[Gluster-devel] gluster-deploy tool

Paul Cuzner pcuzner at redhat.com
Mon Dec 23 23:35:13 UTC 2013


Just a quick update as to where the deploy tool is currently.

Here's what's been implemented so far;

- cluster can be built in two ways
  - via a subnet scan - the user chooses which subnet to scan (based on the subnets seen on the server running the tool)
  - via a config file that supplies the nodes to use in the cluster (-f invocation parameter)
- ssh keys are propagated from the node running the deploy tool to the other nodes
- ssh is then used to discovery free i.e. unused disks using a python script.
- user then selects the disks they want to use
- Once the disks are selected, you define how you want the bricks formatted
  - lvm (default)
  - lvm with dm-thinp
  - btrfs (not supported yet, but soon!)
  NB When you choose snapshot support (lvm + dm-thinp or btrfs), confirmation is required since these are 'future' features

- With the bricks formatted, you can define a volume
  - tuning presets are provided for cinder, glance, ovirt/rhev, and hadoop
  - dist and dist-repl type volumes are supported
  - for repl volumes, the UI prevents bricks from the same server being assigned to the same replica set
  - UI shows a summary of the capacity expectation for the volume given the brick configuration and volume type
- the UI polls for updates during long running commands to keep the user informed of progress (subnet scan, brick formatting and vol create)
- the tool provides a summary to stdout, and logs to a separate log file (currently in debug mode)

I find it useful when testing different configs, especially coupled with the undodeploy script in the utils folder. This allows you to stand-up and tear-down different test configurations very quickly, especially when you use the new config file option (-f).

As usual - the code is on the forge - https://forge.gluster.org/gluster-deploy

Have a great Christmas and New Years!


Paul C


More information about the Gluster-devel mailing list