Fullfilling 189

[jbusecke]

#189

[jbusecke]

Ahh brilliant! This confirms that the error was indeed only occuring with the GISS model. I am tracking that smaller issue here and will merge and run this one here once this action run is completed. @scoats please remind me if I should forget to deploy this.

[jbusecke]

Just changed the PR config and iid list to include one OOM failcase from #194. If that works, I am considering switching from Dataflow prime to fixed instances to unblock all these HighResMIP datasets.

[jbusecke]

Ok so it seems that the test job did not OOM, but ran into a test error. Ill leave the worker config this high for now, but maybe worth checking back on that in a bit? This run cost about 12 cents, so not too terrible. Lets run this bad boy and see which ones we might get.