bugfix(op-node): syncClient incorrectly removes peer issue

[welkin22]

Background

When the staticPeer is configured with a DNS address (such as /dns/xxx.com/tcp/xxx/p2p/xxx), a peer can obtain multiple IP addresses through domain name resolution. However, in Libp2p, dialing concurrently can result in one peer corresponding to multiple connections, which is exacerbated by multiple IP addresses. If one of the connections is disconnected due to an unstable network or other reasons, the Notifiee.Disconnected event is triggered. At this time, the peer is removed from the syncClient, and the processing flow of the peerLoop exits. The log displays the message stopped syncing loop of peer. However, the peer still has other connections that can be used for p2p synchronization. Therefore, we should not remove this peer unless we determine that all of its connections are disconnected and there are no available connections, just like the check performed by the monitorStaticPeers function.

Solution

Add a condition to determine when to remove a peer from the syncClient: only remove the peer if it cannot establish a connection.

[github-actions]

This PR is stale because it has been open 14 days with no activity. Remove stale label or comment or this will be closed in 5 days.

[welkin22]

The same changes have been submitted to the Optimism chain: ethereum-optimism/optimism#7348 and were approved