I bought a new HDD a few days ago, to get some additional space on my fileserver. I have some 3 disk software raid5 running on it, build with mdadm. I read that kernel 2.6.17 supports raid5 grow (used disks), so I build a new 2.6.17 kernel and got it run.
Here is what I did to grow the raid5 array:
mdadm --grow /dev/md0 --raid-disks=4 --backup-file=/tmp/raid5backupfile
My server is working on the grow-process for the next 400 minutes (by the time I write this post : ), so all looks good ... BUT I was not asked to input a new drive to be added, nor I found something in the manpages which told me, how to add the new disk to the grow process. If I cat /proc/mdstat it shows me, that the array is still reshaping, and that it has [4/3] devices [UUU_].
After adding the new, 4th disk with:
mdadm --manage /dev/md0 --add /dev/hdc
the new disk was shown as spare in the array. I think this means for me, that there will be a resyncing after the reshaping, which will probably take additional 180 minutes.
So my question here is: is there a way to grow and sync a raid5 in one step? Because the way I did it (or am doing it ... it is still reshaping) is not very secure, since while the new 4th disk is not synced, the array is not reliable to diskfailure (I did a backup, because reshaping is marked experimental, but anyway).
Would be great if someone knows a neat solution
BTW, sry for my bad english o.0