ned14
commented
10 years ago I'd strongly recommend NOT to merge my branch into Maidsafe next. The code is neither safe nor stable. I only sent you this code for testnet purposes only, it is not fit to be used otherwise.
If you can merge just your improvements only to Maidsafe next for now, I can merge your changes to my experimental branch. I can then derive from that a new interim_merge_of_unsafe_packet_loss_tolerance branch which has SYN cookie disabled if you need such a thing.
The current showstopper for the SYN cookie implementation is that something is rarely opening a spurious second connection to the same endpoint which invokes the DDoS protection against one of those connections. If the DDoS excluded connection is the one actually doing the transfer, that kills the connection.
The source of this spurious second connection is ancient, and I believe is why your former SYN cookie protection was disabled because it similarly cut out connectivity.
maqi
https://maidsafe.atlassian.net/browse/MAID-22
this will aslo merge the work from Niall in the branch of interim_merge_of_unsafe_packet_loss_tolerance once bad_alloc branch got merged into next, interim_merge_of_unsafe_packet_loss_tolerance can also be considered as approved and merged
the verification involves using routing_node network across droplets