Can you please let me know what approach and steps would be in case the Splunk HEC implementation on indexers runs into some issues?
This also includes getting syslog data from
syslog-ng servers into indexers using HEC.
You can use useack functionality with HEC if you want.
See opensourced hec client
If all indexers are down or not available, the source should stop sending and either queue or propagate the issue down the chain (ie to stop eating new data).
For Syslog, Splunk Connect for Syslog has this functionality for example, see disk buffer here