Hi all,
I am currently trying to set up a paleoclimate suite of the HadGEM3-GC5 model on MONSOON3. To save spin-up time, I attempted to run the GC5 suite using the restart dump from a GC3 paleo suite that has already completed a spin-up of more than a thousand years.
The model runs normally during the first month. However, the following error occurs at the end of the month:
5260 ???!!!???!!!???!!!???!!!???!!! ERROR ???!!!???!!!???!!!???!!!???!!!
5261 ? Error code: 24
5262 ? Error from routine: WRITHEAD
5263 ? Error message: WRITHEAD: Addressing conflict
5264 ? Error from processor: 103
5265 ? Error number: 16
5209 ???
I suspect this may be caused by a version mismatch between the restart dump and the model configuration, but I am not sure how to resolve it.
Any advice or suggestions would be greatly appreciated.
Best,
Zikun
Hi Zikun,
If there was discrepancy between fields in the GC3 dump and GC5 model settings the run would have failed in Recon or first atmos timestep.
More information on the error should be available in the ~/cylc-run/suite/runX/work/date-time/atmos (or atmos_main or coupled)/pe_output/atmos,fort6.pe000 file.
A link to the changes made to the workflow (from roses-u trac) will also help in detecting any issues.
Mohit
Hi Mohit,
Thank you very much for your reply, and apologies for my late response. I didn’t keep the full output, so it took me some time to revisit this point.
Below is the longer output around the ERROR message:
26275451 FIXED LENGTH HEADER
26275452 -------------------
26275453 Dump format version 20
26275454 UM Version No 1308
26275455 Atmospheric data
26275456 Charney-Phillips on radius levels
26275457 Over global domain
26275458 Instantaneous dump
26275459 Exp No =-32768 Run Id = 0
26275460 360-day calendar
26275461 V-AT-POLES/ENDGAME grid
26275462 Record: Year Month Day Hour Min Sec DayNo
26275463 Data time = 3201 1 1 0 0 0 1
26275464 Validity time = 3201 2 1 0 0 0 31
26275465 Creation time = 2026 3 10 17 47 30 -32768
26275466 Start 1st dim 2nd dim 1st parm 2nd parm
26275467 Integer Consts 257 46 46
26275468 Real Consts 303 38 38
26275469 Level Dep Consts 341 86 8 86 8
26275470 Row Dep Consts -32768 -32768 -32768 145 0
26275471 Column Dep Consts -32768 -32768 -32768 192 0
26275472 Fields of Consts 1029 12348 1 12348 1
26275473 Extra Consts -32768 -32768 0
26275474 History Block -32768 -32768 0
26275475 CFI No 1 -32768 -32768 0
26275476 CFI No 2 -32768 -32768 0
26275477 CFI No 3 -32768 -32768 0
26275478 Lookup Tables 25725 64 17081 64 17081
26275479 Model Data 2097153 448633944 7524988
26275480
26275481 LEVEL DEPENDENT CONSTANTS
26275482 688 64-bit words long
26275483
26275484 FIELD DEPENDENT CONSTANTS
26275485 12348 64-bit words long
26275486 ******FATAL ERROR WHEN READING MODEL DUMP******
26275487 Conflict between start position of lookup table
26275488 Block and Pointer in Fixed Length Header: fixhd(150) = 25725
26275489 Current position in file = 13377 words in
26275490 ***********************************************
26275491
26275492 ????????????????????????????????????????????????????????????????????????????????
26275493 ???!!!???!!!???!!!???!!!???!!! ERROR ???!!!???!!!???!!!???!!!???!!!
26275494 ? Error code: 24
26275495 ? Error from routine: WRITHEAD
26275496 ? Error message: WRITHEAD: Addressing conflict
26275497 ? Error from processor: 0
26275498 ? Error number: 16
26275499 ????????????????????????????????????????????????????????????????????????????????
From what I can see, this error seems to be caused by a mismatch between the lookup table pointers in the header and the actual dump file structure.
I have committed my modifications to Trac; the suite ID is u-dv769.
Thanks again for your help!
Best,
Zikun
Hi Zikun,
I think this is an issue we have seen before when using a restart from an old UM version. I can’t remember the details but there is some mismatch with the dump headers.
We have a branch:
fcm:um.x-br/dev/simonwilson/vn11.6_stochastic_header
You will need to upgrade it to the UM version your suite is using, then rerun the build and recon.
Annette
Hi Annette,
Many thanks!
I’ll have a go and get back to you with the result.
Best,
Zikun