I have made some minor code changes to my UKESM suite u-ce487. The atmos_main task crashes with the unrecoverable library error: ‘A WRITE operation tried to write a record that was too long’.
I don’t know if the error is caused by my code changes, which I’ve double-checked, my choice of diagnostics, or some problem related to the python scripts that perturb values of namelists in my suite.
Could you advise me how I might explore this further please?
Thanks for the suggestion Grenville. I changed the setting of ‘PRINT_STATUS’ in
um->Runtime_Controls->Atmosphere_Only
so that ‘Extra diagnostic messages’ is false.
But, the same error is repeated in my ‘u-ce487’ atmos_main log file. Any idea how I can explore this further?
I understood from your earlier suggestion I should set the diagnostic messages to FALSE instead of TRUE, hence the PRINT_STATUS was still set to PrStatus_Diag. But, I have now set the PRINT_STATUS to ‘Normal’ and set the ios_verbosity to 2 (Normal).
This will likely take a while to run, so hopefully the ticket isn’t automatically clsoed before I get results of this test.
My suite is running now, so reducing the verbosity of output and/or reducing the number of diagnostic messages seems to have done the trick. If you have time, it would be useful to have some insight into why this happens, since it is a counter-intuitive solution.
It’s trying to write a message but the message buffer it writes to is too small to hold it. The message maybe informative only (which seems to be the case since your model is running) - but extra diagnostics can be useful if it writes an error message short enough to fit in the buffer. I guess no one has found the time to fix the buffer problem.