NFD

Here's the relevant earlier discussion:

Eric Newberry wrote:

I ran a test, but it appears that onLostPacket is being called for every frame, even if it has been successfully acknowledged on its first transmission.

Klaus Schneider wrote:

What might cause this bug?

Eric Newberry wrote:

No idea. The timeout calling onLpLostPacket should be cancelled when its frame goes out of scope (when the UnackedFrag object containing it is deleted). Given that the initial RTO is 1 second and ndnping is reporting an RTT in the 10s of milliseconds on the link, I am at a loss.

Davide Pesavento wrote:

Klaus Schneider wrote:

Looks like we need a more formal integration test for this?

Not necessarily an integration test. A unit test should be able to catch this kind of bugs.

Can you give a specific, step-by-step instruction of your testing? (i.e. list the terminal commands and files that need editing).

Davide Pesavento wrote:

So there's no bug here, correct?

Not to my knowledge. Here's how I performed my testing (if my memory serves right):

1. Set up a VirtualBox VM with a host-only network adapter, allowing it to be reached over the network from the host machine. Install an unmodified version of NFD on the VM.
2. On the host machine, modify, compile, and install NFD to include a message to cout in BestRouteStrategy2::onDroppedInterest (creating this function in the process), as well as in LpReliability::onLpPacketLost to print before onDroppedInterest is signaled.
3. Start NFD both on the host machine and the VM.
4. Create a UDP face on the host machine, with the VM as the remote endpoint, enabling reliability: nfdc face create udp4://[remote-ip-addr] reliability on
5. Run the corresponding nfdc command on the VM to enable reliability from that end of the connection.
6. Use tc on the VM to simulate 10% packet loss on the host-only interface: sudo tc qdisc add dev enp3s0 root netem loss 10%
7. Start ndnpingserver on the VM to serve a particular prefix.
8. Start ndnping on the host machine for the prefix mentioned above.
9. Examine the output of NFD on the host machine to ensure that both messages are being printed to cout.

Great. I suggest rejecting this task unless someone can describe an integration test scenario that tests something not already covered by unit tests.

Davide Pesavento wrote:

Oh wait, this task is about unit tests, not integration tests. LpReliability already has numerous unit tests though, is there something that is not covered by them? Please explain in the description exactly what tests are needed, if any, otherwise reject.

We could add a unit test to ensure that the loss timeout is cancelled when an Ack in received for the packet. I already have this test case written, so I'll go ahead and push it for code review.

Eric Newberry wrote:

Davide Pesavento wrote:

Oh wait, this task is about unit tests, not integration tests. LpReliability already has numerous unit tests though, is there something that is not covered by them? Please explain in the description exactly what tests are needed, if any, otherwise reject.

We could add a unit test to ensure that the loss timeout is cancelled when an Ack in received for the packet. I already have this test case written, so I'll go ahead and push it for code review.

Sounds good.