tcp accept queue overflow/drops + tcp syn queue cookies full

[vancanneyt]

Since I have netdata installed, every time I have a traffic spike on the server, due to a site sending out +50k of push notifications, netdata sends some e-mails with the following contents:
1m tcp syn queue cookies = 452.6 cookies
1m tcp syn queue cookies (was critical for 1 minute and 21 seconds)
1m tcp accept queue overflows = 101.9 overflows
1m tcp accept queue overflows (was critical for 1 minute and 39 seconds)
1m tcp accept queue drops = 101.9 drops
1m tcp accept queue drops (was critical for 1 minute and 39 seconds)

I already used google to get more help before I post this topic and did some modifications to /etc/sysctl.conf as documented on https://help.directadmin.com/item.php?id=495 and https://easyengine.io/tutorials/linux/sysctl-conf/ and loaded those changes to the kernel. But still after implementing those modifications, these mails from netdata monitoring keeps coming. During those events, the server was still responsive, no high load whatsoever. (OS=CentOS 8)
So I was wondering what can I do next to further improve this and prevent that connections might be dropped?