Change tracking failure

[davis157]

We have a Server 2012 R2 Hyper-V cluster with an EqualLogic iSCSI SAN. Recently, VM backups with application aware processing enabled that target the repository started completing with the warning, "Failed to flush change tracking data. Using full scan incremental run." We weren't running the most recent version of B&R, so we applied the newest update (so we're now at v9.5.0.1038), but it hasn't helped. I found some old discussions that indicated rebooting your cluster nodes would address this, but we recently took care of that through normal patching schedules, and it hasn't helped.

Does anyone have any ideas?

Thanks,
Mike

[nmdange]

I have only seen this happen consistently when you have at least one node in the cluster in a paused state.

[davis157]

Yes, that is the case here. But the paused node has been in that state much longer than when these warnings started to show up. Should I infer that if the node is resumed, these messages should go away? Is there a way to get Veeam to "ignore" this?

Thanks

[foggy]

Veeam B&R cannot trust CBT data in case the node is paused, hence the described behavior.

[davis157]

OK, it turned out the node wasn't paused, but there was a connectivity problem as far as Veeam was concerned. We resolved that, and backups since then are finishing normally.

Thanks