mailq|fgrep .science|sed 's/\*.*//'|postsuper -d -
mailq|fgrep .work|sed 's/\*.*//'|postsuper -d -
mailq|fgrep .link|sed 's/\*.*//'|postsuper -d -
mailq|fgrep .club|sed 's/\*.*//'postsuper -d -
mailq|fgrep .ninja|sed 's/\*.*//'|postsuper -d -
postsuper -d ALL deferred
... and any other domains / email addresses after the "fgrep" that look suspicious. pflogsumm is good for getting metrics on repeat offenders. Its tricky to avoid this happening in the first place.
Other References (some outdated)
There is a real issue with the output of the mailq / "postqueue -p" output, in terms of making something usable to check against. Here's a modified example from my logs, regarding a spam message that's failing to be delivered: each entry in the outputted text file has a blank line after; the Perl scripts floating around out there try to accommodate this, but poorly. A Python/Ruby script might work better for this....
91ADA1DDAFC* 12156 Wed Jun 17 13:14:32 email@example.com
There's an awesome RHEL / CentOS repo maintained with current Postfix builds. Was able to update 2.3 to 2.11 without immediately borking config files!
You can define a PCRE whitelist/blacklist of domains and addresses, and refer to it in main.cf. You don't have to run "postmap" on this after updating it either.
Sample entries to add....