number: 37

I’ve also noticed that cylc-run/u-dp788/share/data/ and cylc-run/u-dq291/share/data appear different:

ajw1g19@ln01:~> ls cylc-run/u-dp788/share/data/
total 3.3G
drwxr-sr-x 4 ajw1g19 4.0K May 14 16:46 .
drwxr-sr-x 7 ajw1g19 4.0K May 16 10:52 ..
lrwxrwxrwx 1 ajw1g19   59 May 13 21:40 dp788a.ainitial -> /work/n02/n02/ajw1g19/um/restart_dumps/dd727a.da20500101_00
-rw-r--r-- 1 ajw1g19 3.3G May 28 15:25 dp788a.astart
-rw-r--r-- 1 ajw1g19  534 May 28 21:36 dp788_wallclock.list
drwxr-sr-x 4 ajw1g19 4.0K May  9 15:47 etc
drwxr-sr-x 4 ajw1g19 4.0K May 28 21:40 History_Data

ajw1g19@ln01:~> ls cylc-run/u-dq291/share/data/
total 16K
drwxr-sr-x 4 ajw1g19 4.0K May 28 15:26 .
drwxr-sr-x 7 ajw1g19 4.0K May 28 15:13 ..
lrwxrwxrwx 1 ajw1g19   59 May 28 15:13 dq291a.ainitial -> /work/n02/n02/ajw1g19/um/restart_dumps/dd727a.da20500101_00
lrwxrwxrwx 1 ajw1g19   59 May 28 12:08 dq291a.astart -> /work/n02/n02/ajw1g19/um/restart_dumps/dd727a.da20500101_00
drwxr-sr-x 4 ajw1g19 4.0K May 28 12:13 etc
drwxr-sr-x 3 ajw1g19 4.0K May 28 17:24 History_Data

I’d like to know, firstly, why the astart file for u-dq291 is not being created properly, and secondly, whether that has any connection to the error causing the model to fail.

If it’s not connected, then any advice on the error would also be appreciated.

Looking forward to hearing from you.

Alfred

2

Alfred

I don’t know why dq291a.astart -> /work/n02/n02/ajw1g19/um/restart_dumps/dd727a.da20500101_00 - I can only guess that you made the link some point?
The reconfiguration takes dq291a.ainitial and reconfigures it to dq291a.astart, which in this case are the same file. I don’t know what state /work/n02/n02/ajw1g19/um/restart_dumps/dd727a.da20500101_00 will be in now.
I’d delete the link and start again.

Grenville

3

Hi Alfred,

FYI re the symlink: If this suite is run without reconfiguration there is an optional override file (app/install_cold/opt/rose-app-norecon.conf) that gets used which puts the symlink from astart → ainitial. If you then subsequently switch on reconfiguration and don’t do a clean run, then the system doesn’t remove the old symlink and the reconfiguration doesn’t squawk about the symlink being there.

Cheers,
Ros.

4

Hi Both,

Thank you! Removing the symlink and starting again fixed the astart issue. Unfortunately the model is still crashing on timestep 1 with the convection error I copied in above. I assume it’s connected to the surface fractions and my having l_triffid=true and l_veg_compete=false.

I don’t suppose you have encountered this error before or have an idea as to what might be causing it?

Thanks again,
Alfred