[DRBD-user] Upgrade from 8.3.3 to 8.3.7 caused full resync

Brian Marshall brian at netcents.com
Sun May 23 09:10:48 CEST 2010

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


Recently I did an upgrade from 8.3.3 to 8.3.7 on both ends of my drbd 
mirrored volume and was surprised to see that the volumes, which had 
been in sync, had gone into a full resync.

I had successfully compiled the drbd module and tools from source on 
both ends and tested with drbdadm -V to make sure both ends had the same 
version installed and working. I then restarted the nodes (primary 
first), which is when the full resync kicked off. Does an upgrade kill 
off the sync status bitmap?

Under normal circumstances this would not be a big deal, but my nodes 
are 2TB and connected via a 12Mbit long-haul wireless link. Syncing 
looks like it will take somewhere between 2 and 4 weeks to finish. There 
are also occasional service interruptions which reset the sync back to 
the beginning again.

My question is this: Is there any way to set the secondary node to valid 
  (the reverse of the invalidate-remote command?). Manually setting the 
bitmap, perhaps? That way I could run a validate and have only the 
changed blocks transferred over the link. Having the validate command 
kick off a full block-by-block checksum verify instead of failing with 
"Can not start OV/resync since it is already active" would be very useful.
Also, as a enhancement request: could a config option be added to have 
syncs resume from where they left off given a service interruption? I 
could probably live with the resync time if it didn't start right from 
the beginning everytime there was a hiccup.
Thanks for the great product and all your help!

    -Brian Marshall

More information about the drbd-user mailing list