Replicator error when connect a third device using Peer-to-Peer Sync and MultiPeerConnectivity framework.

[GabrieleNardi]

Description Hi, I am using the Peer-to-Peer Sync to replicate Db changes among devices. Actually I am able to replicate data between two device, but when I connect a third device the replicator crashes.

Expected behavior Replication among the three or more devices without crashes.

Logs

Connected to session: <MCSession: 0x600002c3c320 MyPeerID = <MCPeerID: 0x600001f20990 DisplayName = iPhone 11> SecurityIdentity = (null) EncryptionPreference = None ConnectedPeers = (
    "<MCPeerID: 0x600001f3c4a0 DisplayName = iPhone 11 Pro>",
    "<MCPeerID: 0x600001f3c9b0 DisplayName = iPhone XS>"
) Delegate = <CouchbaseTest.MultiPeerConnectivityManager: 0x6000036182a0>>
Received: �=Profile�getCheckpoint�client�cp-MH1LMCIi3y1zKPIqxIV5S3TN+SA=�òÃý¬
2020-02-28 12:01:14.808357+0100 CouchbaseTest[75396:2202630] CouchbaseLite Network ERROR: {BLIPIO#3}==> litecore::blip::BLIPIO ->x-msg-endpt:////_blipsync @0x7f8bfeb188b8
2020-02-28 12:01:14.808495+0100 CouchbaseTest[75396:2202630] CouchbaseLite Network ERROR: {BLIPIO#3} Caught exception handling incoming BLIP message: BLIP protocol error: Bad incoming REQ #1 (already finished)
2020-02-28 12:01:14.808616+0100 CouchbaseTest[75396:2202630] CouchbaseLite Database WARNING: Caught unexpected C++ runtime_error("BLIP protocol error: Bad incoming REQ #1 (already finished)")
2020-02-28 12:01:14.808791+0100 CouchbaseTest[75396:2203037] CouchbaseLite Network WARNING: {C4SocketImpl#4}==> litecore::repl::C4SocketImpl x-msg-endpt:////_blipsync @0x7f8bfeb0b8e0
2020-02-28 12:01:14.808887+0100 CouchbaseTest[75396:2203037] CouchbaseLite Network WARNING: {C4SocketImpl#4} WebSocket closed abnormally with status 4002
2020-02-28 12:01:14.809313+0100 CouchbaseTest[75396:2203037] CouchbaseLite Replicator ERROR: {Repl#5}==> litecore::repl::Replicator /Users/gabrielenardi/Library/Developer/CoreSimulator/Devices/D8C68024-9B88-4C6F-94A4-0205CF17F1C3/data/Containers/Data/Application/9BB2A004-D5F6-468C-BD43-141D0AF5B7FC/Library/Application Support/CouchbaseLite/peerToPeerSync.cblite2/ ->x-msg-endpt:////_blipsync @0x7f8c0101c828
2020-02-28 12:01:14.809447+0100 CouchbaseTest[75396:2203037] CouchbaseLite Replicator ERROR: {Repl#5} Got LiteCore error: LiteCore error 10 "BLIP protocol error: Bad incoming REQ #1 (already finished)"
2020-02-28 12:01:14.809620+0100 CouchbaseTest[75396:2203037] CouchbaseLite Replicator ERROR: {Repl#5} Stopping due to fatal error: LiteCore error 10 "BLIP protocol error: Bad incoming REQ #1 (already finished)"

Platform:

• Device: [iPhone XS, iPhone 11, iPhone 11 Pro]
• OS: iOS13.3
• Couchbase Version: CouchbaseLiteSwiftEnterprise 2.7.0

Additional context I got a live query in viewDidLoad who gets all messages stored in my DB. Below the code:

func getAllTextType(callback: @escaping(([DataChatMessage]) -> Void)) {
        
        let query = QueryBuilder.select(SelectResult.all())
            .from(DataSource.database(self.database))
            .where(Expression.property("type")
            .equalTo(Expression.string("text")))
        
        self.token = query.addChangeListener { (query) in
            
            let messages = query.results?.allResults().compactMap(ModelMapper.resultsToDataChatMessage).sorted(by: { $0.creationDate < $1.creationDate })
            messages.flatMap(callback)
        }
    }

[snej]

That's not a crash. The replicator stopped due to an error in the underlying framing protocol — most likely your P2P code is passing the wrong data.

