Opened 2 years ago

Closed 2 years ago

#2285 closed help (fixed)

Cannot submit job, error: illegal cycle point offset on the right

Reported by: shakka Owned by: um_support
Component: UM Model Keywords: cycle, submission
Cc: Platform: Monsoon2
UM Version: 10.4

Description

Hi CMS,

I am trying to run a new job and keep getting the following error. Are you able to advise?

Thanks,

Ella

[FAIL] cylc validate -v -o /var/tmp/tmp348Amd --strict u-ai781 # return-code=1, stderr=
[FAIL] WARNING: deprecated items were automatically upgraded in 'suite definition':
[FAIL]  * (6.11.0) [cylc][event hooks] -> [cylc][events] - value unchanged
[FAIL]  * (6.11.0) [runtime][root][event hooks] -> [runtime][root][events] - value unchanged
[FAIL]  * (6.11.0) [runtime][HOST_LOCAL][job submission] -> [runtime][HOST_LOCAL][job] - value unchanged
[FAIL]  * (6.11.0) [runtime][HOST_HPC][job submission] -> [runtime][HOST_HPC][job] - value unchanged
[FAIL]  * (6.11.0) [runtime][HOST_IDL][job submission] -> [runtime][HOST_IDL][job] - value unchanged
[FAIL]  * (6.11.0) [runtime][HOST_LOCAL][job][method] -> [runtime][HOST_LOCAL][job][batch system] - value unchanged
[FAIL]  * (6.11.0) [runtime][HOST_HPC][job][method] -> [runtime][HOST_HPC][job][batch system] - value unchanged
[FAIL]  * (6.11.0) [runtime][HOST_IDL][job][method] -> [runtime][HOST_IDL][job][batch system] - value unchanged
[FAIL]  * (6.11.0) [runtime][ARCHIVE][retry delays] -> [runtime][ARCHIVE][job][execution retry delays] - value unchanged
[FAIL] ERROR, illegal cycle point offset on the right: Peninsula_km1p5_Smith_um_recon  =>  install_glm_startdata[-PT12H]

Change History (11)

comment:1 Changed 2 years ago by shakka

Hello,

Any chance anyone could help we with this please?

Ella

comment:2 Changed 2 years ago by grenville

Ella

Has suite ever run successfully? Did you change something after Sep 13th?

Grenville

comment:3 Changed 2 years ago by shakka

Hi Grenville,

Thanks for getting back so quickly.

Yes this is the suite I have been using for previous work. I have changed the following:

  • updated my charging code from cascade to polar
  • updated the location of my startdumps accordingly
  • changed the cycle end and start points for my new case study
  • changed the string formatting of the glm startdumps to match the 2011 startdumps I have to projects/polar/elgil/startdumps/YYYYmmdd_qwqgHH.T+0
  • added a new output stream with different STASH requests

Can you see any problems there?

Ella

comment:4 Changed 2 years ago by grenville

Hi Ella

I can't see why these changes would cause any problem - did you commit the working version? Can you revert to a working version so we can try to isolate where the error has crept in?

Grenville

comment:5 Changed 2 years ago by grenville

Ella

In the logs from Sept 13 and in suite.rc.processed from Sept 13, there is no reference to Peninsula_km1p5_Smith_um_recon — when was Smith introduced?

Grenville

comment:6 Changed 2 years ago by shakka

Ah yes, I also renamed the run Smith last week in anticipation of changing the cloud microphysics options. That couldn't be it, could it? All the 'smoothed' and 'ctrl' runs successfully completed. On your other question, I didn't commit the working version, but I have a record of the previous set-up, so it wouldn't take long to change back.

comment:7 Changed 2 years ago by grenville

Ella

I doubt Smith caused the problem — what is the most recent successful run - ie which zipped log file?

Don't change it all back in this suite —- make a copy first

Grenville

comment:8 Changed 2 years ago by shakka

Hi Grenville,

I've made a new copy. The run on the 8th September definitely worked (log.20170908T103402Z.tar.gz). Shall I see if reverting any of those changes has any effect?

Ella

comment:9 Changed 2 years ago by ros

Hi Ella,

We've found the problem. In suite conf → Nested Region 1 the rg01_runlen needs to be an integer multiple of CRUN_LEN (suite conf → General run options) otherwise the cylc graph is incompletely written. See also the help panel for `rg01_runlen'.

Would be useful if the suite did a check on this as the error message is less than helpful!

Cheers,
Ros.

Last edited 2 years ago by ros (previous) (diff)

comment:10 Changed 2 years ago by shakka

Hi Ros and Grenville,

Thanks for your help, that makes sense. The suite has successfully submitted. I think I misread the help box as meaning CRUN_LEN needed to be a multiple of rg01_runlen, not the other way around. Doh!

Thanks again for all your help.

Ella

comment:11 Changed 2 years ago by shakka

  • Resolution set to fixed
  • Status changed from new to closed
Note: See TracTickets for help on using tickets.