Open
Description
When hyperdrive-publisher gets started as github action, it may very well be possible that (by accident, misconfiguration or parallelization) the script gets started at the same time twice. As far as I can see this may cause forks in the hypercore (corrupt the data) and that is not such a good thing.
I can a few ways to mitigate this issue and I am wondering how relevant each would be.
- Add a lock file to the file system that prevents parallel starts on a OS/process level.
- Add a feature to the pinning host/server (in conjunction with Add target identification support #11) that prevents them to connect two two seeding pees at the same time.
- Before syncing of the file system: connect to the dht to see if there is another peer that also possesses the secret key.