How important are the TCPOutAutoLB Health Warnings?

[Glasses2]

Hi,

I have seen yellow and red health warnings for TCPOutAutoLB-0 for sometime. 

We identified a few issues like a HF that was sending loads of data and locking to 1 single Indexer for a long time.    But after some HF config changes (i.e. parallel pipelines and asynchronous fwding) the warning frequency has been reduced, but I still see these yellow warnings randomly. 

How worried should you be if you see yellow or red health warnings like this >

TCPOutAutoLB-0 with RootCause  "more than <20% or 50%> of forwarding destinations have failed.." 

These warnings eventually they clear.   

When I check indexer saturation a few will be a 100% but I cannot tell for how long.  The 100% saturation seems to be very short lived.

Do these alerts fire based on count alone or does it have a saturation time check as well, like the 100% saturation state needs to be sustained for more than 30 seconds?

Are these typical warnings to just live with or is it something you never want to see?  or when is it a problem?

Is there a setting to increase the indexer HW use to process the events faster and clear blocks?

Thank you