FAQ: LeoFS Limits


  • LeoFS have covered almost major AWS S3-API but not all APIs.
  • If you use S3's Multi Part Upload API, the size of a part of an object must be less than the size of a chunked object in LeoFS.
  • When using the multi datacenter replication feature, we have supported up to 2 clusters with LeoFS v1.2, but we're going to support over 3 clusters replication with LeoFS v2.0.
  • When you run recover-node while another recover-node is already working in-progress, objects that are not transmitted to the target node yet in the former recover-node are canceled and the latter recover-node will take place. To ensure that every recover-node completes its job, we'd recommend you to run recover-node one-by-one.

  • Related Links:


  • When you upgrade your LeoFS, you can NOT change the metadata storage as |KVS| - bitcask or leveldb can be used in LeoFS - used by LeoFS Storage. We recommend users to replace bitcask with leveldb by using |leofs_utils/tools/b2l|.

  • Related Links:

NFS Support

  • NFS implemantation with LeoFS v1.1 is a subset of NFS v3. Authentication, and Owner/Permission management are NOT covered.
  • The ls command may take too much time when the target directory have lots of child. We're planning to provide better performance with LeoFS v.2.0.
  • If you use LeoFS with NFS, you should set the size of a chunked object in LeoFS to 1MB (1048576Bytes) - large_object.chunked_obj_len = 1048576 (leo_gateway.conf), otherwise the efficiency of disk utilization can be decreased.
  • What are the requirements to run LeoFS with NFS?

    • LeoGateway (NFS Server):
      • We've supporeted the targets Debian 6, Ubuntu-Server 14.04 LTS or Higher and CentOS 6.5/7.0 as LeoFS does, but should work on most linux platforms. In addition, We've confirmed LeoFS with NFS works properly on the latest FreeBSD and SmartOS by using NFS Integration Test Tool.
  • Related Links: