Opened 11 days ago

Last modified 4 days ago

#2757 new help

Atmosphere only job not running

Reported by: l.j.wilcox Owned by: um_support
Priority: normal Component: UM Model
Keywords: Cc:
Platform: UM Version: 8.5

Description

I'm trying to set up a new atmosphere only job with version8.5 and am having trouble getting it going. I've managed to run the model for a few months. I accidentally set my initial run as a CRUN rather than an NRUN, which I think is why it stopped after a few months. I've now changed the job to an NRUN and set it to start from one of the dumps it created. It looks like it does something, as the output files have been updated, but it fails after about 45 minutes. Are you able to see what I'm doing wrong?

The job is xohsb. The leave file is /home/n02/n02/laura/um/umui_out/xohsb000.xohsb.d19036.t101458.leave

Thank you!

Change History (4)

comment:1 Changed 11 days ago by grenville

Laura

See /work/n02/n02/laura/um/dataw/xohsb/pe_output/xohsb.fort6.pe115 for example:

??!!!???!!!???!!!???!!!???!!!???!!! ERROR ???!!!???!!!???!!!???!!!???!!!???!!!?
? Error in routine: glue_conv_5a
? Error Code: 3
? Error Message: Mid conv went to the top of the model at point 45 in seg on call 1
? Error generated from processor: 115
? This run generated 19 warnings
????????????????????????????????????????????????????????????????????????????????

The model has failed.

Grenville

comment:2 Changed 11 days ago by l.j.wilcox

Thanks Grenville. That doesn't sound good…

There's loads of those .pe files. How did you know to look in pe115 for the error message?

I've not run the model like this before. What's the best approach to dealing with this kind of problem so early in a simulation? Should I do a reconfiguration run and cross my fingers? Try starting again with a completely different dump? Worry about my ancillary files…?

Thanks!
Laura

comment:3 Changed 11 days ago by grenville

Hi Laura

Bit of hacking about (some of those files have different sizes too). It may be worth shortening the time step, or perturbing the dump, or using a different one. If that doesn't help, it'd be worth writing out some dumps leading up to the failure - they may point to a problem.

Grenville

comment:4 Changed 4 days ago by grenville

Laura

Has anything worked?

Grenville

Note: See TracTickets for help on using tickets.