A system at an office has its testnet node stuck at a particular block.
The logs report "requesting block" over and over again for the same block and after each one reports "socket recv error" and that peer disconnects. Then it connects a new peer and the same thing happens. There are no log entries about INVALID blocks.
As a result the system hasn't had a new block in two days, but all the explorers report that the chain has continued to move advance.
The current block this node is on is a member of the best chain reported by the explorers, but thousands behind.
This node is running on Ubuntu in the US. The office network has a sonicwall firewall.