I am working on suite u-db437 and I have switched to N768 res for the global model recently which is the driving model. Earlier I was using the N216 res for which the model ran successfully. Only after switching to N768, i am unable to get past the following error. I was hoping that there should never really be any issue with running the glm n768 with ERA5, and such errors pop up in the high-res inner nested domain.
Error code: 1
? Error from routine: EG_BICGSTAB
? Error message: NaNs in error term in BiCGstab after 1 iterations
? This is a common point for the model to fail if it
? has ingested or developed NaNs or infinities
? elsewhere in the code.
? See the following URL for more information:
? https://code.metoffice.gov.uk/trac/um/wiki/KnownUMFailurePoints
? Error from processor: 765
? Error number: 50
The error is in the global model - happening so soon usually indicates a problem with input data. You can see in /home/n02/n02/aurocumulus/cylc-run/u-db437/work/20190113T1200Z/glm_um_fcst_000/pe_output/umgla.fort6.pe0000 that there are NaNs and Infinities in the model.
It might be worth reducing the time step (that’ll be set in the n768 override file)
Thanks for the suggestion. I had turned on extra diagnostics to see which of the physics modules was running into trouble. The wind attributes (s_u, s_v, etc) were yielding erroneous values like NaNs etc which was causing the BICGSTAB. The ics file to the global model was also not opening properly meaning it didn’t do the recon properly. Could be a problem with the ancil files which are being ingested at the first time step. So I had a look at the glm_um ancils in the gui and found two ancils at N216 in place along with others (for a N768 run). So had to modify those two in a way the others were listed. And I have given it a go. Will let you know how I get on with this modification. Thanks anyway.
Best Sid
This works!!- So, if anyone is getting Bicgstab at the very first step in the global model (which is quite rare)- might want to check the ancils and their paths (in glm_um in the gui)
This is in continuation to the same run. The N768 glm_um_fcst worked successfully after the modifications (posted above). But now I am getting the following error at the inner high-res recon step.
Error code: 10
? Error from routine: READ_NML_ITEMS
? Error message: stash_req is missing or incompatible in items namelist.
? Please check entries.
? Error from processor: 0
? Error number: 2
I had a look at the rose-app-ra3.conf file (since I am running ra3 setting with casim) and I can see that there are a few ancillary items being called in the first few lines (with their source number and stash requests mentioned). I have a feeling that the inner domain recon is failing because of a few missing ancils which is causing the model to report an error related to missing STASHES. I am running the suite u-db437. I would be grateful for any help on this issue. How do I find which extra ancils are not present if this is the problem?
Best Sid
