IO error: lock hold by current process

[bfollington]

Encountered by @justinabrahms after trying to sync (with a period of inactivity since last sync).

Subsequent attempts failed with different messages, seemingly intermittently:

[bfollington]

Current vague suspicion: large history causing issues, perhaps resulting in a slow rocksdb teardown and multiple attempts to connect at once?

[bfollington]

I've been doing some investiation into whether it's possible to construct two noosphere instances in Subconscious, since this would be the most straightforward explanation of the error we're seeing here.

I've come up empty handed. Even if the issue were to do with disposing and recreating our noosphere reference (i.e. #1007) I can't trigger disposal by performing a regular sync.

Next steps: add more detailed logging to the app and ship and update to see if we can learn more about what's happening on @justinabrahms device.

[jsantell]

WRT the LOCK file issue: it could be possible on backgrounding, rocks was closing its background threads (which fcntl will release the LOCK file upon completing), and on reopening Subconscious, the database (lock) was still in use -- subconsciousnetwork/noosphere#734 may fix this issue in this case