NodeFSStorageAdapter lock exception on large number of updates.

[cristianmadularu]

Hi guys, I am noticing a problem when using the NodeFSStorageAdapter where it seems to crash due to a locking issue. It seems to happen when we are doing a large number of document updates.
I should add that when that happens, it seems to corrupt something since restarting the process causes all the peers (including the new process) to run some infinite sync loop which never ends until the process crashes again with the same error. In order to 'fix' the problem, we delete all the data from that node. After that, on restart, it is able to normally sync all its data from the other nodes.
Any thoughts? Thank you!

[pvh]

Neat! This is probably pretty reproducible with a test. I'm guessing it's a file system handle exhaustion issue? Probably we can correct this with a little bit of back-pressure, a queue, or some retry logic. Can you get me either a failing test case or an example repo? Thanks!

[cristianmadularu]

I wish I could :(. We are still fighting this random infinite sync loop. It seems to happen when a new peer (a web app - probably not relevant) connects or disconnects. When that happens (the infinite sync loop), it takes about 30 seconds for the Node peer to throw that error.