Skip to content

Latest commit

 

History

History
86 lines (67 loc) · 3.68 KB

README.md

File metadata and controls

86 lines (67 loc) · 3.68 KB

Remote configuration of a LIO-based storage appliance

targetd turns Linux into a remotely-configurable storage appliance. It supports an HTTP/jsonrpc-2.0 interface to let a remote administrator allocate volumes from an LVM volume group, and export those volumes over iSCSI. It also has the ability to create remote file systems and export those file systems via NFS/CIFS (work in progress).

targetd's sister project is libStorageMgmt, which allows admins to configure storage arrays (including targetd) in an array-neutral manner.

targetd development

targetd is licensed under the GPLv3. Contributions are welcome. Please format code changes using: black -t py36 . before submitting your request as CI checks formatting.

NOTE: targetd is STORAGE-RELATED software, and may be used to remove volumes and file systems without warning from the resources it is configured to use. Please take care in its use.

Getting Started

targetd has these Python library dependencies:

All of these are available in Fedora Rawhide and recent Ubuntu versions.

Configuring targetd

A configuration file may be placed at /etc/target/targetd.yaml, and is in YAML format. Here's an example:

user: "foo" # strings quoted, or not
password: bar
ssl: false
target_name: iqn.2003-01.org.example.mach1:1234

block_pools: [vg-targetd/thin_pool, vg-targetd-too/thin_pool]
fs_pools: [/mnt/btrfs]

portal_addresses: ["192.168.0.10"]

targetd defaults to using the "vg-targetd/thin_pool" volume group and thin pool logical volume, and username 'admin'. The admin password does not have a default -- each installation must set it. Use the portal_addresses parameter to set explicit addresses that LIO should direct iSCSI connections to, this is useful if you are using a proxy such that LIO cannot correctly detect the public address (e.g. a Kubernetes service). The default behavior is to listen on all addresses (0.0.0.0).

Then, in the root of the source directory, do the following as root:

# export PYTHONPATH=`pwd`
# ./scripts/targetd`

client.py is a basic testing script, to get started making API calls.

Docker

targetd can be run in a Docker container. This requires mounting sensitive host directories and granting privileged access in order to set up LVM volumes. Use the following command:

docker build -t targetd -f docker/Dockerfile .
docker run --privileged -v /etc/target:/etc/target -v /sys/kernel/config:/sys/kernel/config -v /run/lvm:/run/lvm -v /lib/modules:/lib/modules -v /dev:/dev -p 18700:18700 -p 3260:3260 targetd

where your config is stored at /etc/target on the host machine.

Clients/libraries utilizing JSON API