Aidan
(Aidan Heerdegen, ACCESS-NRI Release Team Lead)
22
It may well be the problem. In general I advise people to never load anything like this in .bashrc as this sourced every time a shell is started. This means it is difficult to provide an isolated environment with module purge etc and load just the modules a user needs because any new shell will reload .bashrc. Often a new shell is started for all sorts of innocuous tasks.
You can add this sort of autoload to .profile or .bash_profile as this is sourced only for login shells. Personally I don’t do this either. I prefer to make an alias that loads things I need a lot, and a single command can then set up your interactive shell environment as you like.
It might be!
Aidan suggestion is valid.
I load the conda/analysis3 too but in my .bash_profile and only in interactive shells, therefore it is not loaded automatically when the model runs.
But in your case it is.
I am still not sure how that could mess up with the environment (since you are loading the same environment anyway) but it might be worth trying to run the model again after having removed the module imports from your ~/.bashrc.
mmm okay good to know, thanks @Aidan. I tried to re-run it last night with those lines removed from my .bashrc script but it didn’t complete (didn’t get to the pre-processing step). I’ll wait until after the nci maintenance and try this again.
Rerunning the model again with the load conda/analysis3 commented out of my .bashrc file worked! I now have nicely collated output files in my scratch directory. Thanks @Aidan and @atteggiani!
2 Likes
Aidan
(Aidan Heerdegen, ACCESS-NRI Release Team Lead)
26