The script analyze_new_cluster.sh output by pg_upgrade contains several
"sleep" calls (see contrib/pg_upgrade/check.c). What is the point of
this? If the purpose of this script is to get the database operational
again as soon as possible, waiting a few seconds doing nothing surely
isn't helping.
I could maybe see the point of waiting a bit between the different
vacuumdb calls, to catch some breath, but the one before the first call
to vacuumdb is highly dubious to me.