[Gluster-devel] gluster rebalance progress

[Anand Babu Periasamy]

On Tue, Jun 14, 2011 at 12:24 AM, Emmanuel Dreyfus <manu at netbsd.org> wrote:
> Hi
>
> Since I am porting glusterfs, I am never fully sure it really works as
> intended. When I try gluster volume rebalance, I get no error reported,
> but the operation seems to take forever.
>
> Here is what it says:
> silo# gluster volume rebalance gfs status
> rebalance step 1: layout fix in progress: fixed layout 14204
>
> How can I know it is really busy and not stuck somewhere? How long is
> it supposed to take?
>
> --
> Emmanuel Dreyfus
> manu at netbsd.org

Hi Emmanuel, once your re-balance completes, you will see "completed"
status message. Total time depends on amount of data in your volume.
Eg:
# gluster volume rebalance test-volume status
Rebalance completed: rebalanced 3107 files of size 1428745 (total
files scanned 6000)

It will be nice to show progress (time estimate or at least
files-scanned/total-files). To implement this feature, we first need
to crawl the name space to count total number of files and amount of
data. Crawling the name space itself is a slow process. A better
approach is to implement something like the de-fragment tool in NTFS.
Instead of doing rebalance in 2 complete phases, it should become
discrete. Users should look at this tool as an optimizer (for capacity
or performance). You should be able to run or stop any time you want.
Longer you run, more it optimizes, but beyond some point it gives
diminishing returns.  There is a lot of scope for improvement here.
Quota, marker and profiling features can provide the necessary
intelligence.

-- 
Anand Babu Periasamy
Blog [http://www.unlocksmith.org]

Imagination is more important than knowledge --Albert Einstein