FAQ: LeoFS Limits

Features

  • LeoFS have covered almost major Amazon S3 REST API but not all.
  • If you use Amazon S3 multipart 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 DC feature, we have supported up to 2 clusters with LeoFS v1.2.8 or lowers, but we will support over 3 clusters with LeoFS v1.4.
See Also:

Operations

  • 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.
See Also:

NFS Support

  • NFS implemantation with LeoFS v1.1 is a subset of NFS v3. Lock manager protocol, 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.1.4.

  • 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?

    • LeoFS Gateway(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.

    • NFS Client

      Since we still have not implemented NFS Lock Manager, you can only use NFS Clients which support disabling lock option(the linux client might be the only option for now). After we will finish implementing NFS Lock Manager, You can use NFS Client on other xnix platforms as well.

See Also: