Not sending response to the alloc requests with large bucket size for big hard disk

[greg1968]

Package Versions

/home/greg10/.nvm/versions/node/v8.9.4/lib
└── (empty)

v8.9.4


### Expected Behavior

If the bucket size is OK, the node should send a response to the alloc request.
example:
{"level":"info","message":"handling alloc request from 3439d79e19b26af8e9bab492d161f45c48beca11 hash 7da59fcba549d1e15d9d0528865c63773c2608d9 size 101","timestamp":"2018-02-22T01:16:09.878Z"}
{"level":"info","message":"Sending alloc response hash 7da59fcba549d1e15d9d0528865c63773c2608d9 size 101","timestamp":"2018-02-22T01:16:09.964Z"}


### Actual Behavior
It some times does not send response to large KFS bucket, even the farmer's bucket has enough free space. See the log:
{"level":"info","message":"handling alloc request from 872f559aec9973e1b336e9b1befe3341534b1ef6 hash a09c7cfbce65609699b992931fee03405a221d85 size 13974057527","timestamp":"2018-02-22T02:04:58.584Z"}
{"level":"debug","message":"negotiator returned: false","timestamp":"2018-02-22T02:04:58.584Z"}
{"level":"debug","message":"max KFS bucket size 34359738247, used 0, free 34359738247, shard size 13974057527","timestamp":"2018-02-22T02:04:58.771Z"}
{"level":"debug","message":"we have enough free space: true","timestamp":"2018-02-22T02:04:58.771Z"}
{"level":"debug","message":"not sending an offer for the contract","timestamp":"2018-02-22T02:04:58.772Z"}
{"level":"info","message":"replying to message to 1a461d17ac8523f09d7c620d9cdd0d37736ec546","timestamp":"2018-02-22T02:04:58.772Z"}

### Steps to Reproduce

Please include the steps the reproduce the issue, numbered below. Include as
much detail as possible.

1. I have a storage configuration of 120GB
2. After running for some time and the shared storage is 100M, as I see in my area 80% of the alloc request are more than 2GB, I set the configuration file for 8TB.
See the line below, which is my setting:
"storageAllocation": "8TB", 
3. Now I still can not receive those  alloc request for large chunks of data with size of more than 1GB.

[littleskunk]

Works as expected. See #752 for more details.

Default maxShardSize is 4GB on the farmer and bridge side. The new bridge version is not deployed.

[greg1968]

Do you mean that it is the expected behavior for the current version and it is the limitation that it can only accept buckets size that is less than 4 GB?

[littleskunk]

More or less yes. As soon as the new bridge version is deployed the renter will not be able to upload bigger shards.