HEC tuning for Splunk Cloud

[cwsandcreek-oss]

Hi there. We have an SC4S instance that shows drops when running syslog-ng-ctl stats and looking at the HEC destination. If I run an "ss" on the host and look at the HEC outbound connections, there only appear to be 2 or 3 that are actually showing any buffer size to them regardless of the number of workers we try. Is there a way to tune this or are there any guidelines to troubleshooting dropped HEC events? Thanks