...The mfs works by storing a reference to the root node's CID in LevelDB. LevelDB does not support concurrent access so there are read/write locks around bits of the code that modify the the root node's CID. A lock is kept on the main thread and any requests to read/write from workers or the main thread itself are queued pending release of the lock by the existing holder. Reads are executed together, writes are executed sequentially, and prevent any reads from starting. If you are using IPFS in a single process or with the node cluster module this should be completely transparent. ...