This behavior has manifest on two new validators that were onboarded to TestNet this week, but not on the other validators on that network. TestNet is running indy-node=1.6.78.
valNode01 is a node that was once on the TestNet, but it has been removed with a ledger entry with services= written to the ledger. The issue is that in spite of this, this node is shown as if it is an active node that is unreachable in validator-info. In the logs, you can see that the new validators are attempting to contact it as well. This occurs on these 2 new validators, but not on the other ones in the pool. At least for these 2 validators, this is a drag on consensus, and could also influence client behavior as well.
The pool ledgers are in sync on all nodes.
validator-info output, the pool ledger, and logs from one of the new nodes are attached.