[DRBD-user] Unrelated data, aborting! after split brain

Alex Kuehne alex at movx.de
Mon Jan 24 16:38:48 CET 2011

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


Quoting Alex Kuehne <alex at movx.de>:

> Hi guys,
>
> In order to prepare a DRBD cluster for production, I'm playing  
> around with split brain scenarios. First test worked after  
> instructions from  
> http://www.drbd.org/users-guide/s-resolve-split-brain.html, but  
> second test gives me that error after drbdadm connect:
>
> kernel: block drbd1: uuid_compare()=-1000 by rule 100
> kernel: block drbd1: Unrelated data, aborting!
>
> I don't know why both nodes don't like each other anymore. I'm  
> curios to understand what happened here.
>
> Here are my data:
>
> DRBD ver 8.3.10rc2
>
> on split brain survivor:
> # cat /proc/drbd
> version: 8.3.10rc2 (api:88/proto:86-96)
> GIT-hash: a38b219864715e807e785488d5907a273f54adb6 build by  
> root at std11526-vm03, 2011-01-14 07:08:33
>
>  1: cs:StandAlone ro:Primary/Unknown ds:UpToDate/DUnknown   r-----
>     ns:0 nr:0 dw:222947 dr:2450 al:25 bm:29 lo:0 pe:0 ua:0 ap:0 ep:1  
> wo:b oos:7988
>
> on split brain victim:
> # cat /proc/drbd
> version: 8.3.10rc2 (api:88/proto:86-96)
> GIT-hash: a38b219864715e807e785488d5907a273f54adb6 build by  
> root at std11526-vm03, 2011-01-14 07:08:33
>
>  1: cs:StandAlone ro:Secondary/Unknown ds:UpToDate/DUnknown   r-----
>     ns:0 nr:0 dw:0 dr:208 al:0 bm:0 lo:0 pe:0 ua:0 ap:0 ep:1 wo:b oos:0
>
> My DRBD conf here: http://p.0wnz.at/205018, global_common.conf is  
> used as shipped.
>
> Here is where I suspect the problem:
>
> kernel: block drbd1: drbd_sync_handshake:
> kernel: block drbd1: self  
> 26B93DADD512C783:65ED83424EBC4E55:C1A7166A8B12AF84:C1A6166A8B12AF85  
> bits:2166 flags:0
> kernel: block drbd1: peer  
> B2D0D9A2D3A1044E:52265F28809CCB48:0001000000000000:06947CB397C2C578  
> bits:0 flags:1
> kernel: block drbd1: uuid_compare()=-1000 by rule 100
> kernel: block drbd1: Unrelated data, aborting!
> kernel: block drbd1: conn( WFReportParams -> Disconnecting )
>
> So what can I do to get both nodes to sync again, without having to  
> recreate the whole mirror?
>
> Best regards,
> Alex Kuehne

Ok for the record I did the following on split brain victim:

drbdadm detach drbd-sr1
dd if=/dev/zero of=/dev/sda3 bs=4096 count=1000
drbdadm create-md drbd-sr1
drbdadm up drbd-sr1

where "drbd-sr1" is my DRBD mirror resource. After that everything was  
working fine again.

BR,
Alex Kuehne



More information about the drbd-user mailing list