BLIP protocol error: Bad incoming REQ #1 (already finished)

This error means it received a duplicate message ID. Make sure you're not passing along the same message twice, or passing a message to the wrong replicator instance.

[GabrieleNardi]

Thank you for the answer.

Im in this situation: Device A (ActivePeer) which start connection with Device B (PassivePeer).

Then how I should insert the third device? Can I directly connect to A? Making Device A being Active or Passive Peer for Device C?

I think the error comes because I'm trying to make a Multipeer chat, and every devices has all the history. So Device B and Device C trying to push to Device A the same messages. Can this cause the error?

Are there some updated examples of a Multipeer push and pull replication? I only found Photo-Drop but the code is obsolete (about 5 years ago).

PS: Can the the target of MessageEndpoint(uid:, target:, protocolType:, delegate:) be [MCPeerID]? Actually I have a single MCPeerID as target but I need the replication among multiple devices. (So I have a [ActivePeerConnection] in VC. The DB is a single instance, but there are many Replicator as connected devices.)

PPS: Have I to connect all device involved in replication before I create the connection by func createConnection(endpoint: MessageEndpoint) -> MessageEndpointConnection? Or I can add device to session even after the createConnection has triggered? Creating more than one Connection.

[pasin]

Then how I should insert the third device? Can I directly connect to A? Making Device A being Active or Passive Peer for Device C?

It could be either ways when device A connects to C. I think the logic could be the same when A connects to B. The setup could be mesh that each device connects to each other or star which one device is a host like a sync server.

I think the error comes because I'm trying to make a Multipeer chat, and every devices has all the history. So Device B and Device C trying to push to Device A the same messages. Can this cause the error?

If it is the same document, there should be an issue. I think the error that you have is causing by sending a message to a wrong peer.

Are there some updated examples of a Multipeer push and pull replication? I only found Photo-Drop but the code is obsolete (about 5 years ago).

The example was based on 1.x solution. I don't think we have an example that uses MC. CC: @rajagp

PS: Can the the target of MessageEndpoint(uid:, target:, protocolType:, delegate:) be [MCPeerID]? Actually I have a single MCPeerID as target but I need the replication among multiple devices. (So I have a [ActivePeerConnection] in VC. The DB is a single instance, but there are many Replicator as connected devices.)

MessageEndpoint represents 1-1 connection between active peer (replicator) and message endpoint listener (like server). MessageEndpoint cannot connect to multiple message endpoint listeners. If you have one device acting like a host or server running message endpoint listener, the other devices will create replicator connecting to the listener.

The DB is a single instance, but there are many Replicator as connected devices.)

I seems like you should have it backward by having a message endpoint listener on one device and the other will have a replicator connecting to the listener.

PPS: Have I to connect all device involved in replication before I create the connection by func createConnection(endpoint: MessageEndpoint) -> MessageEndpointConnection? Or I can add device to session even after the createConnection has triggered? Creating more than one Connection.

The replicator will call createConnection(endpoint: MessageEndpoint) -> MessageEndpointConnection? to create a connection to the message endpoint listener. You cannot share the same connection with multiple peers. If you have multiple replicators, each replicator will create its own MessageEndpointConnection.

[GabrieleNardi]

If it is the same document, there should be an issue. I think the error that you have is causing by sending a message to a wrong peer.

So in my case where the most of documents are the same on all devices (group chat with 20-30 devices), how can I handle the replication among multiple peers?

To avoid the quoted case can I only create a chain of replication A <-> B <-> C... or maybe a star network where the center is passivePeer (server) and the other are activePeers (clients)?

Are there any other options?