qstat_alive.sh

#!/usr/bin/env bash
set -euo pipefail

CACHE_FILE="$HOME"/.cache/qstat.map

# a 1 minute timed cache
# does not actually handle failures properly. todo
if ! test -e $CACHE_FILE || (( $(find "$CACHE_FILE" -mmin +1 | wc -l) > 0)); then
    CACHE_FILE_TMP="$CACHE_FILE.$$.$RANDOM"
    trap 'rm '"$CACHE_FILE_TMP" EXIT
    qstat | awk '$3 ~ /cromwell/{print $1}' > "$CACHE_FILE_TMP"
    mv "$CACHE_FILE_TMP" "$CACHE_FILE"
    trap '' EXIT
fi

grep -q '^'"$1"'$' "$CACHE_FILE"

I just realised this wont deal with brand new jobs very well - need the cache duraction to be shorter than the check-alive interval. (and hope check-alive is called AFTER an interval, not straight after qsub)