Handle receiving Discovery packets with changed endpoint by gumb0 · Pull Request #5519 · ethereum/aleth

gumb0 · 2019-03-13T15:01:20Z

codecov-io · 2019-03-14T16:09:17Z

Codecov Report

Merging #5519 into master will increase coverage by 0.08%.
The diff coverage is 95.29%.

@@            Coverage Diff             @@
##           master    #5519      +/-   ##
==========================================
+ Coverage    61.8%   61.89%   +0.08%     
==========================================
  Files         344      344              
  Lines       28694    28753      +59     
  Branches     3261     3264       +3     
==========================================
+ Hits        17735    17797      +62     
+ Misses       9797     9789       -8     
- Partials     1162     1167       +5

gumb0 · 2019-03-18T15:57:40Z

libp2p/NodeTable.cpp


    LOG(m_logger) << "Active node " << *_nodeEntry;
-    // TODO: don't activate in case endpoint has changed
-    _nodeEntry->endpoint.setAddress(_endpoint.address());


This part is basically moved to Pong handler (lines 513-516), because we should trust changed endpoint only after valid Pong

halfalicious

Some minor suggestions and a few questions

halfalicious · 2019-03-19T15:58:36Z

libp2p/NodeTable.cpp

                // create or update nodeEntry with new Pong received time
                DEV_GUARDED(x_nodes)
                {
+                    auto const& sourceId = pong.sourceid;


We already get the sourceId from the pong packet:

aleth/libp2p/NodeTable.cpp

Lines 489 to 490 in 0886a34

// in case the node answers with new NodeID, drop the record with the old NodeID

auto const& sourceId = pong.sourceid;

halfalicious · 2019-03-20T03:19:33Z

libp2p/NodeTable.cpp

                        sourceNodeEntry = it->second;
                        sourceNodeEntry->lastPongReceivedTime =
                            RLPXDatagramFace::secondsSinceEpoch();
+


I know this is good practice but do you know if we ever actually hit this case? It seems like we wouldn't given that pongs have to be received within 1 minute of the ping being sent to be considered valid?

Not sure how it is related to Pong timeout, this is the case when there is a record in the bucket of the node table (maybe added long ago), but we receive a valid Pong with the same NodeID but from another endpoint.

halfalicious · 2019-03-20T03:20:08Z

libp2p/NodeTable.cpp

                }
+                if (sourceNodeEntry->endpoint != _from)
+                {
+                    LOG(m_logger) << "Neighbours packet from unexpected endpoint.";


Should we log both endpoints here?

halfalicious · 2019-03-20T03:20:14Z

libp2p/NodeTable.cpp

                }
+                if (sourceNodeEntry->endpoint != _from)
+                {
+                    LOG(m_logger) << "FindNode packet from unexpected endpoint.";


Should we log both endpoints here?

test/unittests/libp2p/net.cpp

halfalicious · 2019-03-20T03:40:24Z

test/unittests/libp2p/net.cpp

+    auto findNode = dynamic_cast<FindNode const&>(*findNodeDatagram);
+
+    // send Neighbours through endpoint 2
+    // TODO fill nearest with one node


TODO? I think you already do this?

halfalicious · 2019-03-20T03:42:13Z

test/unittests/libp2p/net.cpp

+    NodeIPEndpoint neighbourEndpoint{boost::asio::ip::address::from_string("200.200.200.200"),
+        c_defaultListenPort, c_defaultListenPort};
+    vector<shared_ptr<NodeEntry>> nearest{make_shared<NodeEntry>(nodeTable->m_hostNodeID,
+        KeyPair::create().pub(), neighbourEndpoint, RLPXDatagramFace::secondsSinceEpoch(), 0)};


Would be nice to comment the 0, especially since there are a lot of ctor args

halfalicious · 2019-03-20T03:42:22Z

test/unittests/libp2p/net.cpp

+        c_defaultListenPort, c_defaultListenPort};
+    vector<shared_ptr<NodeEntry>> nearest{make_shared<NodeEntry>(nodeTable->m_hostNodeID,
+        KeyPair::create().pub(), neighbourEndpoint, RLPXDatagramFace::secondsSinceEpoch(), 0)};
+    Neighbours neighbours(nodeTable->m_hostNodeEndpoint, nearest);


(Nit) {} initializers

halfalicious · 2019-03-20T03:42:40Z

test/unittests/libp2p/net.cpp

+    nodeSocketHost2.socket->send(neighbours);
+
+    // Wait for Neighbours to be received
+    nodeTable->packetsReceived.pop(chrono::seconds(5));


Should we validate the received packet?

gumb0 · 2019-03-25T13:59:12Z

Addresses review issues, rebased and added CHANGELOG entry.

gumb0 added the in progress label Mar 13, 2019

gumb0 force-pushed the changed-endpoint branch from f246aef to e757bbb Compare March 14, 2019 16:09

gumb0 force-pushed the changed-endpoint branch 3 times, most recently from 7a9d46b to 0886a34 Compare March 18, 2019 15:47

gumb0 commented Mar 18, 2019

View reviewed changes

gumb0 requested review from chfast and halfalicious March 18, 2019 15:58

gumb0 removed the in progress label Mar 18, 2019

halfalicious suggested changes Mar 20, 2019

View reviewed changes

gumb0 force-pushed the changed-endpoint branch from 7f5c400 to 8163800 Compare March 25, 2019 13:58

gumb0 requested a review from halfalicious March 25, 2019 13:59

chfast approved these changes Mar 25, 2019

View reviewed changes

halfalicious approved these changes Mar 26, 2019

View reviewed changes

gumb0 force-pushed the changed-endpoint branch from 8163800 to 2f589f7 Compare March 26, 2019 10:19

gumb0 added 5 commits March 26, 2019 11:32

m_sentPings maps pinged udp::endpoint to NodeValidation

9343c0b

Check whether node's endpoint changed when handling discovery packets

a1d3b8d

Minor improvements

5719e7d

Add changelog message

bca8cb6

Fix build after bad rebase

28984b4

gumb0 force-pushed the changed-endpoint branch from 2f589f7 to 28984b4 Compare March 26, 2019 10:48

gumb0 merged commit 9f2e689 into master Mar 26, 2019

gumb0 deleted the changed-endpoint branch March 26, 2019 11:41

	// in case the node answers with new NodeID, drop the record with the old NodeID
	auto const& sourceId = pong.sourceid;

Conversation

gumb0 commented Mar 13, 2019

Uh oh!

codecov-io commented Mar 14, 2019 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

gumb0 Mar 18, 2019 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

halfalicious left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

gumb0 commented Mar 25, 2019

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

codecov-io commented Mar 14, 2019 •

edited

Loading

gumb0 Mar 18, 2019 •

edited

Loading

halfalicious left a comment •

edited

Loading