Time required to change from RAID 1 to RAID 10

1

Greetings,

       I am using Xeon single core processor with 4 X 10 TB Hdd, Earlier 2 of the 10 TB hardisk was configured to be single pool with RAID 1. Now two more 10 Tb hardisk has to be added  to same pool with RAID 10. I have started this process. It's been 48 hrs and the NAS box is still continuing. Does anyone has any idea how much time does such process take. Meanwhile I cannot access NAS box by web gui. Adding a console also shows blank screen. However it is responding to pings.

Any pointers,
Regards,
Ashima

2

I wouldnt expect that the UI should become unresponsive during a raid level change, it sounds to me like something’s gone wrong. (Btrfs should be able to do live raid level changes transparently)

When you say ‘adding a console’, do you mean are you referring to a local monitor or ssh? If not SSH, could you try it and advise results?

Short answer: I think your NAS has crashed. A reboot may be necessary.
Btrfs is reasonably good at not breaking when rebooting during an operation (see my own post history for evidence), however there is a possibility of data loss.

Assuming you come back up good, make sure you balance after the raid level change!

3

Thanks @Haioken for replying.

          I have connected a local monitor but it shows black screen.  Even SSH not working. Though I am not sure SSH was enabled in the box.

Regards
Ashima

4

It sounds like a nasty situation.
Unfortunately, with no local shell, ssh nor gui, there’s no debug information available.

Be warned: While I try to provide helpful information and insights, I am just a user like you - with a decent linux knowledgebase. My answers may not be the best advice for a given situation, and may break your stuff, burn down your house or kill your dog. If this is the case, they do not reflect on Rockstor or its developers.

I’m fairly certain that it’s not supposed to become non-responsive during RAID level changes (at least, not by btrfs design), so I personally would suggest a hard reboot, then (from shell):

btrfs balance status /mnt2
btrfs fi show

The first will tell us the balance state from the raid conversion. If it had not completed, it should try to continue.

If it reports incomplete and doesn’t continue:

btrfs balance resume /mnt2

The second will tell us how balanced the disks actually are. As your disks are all 10Tb, the usage should theoretically be very close to even.

I myself had some issues with system lockups during a balance when adding disks and re-balancing, so I feel your pain. Mine stalled on balance 3 times requiring a hard reboot each time.

If your disks continue balancing, ensure that you have a monitor and keyboard attached at all times during the balance.

You can monitor the balance status live with:

watch -n 10 btrfs balance status /mnt2

Look for the line like:

450 out of about 750 chunks balanced (451 considered), 40% left

If the values dont change for more than (at a guess, I don’t have 10tb disks!) 15 minutes, it’s probably stalled and requiring a hard reboot.

Also worth watching ps, top and syslog in other tty’s. I was seeing btrfs processes stuck in uninterriptable sleep, increasing io wait as a result when I experienced balacing issues.

Hope this helps.