[DRBD-user] correct way to wipe drbd fs

Lars Ellenberg lars.ellenberg at linbit.com
Tue Feb 9 16:00:15 CET 2016

Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.


On Tue, Feb 09, 2016 at 05:22:46AM +0100, krautus at kr916.org wrote:
> Hi all, as you can imagine i need some help :)
> I reinstalled the OS on a server, from debian 7 with drbd
> to debian 8 without drbd
> (I actually don't need drbd anymore on this server)
> 
> The installation went fine, as well as booting and performance,
> but .. I'm getting a few of this line in the kern logs:
> systemd-gpt-auto-generator[18808]: Failed to determine partition table type of /dev/sda: Input/output error
> around 30 times and then it stops.
> 
> So after some googling I've tried this:
> 
> root at pcmr:~# lsblk -fs
> NAME  FSTYPE LABEL UUID                                 MOUNTPOINT
> sda1  xfs          1490d770-8c7a-41b9-9af0-1761f7cde4e1 /boot
> └─sda drbd         fda4aa032add9632                     
> sda2  xfs          26d59afd-a2c4-488d-a286-9573e2730f84 /
> └─sda drbd         fda4aa032add9632                     
> sda3  btrfs        c7ce9dc4-f17f-417b-9660-58c7a5400aa7 /var
> └─sda drbd         fda4aa032add9632                     
> sda4  swap         366cf7e2-fada-4b64-b2cb-faae467914cf [SWAP]
> └─sda drbd         fda4aa032add9632                     
> 
> and as you can see there are some drbd bits which survived the format.
> 
> It looks like I need to use wipefs
> but since I access this server only thru a kvm (and must upload the iso from my home network, very slow ..),
> I'd like to ask you: what's the correct way to clean an hdd from a previous drbd cluster ?
> Can I do it directly from the OS or do I need to run a live-cd ?

afaik, libblkid picks up on the "DRBD"-ness of a device
by looking at its last aligned 4k block,
specifically checking that the four bytes
starting from offset 60 decimal, 0x3c hex,
in that block are hex [83 74 02 6b].

Which is a crude but effective heuristic.
It does not catch all kinds of DRBD meta data,
but at least covers most typical deployments.

If you flip just one bit there,
it should stop being recognized as "DRBD".

wipefs usually just zeroes those four bytes.
But ask it what it sees at whic offsets,
wipefs /dev/sda should tell you.

There is also the "wipe-md" subcommand to drbdmeta,
so while you are "un-deploying" DRBD,
you could start with "drbdadm wipe-md <resource>".

> Thanks for your help, btw i use drbd from around 5 years on many
> servers, also under quite heavy load and never had a single problem,
> never lost a bit *thumbs up*

Cheers,

-- 
: Lars Ellenberg
: http://www.LINBIT.com | Your Way to High Availability
: DRBD, Linux-HA  and  Pacemaker support and consulting

DRBD® and LINBIT® are registered trademarks of LINBIT, Austria.
__
please don't Cc me, but send to list   --   I'm subscribed



More information about the drbd-user mailing list