Evidenced by check_trafficserver_log_fifo_analytics_tls and check_trafficserver_log_fifo_notpurge_backend unknown status on May 30.
Logs from mtail are the same across the DC a couple days following the upgrade:
May 30 21:26:19 cp5011 atsmtail-tls: E0530 21:26:19.037737 46542 log_watcher.go:198] fsnotify error: fsnotify queue overflow
Issuing a restart renders many duplicates of this entry:
Jun 01 19:45:09 cp5011 atsmtail-tls: 2020/06/01 19:45:08 Unable to read from socket: dial unix /srv/trafficserver/tls/var/run/analytics.sock: connect: connection refused
It necessitates an additional restart to get metrics flowing again.