Project

General

Profile

Actions

Bug #3976

closed

NLSR not converging due to NFD faces/create semantics change

Added by Ashlesh Gawande about 7 years ago. Updated almost 7 years ago.

Status:
Closed
Priority:
Urgent
Assignee:
-
Target version:
Start date:
02/22/2017
Due date:
% Done:

100%

Estimated time:

Description

NLSR currently creates adjacency by invoking NFD faces/create command and expecting 200 response.
Since #3232 https://gerrit.named-data.net/3012, NFD faces/create command will respond with code 409 if a face with same RemoteUri already exists, breaking the expectation of NLSR.
To solve this problem, NLSR should either use faces/query dataset to find existing face, or inspect the response body that comes with code 409 to determine whether the existing face matches the expectation.


Related issues 1 (0 open1 closed)

Related to NFD - Bug #3232: Inaccurate log message when changing FacePersistencyClosedYanbiao Li09/30/2015

Actions
Actions #1

Updated by Junxiao Shi about 7 years ago

  • Related to Bug #3232: Inaccurate log message when changing FacePersistency added
Actions #2

Updated by Junxiao Shi about 7 years ago

  • Subject changed from NLSR not converging with latest NFD to NLSR not converging due to NFD faces/create semantics change
  • Description updated (diff)
  • Priority changed from High to Urgent
Actions #3

Updated by Ashlesh Gawande about 7 years ago

If I do the following fix:

void
HelloProtocol::onRegistrationFailure(const ndn::nfd::ControlResponse& response,
                                     const ndn::Name& name, const ndn::time::milliseconds& timeout)
{
  _LOG_DEBUG(response.getText() << " (code: " << response.getCode() << ")");

  // Temporary fix until #2954
  if (response.getText() == "Face with remote URI already exists") {
    ndn::nfd::ControlParameters faceParams(response.getBody());
    _LOG_WARN(response.getText() << " (code: " << response.getCode() << ") " << faceParams.getFaceId());
    onRegistrationSuccess(faceParams, name, timeout);
    return;
  }

NLSR still does not converge.
In a two node topology, node b's NLSR sends HELLO interest to node a.
But NFD's best route strategy says no route:

1487969928.411582 DEBUG: [ContentStore] find /ndn/a-site/%C1.Router/cs/a/NLSR/INFO/%07%1E%08%03ndn%08%06b-site%08%08%C1.Router%08%02cs%08%01b L
1487969928.411616 DEBUG: [ContentStore]   no-match
1487969928.411621 DEBUG: [Forwarder] onContentStoreMiss interest=/ndn/a-site/%C1.Router/cs/a/NLSR/INFO/%07%1E%08%03ndn%08%06b-site%08%08%C1.Router%08%02cs%08%01b
1487969928.411636 TRACE: [Strategy] lookupFib noLinkObject found=/
1487969928.411668 DEBUG: [BestRouteStrategy2] /ndn/a-site/%C1.Router/cs/a/NLSR/INFO/%07%1E%08%03ndn%08%06b-site%08%08%C1.Router%08%02cs%08%01b?ndn.MustBeFresh=1&ndn.InterestLifetime=3000&ndn.Nonce=3969901325 from=257 noNextHop
1487969928.411683 DEBUG: [Forwarder] onOutgoingNack face=257 nack=/ndn/a-site/%C1.Router/cs/a/NLSR/INFO/%07%1E%08%03ndn%08%06b-site%08%08%C1.Router%08%02cs%08%01b~NoRoute OK
1487969928.411696 TRACE: [LinkService] [id=257,local=unix:///run/b.sock,remote=fd://21] sendNack

I think I am missing some step here?

Also face from a to b is persistent,
but face from b to a is on-demand.
(a starts before b, and its HELLO reaches b)

Actions #4

Updated by Ashlesh Gawande about 7 years ago

Oh I think I need to update the persistency:
https://redmine.named-data.net/issues/3232#note-1

Actions #5

Updated by Ashlesh Gawande about 7 years ago

Okay so this does not work because I am not actually calling registerPrefixInNfd.

Actions #6

Updated by Junxiao Shi about 7 years ago

Reply to note-3:

  • Never rely on (or even look at) response.getText(). The correct condition is response.getCode() == 409.
  • Code 409 from faces/create just means face already exists. Routing convergence depends on not only the existence of face, but also the existence of a route to reach the peer NLSR in NFD-RIB. You are getting Nack due the lack of a RIB route.
  • The correct place to apply a quick fix is in FaceController::createFaceInNfd. Instead of calling onFailure directly, when encountering code 409 and the conflicting face has the correct LocalUri and RemoteUri, treat it as success.
Actions #7

Updated by Ashlesh Gawande about 7 years ago

Thanks Junxiao, this is better!
https://gerrit.named-data.net/#/c/3724/

Also one question I have is, so for a two node topology NLSR on node a starts first and does not get a 409.
But node b does, so when you create a udp face from one NFD it is also created at the other NFD?

Actions #8

Updated by Ashlesh Gawande about 7 years ago

  • Status changed from New to Code review
  • % Done changed from 0 to 90
Actions #9

Updated by Ashlesh Gawande about 7 years ago

  • Status changed from Code review to Closed
  • % Done changed from 90 to 100
Actions #10

Updated by Ashlesh Gawande almost 7 years ago

  • Target version set to v0.3.2
Actions

Also available in: Atom PDF