bug: constantly disconnects from a peer

[weboko]

This is a bug report

Problem

Right now js-waku node instance constantly connects and disconnects to a remote peer:

As @danisharora099 pointed out - it is most likely due to metadata protocol dropping connection and connection manager trying to re-connect.

We should:

• improve logging to understand why disconnect happens;
• investigate why this happens and avoid connecting to peers connection to which will be dropped;

Notes

Version:

  "name": "@waku/sdk",
  "version": "0.0.24",

[vpavlin]

I am seeting these errors in light-js example which might be related

Peer connected with peer id: 16Uiu2HAm2M7xs7cLPc3jamawkEqbr7cUJX11uvY7LxQ6WFUdUKUT
libp2p-DmEl0riZ.js:6360 waku:info:connection-manager Connected to peer 16Uiu2HAm2M7xs7cLPc3jamawkEqbr7cUJX11uvY7LxQ6WFUdUKUT +363ms
libp2p-DmEl0riZ.js:6360 waku:info:stream-manager:/vac/waku/metadata/1.0.0 Preemptively opening a stream to 16Uiu2HAm2M7xs7cLPc3jamawkEqbr7cUJX11uvY7LxQ6WFUdUKUT +1s
libp2p-DmEl0riZ.js:6360 waku:info:stream-manager:/vac/waku/metadata/1.0.0 Preemptively opening a stream to 16Uiu2HAm2M7xs7cLPc3jamawkEqbr7cUJX11uvY7LxQ6WFUdUKUT +1ms
libp2p-DmEl0riZ.js:6360 waku:info:stream-manager:/vac/waku/metadata/1.0.0 Preemptively opening a stream to 16Uiu2HAm2M7xs7cLPc3jamawkEqbr7cUJX11uvY7LxQ6WFUdUKUT +2ms
index.html:150 Uncaught (in promise) TypeError: Cannot read properties of undefined (reading 'multiaddr')
    at index.html:150:54
    at Array.map (<anonymous>)
    at updatePeersList (index.html:148:38)
libp2p-DmEl0riZ.js:6360 waku:info:stream-manager:/vac/waku/metadata/1.0.0 Preemptively opening a stream to 16Uiu2HAm2M7xs7cLPc3jamawkEqbr7cUJX11uvY7LxQ6WFUdUKUT +767ms
libp2p-DmEl0riZ.js:6360 waku:info:stream-manager:/vac/waku/store/2.0.0-beta4 Preemptively opening a stream to 16Uiu2HAm2M7xs7cLPc3jamawkEqbr7cUJX11uvY7LxQ6WFUdUKUT +2s
libp2p-DmEl0riZ.js:6360 waku:info:stream-manager:/vac/waku/metadata/1.0.0 Preemptively opening a stream to 16Uiu2HAm2M7xs7cLPc3jamawkEqbr7cUJX11uvY7LxQ6WFUdUKUT +1ms
libp2p-DmEl0riZ.js:6360 waku:info:stream-manager:/vac/waku/store/2.0.0-beta4 Preemptively opening a stream to 16Uiu2HAm2M7xs7cLPc3jamawkEqbr7cUJX11uvY7LxQ6WFUdUKUT +1ms
libp2p-DmEl0riZ.js:6360 waku:info:stream-manager:/vac/waku/metadata/1.0.0 Preemptively opening a stream to 16Uiu2HAm2M7xs7cLPc3jamawkEqbr7cUJX11uvY7LxQ6WFUdUKUT +2ms
libp2p-DmEl0riZ.js:6360 waku:info:stream-manager:/vac/waku/store/2.0.0-beta4 Preemptively opening a stream to 16Uiu2HAm2M7xs7cLPc3jamawkEqbr7cUJX11uvY7LxQ6WFUdUKUT +2ms
libp2p-DmEl0riZ.js:6360 waku:error:stream-manager:/vac/waku/metadata/1.0.0 Failed to prepare a new stream for 16Uiu2HAm2M7xs7cLPc3jamawkEqbr7cUJX11uvY7LxQ6WFUdUKUT +2s
libp2p-DmEl0riZ.js:6360 waku:error:stream-manager:/vac/waku/metadata/1.0.0 Failed to prepare a new stream for 16Uiu2HAm2M7xs7cLPc3jamawkEqbr7cUJX11uvY7LxQ6WFUdUKUT +1ms
libp2p-DmEl0riZ.js:6360 waku:error:stream-manager:/vac/waku/metadata/1.0.0 Failed to prepare a new stream for 16Uiu2HAm2M7xs7cLPc3jamawkEqbr7cUJX11uvY7LxQ6WFUdUKUT +0ms
libp2p-DmEl0riZ.js:6360 waku:error:stream-manager:/vac/waku/store/2.0.0-beta4 Failed to prepare a new stream for 16Uiu2HAm2M7xs7cLPc3jamawkEqbr7cUJX11uvY7LxQ6WFUdUKUT +2s
libp2p-DmEl0riZ.js:6360 waku:error:stream-manager:/vac/waku/metadata/1.0.0 Failed to prepare a new stream for 16Uiu2HAm2M7xs7cLPc3jamawkEqbr7cUJX11uvY7LxQ6WFUdUKUT +0ms
libp2p-DmEl0riZ.js:6360 waku:error:stream-manager:/vac/waku/store/2.0.0-beta4 Failed to prepare a new stream for 16Uiu2HAm2M7xs7cLPc3jamawkEqbr7cUJX11uvY7LxQ6WFUdUKUT +0ms
libp2p-DmEl0riZ.js:6360 waku:error:stream-manager:/vac/waku/metadata/1.0.0 Failed to prepare a new stream for 16Uiu2HAm2M7xs7cLPc3jamawkEqbr7cUJX11uvY7LxQ6WFUdUKUT +0ms
libp2p-DmEl0riZ.js:6360 waku:error:stream-manager:/vac/waku/store/2.0.0-beta4 Failed to prepare a new stream for 16Uiu2HAm2M7xs7cLPc3jamawkEqbr7cUJX11uvY7LxQ6WFUdUKUT +0ms
index.html:160 

[danisharora099]

After some work on #1994,
some updates on findings re disconnections:

• by default, we are currently bootstrapping into the fleet with TWN, while by default on js-waku, we used the DefaultPubsubTopic
• the js-waku node connects to these nodes, the metadata protocol then hangs up the connection because of shard mismatch

---

few action items come out of this issue:

• map the correct bootstrap fleet of nodes, according to the pubsub topic configured on the js-waku node
• adds logs around disconnection due to shard mismatch
• check why redials aren't prevented after disconnection due to shard mismatch (shardInfo should be updated in the peerStore)

cc @vpavlin @weboko

[danisharora099]

closing this issue in favour of the above issues.

if you think something is missed, please feel free to reopen the issue

cc @vpavlin @weboko