In the past 24 hours, I've fixed two extremely hairy problems with git annex get -J. One was a locking problem. And the other involved thundering herds and ssh connection multiplexing and inherited file descriptors and races, and ... That took 4 hours of investigation to understand well enough to fix it.

Neither of those involved the ssh P2P changes, other than perhaps they exposed one of the issues more than it was exposed before, but on the plus side I've been testing that new code quite a lot as I worked on them..

Today's work was supported by the NSF-funded DataLad project.