dont trigger noSpaceLeft on huge shards

[littleskunk]

• Closes #752

[braydonf]

So without this, I do not think spaceAvailable will ever be set to false, as we're not running runSpaceCheck to start the interval that will check it due to that method currently using too much CPU and memory on large databases.

We are running the connectBridges interval that will update a bridge contact info for a farmer to have spaceAvailable if the values differ (https://github.com/littleskunk/core/blob/644d4af70317eba050e55b751fff074a7866a18e/lib/network/farmer.js#L380-L385).

[littleskunk]

Can you link that interval? I haven't seen something like that. The connectBridges is called only once on startup as far as I know.

[braydonf]

Yeah it's here: https://github.com/littleskunk/core/blob/644d4af70317eba050e55b751fff074a7866a18e/lib/network/farmer.js#L313-L319

[littleskunk]

How about this: https://github.com/Storj/core/blob/9691c6a17991a0487e1069a1000db5035b1c57ba/lib/network/farmer.js#L284-L286

As far as I can see the farmer is already connected with the bridge and will skip every additional call?

[braydonf]

Ah yeah, that could be it.

[braydonf]

The check for differences in contact data is after that check in _connectBridge

[braydonf]

I probably neglected to write an integration test for all those methods being used together, and hence the oversight.

[braydonf]

Okay, merging this and we can fix updating spaceAvailable in a new PR, and then we can make a release

[braydonf]

Actually, @littleskunk the connected flag gets set to false when noSpaceLeft is called, which should trigger the check and update: https://github.com/Storj/core/blob/master/lib/network/farmer.js#L365

[braydonf]

We may just need to fix runSpaceCheck and re-enable that interval.

[braydonf]

Continuing on this with a new PR https://github.com/Storj/core/pull/755