Conflicts when joining multiple clusters

Can a single ipfs node/daemon be controlled by multiple cluster peers? It is mentioned on https://collab.ipfscluster.io/ that we can “Follow as many collaborative clusters as you want, the different follower peers will run in parallel”. What will happen when there’s a conflict? For instance, one cluster peer tells you to unpin a file, the other tells you to pin it.

Also, is it possible for both a host for a collaborative cluster and a follower for another collaborative cluster to talk to the same IPFS node ?

The CID will be unpinned first. The other cluster will realize that the CID that should be pinned was unpinned and re-pin it again. The first cluster will not do anything else since that CID is not its problem anymore.

Yes, why not.

1 Like

The other cluster will realize that the CID that should be pinned was unpinned and re-pin it again.

Is there a risk of garbage collection before the second cluster “realizes” that the CID should be pinned? Also what is the realization process like? Is this feature well supported?

Yes, if GC is enabled the Kubo can delete the CIDk before the second cluster realizes that it needs to be marked as pinned again.

Running several clusters on top of the same Kubo daemon is something you can do, not something that we have designed or thought as a usecase though.

1 Like