Forum
Welcome, Guest
Username: Password: Remember me
  • Page:
  • 1

TOPIC: Recovery from a failed drdb sync

Recovery from a failed drdb sync 3 months 3 weeks ago #1816

  • Carlos Marchi
  • Carlos Marchi's Avatar Topic Author
  • Offline
  • Posts: 1
Hi there!

First time posting here, first of all, thank you for this amazing software! I have a customer running two servers in sync for about a year, and it already saved his operation from a hardware failure.

During this night, one of these nodes (let's call it server2, and it was the home for VMs) went down for no known reason. The other (server1) assumed as a master, but we then discovered that the storage wasn't in sync since a few months (sorry, definetely a really bad mistake from me not overseeing this). I've manually turned off the node that has assumed as a master, and rebooted the host that have failed during the night. The node is working again with updated data. Summary:

Server2 - Down for about 6 hours, running again with the data version that we want to keep
Server1 - Out of sync, up for about 6 hours, running an outdated version of vms because of the missing sync

But, now I don't know how to safely sync them again, I'm affraid that they will have mismatched data. Since the Server1 went up and was running for a few hours, I'm affraid that this version overtakes do other one, that could have an older timestamp (at least for parts of the disks).

Any advice? I tought about resyncing server 1 from scratch (disable network, destroy the storage and re-create), would be that a safeway?

Thank you!

Please Log in or Create an account to join the conversation.

Recovery from a failed drdb sync 3 months 3 weeks ago #1817

There is a split-brain recovery script that will let you choose which server has the best data. It's in /etc/iscsi-ha/scripts/drbd-sb-tool.

Please Log in or Create an account to join the conversation.

  • Page:
  • 1