[Gluster-users] replace-brick not completing.

Moses Leslie moses at malformed.org
Tue Jun 11 19:58:55 UTC 2013


We have a replica volume with two bricks.

Those two bricks are too small, and I'm replacing them with two much 
larger bricks.

My thought was to use replace-brick on each of them (one at a time) and 
then I would just end up with a larger volume.

The replace-brick action works at first, but after a seemingly random 
amount of data is transfered, the glusterd process on the new brick is 
stuck using 100% of one core, status on the replace-brick shows completed, 
but all the data isn't copied.

The first time it was around 1.8T (of around 2.3), the second time it was 
only 42G.

Is there additional debugging I can do, or is there a different way to get 
where I want?  I want to remove the hardware for the original two bricks 
because it's dev hardware that went into production as a test, it's too 
small to be useful there.



More information about the Gluster-users mailing list