I just had this, my temp workaround to get the server back up ASAP was to disable all logging whilst I did some googling.
I’m on Oracle Linux 8 (Red Hat EL), example solution YMMV
# Set ownership to caddy user and group
sudo chown caddy:caddy /var/log/caddy
# Set appropriate permissions (755 for directory)
sudo chmod 755 /var/log/caddy
# Set correct SELinux context for logs
sudo semanage fcontext -a -t httpd_log_t "/var/log/caddy(/.*)?"
sudo restorecon -Rv /var/log/caddy