Found a problem with logrotation, decribed in this article.
I started getting daily messages:
error: error renaming temp state file /var/lib/logrotate/status.tmp
I assumed this was caused by having both the original pihole logrotate and the new custom logrotate run @ midnight (00:00).
I modified the new custom logrotate cron job:
0 0 * * * root /usr/sbin/logrotate --force --state /var/log/pihole/status /home/pi/custom-pihole/logrotate
@reboot root /usr/sbin/logrotate --state /var/log/pihole/status /home/pi/custom-pihole/logrotate
This will create a separate state file /var/log/pihole/status.
The error disappeared (tested several days).
It should also be noted the cron @reboot job is useless when using TMPFS, since there will be no logfiles in that location after a reboot.