mesoscale - det: Address Bugzilla 1547 - MPMD processes share the same working directory

[malloryprow]

From NCO's Bugzilla 1547

Multiple jobs use shared working directory for MPMD. Please modify the scripts to have isolated working directory for each process.

In the case of extreme IO, please work with NCO/GDIT for the best approach.

More information can be found here.

Suspected mesoscale - det Jobs Using MPMD

• plots/mesoscale/jevs_mesoscale_grid2obs_plots.sh
• plots/mesoscale/jevs_mesoscale_headline_plots.sh
• plots/mesoscale/jevs_mesoscale_precip_plots.sh
• plots/mesoscale/jevs_mesoscale_snowfall_plots.sh
• stats/mesoscale/jevs_mesoscale_nam_grid2obs_stats.sh
• stats/mesoscale/jevs_mesoscale_nam_precip_stats.sh
• stats/mesoscale/jevs_mesoscale_nam_snowfall_stats.sh
• stats/mesoscale/jevs_mesoscale_rap_grid2obs_stats.sh
• stats/mesoscale/jevs_mesoscale_rap_precip_stats.sh
• stats/mesoscale/jevs_mesoscale_rap_snowfall_stats.sh

[PerryShafran-NOAA]

I hope @MarcelCaron-NOAA can help me look into this. As I mentioned during yesterday's meeting, I thought that mesoscale (and cam) det were already compliant because of the many subdirectories I see in the tmp directory, but Marcel thought that they were not compliant.

@malloryprow If I recall you did the precip and snowfall for mesoscale, and so thus I'm guessing that they are not compliant. Hopefully you'll be able to help here, as I'm sure that it looks like what you did for global_det. Mallory - did you only do the precip/snowfall stats, or both stats and plots?

[malloryprow]

Just stats. I can't promise I'll be able to help as I need to do this for global_det as well as PRs. We have time though. We will have to see.

[PerryShafran-NOAA]

I probably wouldn't need much time, if I can easily mimic what you did for global_det. Just to ask questions. I'm going to get the restart finished (that's been a chore) first, then work on the beta5 stuff (probably with Marcel), and then do the bugzilla stuff. I think I'll likely do a different PR for each one.

[malloryprow]

Yes please do all 3 of those in separate PRs.

[MarcelCaron-NOAA]

@PerryShafran-NOAA I'm not sure how simple this bugzilla fix will be for cam, but it should be transferable to mesoscale grid2obs stats and mesoscale plots. I'll be in touch with you about it.