[DRBD-user] MD5-Hash of DRBD Resource

Lionel Sausin ls at numerigraphe.com
Wed Nov 20 10:29:16 CET 2013

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


Le 19/11/2013 22:20, Arnold Krille a écrit :
> On Tue, 19 Nov 2013 14:29:32 +0100 "FDS | Forensik Data Services"
> <office at fds.at> wrote:
>> Just clarifying my understanding of an UpToDate-Status of a DRBD
>> Resource in Dual Primary Mode on 2 nodes.
>> Does a MD5-Hash calculation of both resources lead to equal  MD5
>> values?
> That depends. Prerequisites:
>   - Both primaries are unmounted.
>   - Both primaries are completely synced from the os/fs point of view.
>     (That is "sync" has been run and thus all write-buffers are cleared.)
>
> Then an md5-hash of the drbd-device on both machines should return the
> same. Also md5-hashes of all the underlaying devices should(*) return
> the same.
A MD5 match proves everything is ok BUT a mismatch does not always mean 
trouble.
There may be harmless differences if the layers above "cut corners" and 
modify blocks in memory after they've been sent to DRBD.
RaiserFS and swap memory are known to do this. See "Notes on data 
integrity" at http://www.drbd.org/users-guide/re-drbdconf.html for more 
examples and details.

Lionel Sausin.



More information about the drbd-user mailing list