I’ve added some print statements and can see that the tracers aren’t being activated, but I can’t figure out why. (option codes are currently set to zero for debugging, but I get the same error even when I adjust the option codes).
Yeah they seem to be.
In the /cylc-run/u-dr880/run2/work/1/atmos_cristrat_rigorous_16x8/pe_output/atmos.fort6.pe000 file (for example), they are displayed with the other tracers in sections 34 when listing the prognostic fields:
Another long shot while I think about this further, if you did this all via the terminal, can you activate the GUI (if one exists in Monsoon3) and see if it spots any errors?
I have had the “Missing X of Y required tracers” before and it was because stuff wasn’t being picked up in the new stashmaster but your changes look fine.
If this is a test suite, are all the atoms tasks failing?
I’ve been using the cylc TUI on Monsoon3 to visualise what is failing and it seems to be all the atmos tasks (from looking at the job.err file of each atmos task).
I’ve also had this error previously when working with version 13.4, but it was an issue of a missing line in the rose-suite.conf file that was preventing the stashmaster file from being found. I’ve gone through and checked all files related to the stashmaster and can’t see anything wrong which is why I am a bit stumped
There shouldn’t be a UM vn13.4 suite in my Monsoon3 directory. I previously (on Monsoon2) implemented this scheme in vn13.4 and had it working using suites such as u-dd630 (d/d/6/3/0 (log) – roses-u).
The error I am getting now with the tracers is using a copy of the test suite at vn13.9 on Monsoon3 (u-dr880). I am trying to implement the same scheme that I put in UM vn13.4 to UM vn13.9.
The Stashmaster file for the 9 tracers has value “0” for all option codes - that appears to be the reason why the tracers are not on. Please look at the explanation in
I had set the stashcodes to zero for debugging, but I still get the error when I have them set to what I used in vn13.4 (see cylc-run/u-dr880/run4/log/job/1/atmos_cristrat_rigorous_16x8/01/job.err).
I’m not sure what else to test because in a side-by-side comparison my STASHmaster file looks the same as the STASHmaster in version 13.4 except that I have had to move the tracers up one so that they now start at 34271 instead of 34270.
Yes - that should be the up-to-date UM vn13.4 branch that successfully ran with the new tracers. It was run in many suites for different simulations, but an example suite that worked with the branch is u-dd630.
Thanks again for taking a look at this! Are there any other tests I could be running that would help diagnose the problem?
I’m obviously no UKCA expert, but notice that the list of the names of tracers in tracer_varnames (appearing in here for example tr_lookup(i) = arrayloc(nm_spec_active, tracer_varnames(i)) does not include BIN_1,… BIN_9.
