With the slurm version on the new cluster you have to explicitly requ…

…est email notification for failed runs. Providing the additional argument "FAIL" also works on the old cluster.

scripts/start/submit.R

@@ -132,7 +132,7 @@ submit <- function(cfg, restart = FALSE, stopOnFolderCreateError = TRUE) {
     exitCode <- system(paste0("sbatch --job-name=",
                               cfg$title,
                               " --output=log.txt --open-mode=append", # append for requeued jobs
-                              " --mail-type=END",
+                              " --mail-type=END,FAIL",
                               " --comment=REMIND",
                               " --wrap=\"Rscript prepareAndRun.R \" ",
                               cfg$slurmConfig))