I ran Zimbra for 6 months on a hosted VM then reinstalled/migrated to a bhyve Ubuntu VM on a bare metal FreeBSD server. The migration was due to the fact the original VM was in an IP range on a blacklist and I was going to have to move and reinstall anyway. So save some $$ and host it as a VM on the existing FreeBSD server.
I tried to migrate but zmconfigd and proxy failed to start. With a clean install it worked for awhile and I moved emails with imapsync but after making changes to proxy ports and http vs https vm both methods eventually was not able to restart all services with zmcontrol restart. After a week of trying I finally was backing up step by step and adding new domains and aliases using the web interface and the error cropped up again. I backtracked and removed the last alias added and it was able to restart again. I was not able to find any errors in the logs, the proxy log just didn't have any entries, not being written to and nginx failed to start unknown error. The error seemed very random, it seemed to be taking too long to start, and after adding more aliases it seemed to point to DNS.
I did serveral DNS lookups but if using 127.0.0.1 I got a slow response, localhost would hang till timeout, without specifying I would get a slow response.. I don't use dnsmasq much and followed a howto explaining to change the server-IP entry in /etc/dnsmasq.conf but upon detail reading this should be your upstream DNS server not your host server IP. Fixing this greatly improved all the DNS response times and nslookup mx.domain.com localhost now works.
Solution is if you are experiencing seemingly random proxy restart errors, check your DNS configuration.