Exchange CCR replication fails during backup of Active node (4043501)

Date - 04/2008
Affected Product & Version - NetVault: Backup 8.x
Affected Module & Version - Exchange Server APM 3.1.8
OS Version - Windows 2003 - 64 bit
Application Information - MS Exchange Server 2007


Description:

Exchange 2007 Cluster Continuous Replication (CCR) fails when NetVault: Backup performs an ESE streaming backup
on the Active node of the cluster.

Symptoms:
Cluster Continuous Replication (CCR): is an Exchange 2007 new feature that allows for the asynchronous log shipping
and replay from an active Exchange 2007 Server cluster node to a passive node.
First a SEEDING phase happens were the .edb files and their logs are copied over.
Then the logs are continuously copied and replayed, but only after they are closed on the source.
Usually Microsoft recommends the use of VSS to backup the passive node, reducing I/O and backup requests on the active node.

However, tests from customers reported that anytime NetVault Backup would start an ESE streaming backup of these
replicated Storage Groups and their files, the replication service "Microsoft.Exchange.Cluster.ReplayService.exe"
crashes. The NetVault Backup itself carries on and finishes successfully.
The crash of Microsoft.Exchange.Cluster.ReplayService.exe causes the log shipping to be left far behind and will
require another seeding to go back to a stable situation.

Here is what the MS Exchange 2007 Application event logs show:
Event Type: Information

Event Source: ESE
Event Category: Logging/Recovery
Event ID: 223
Date: 3/17/2008
Time: 12:02:32 AM
User: N/A
Computer: MB2B
Description:
MSExchangeIS (16552) SG2: Starting the backup of log files (range C:\Logs-SG2\E020006A335.log - C:\Logs-SG2\E020006A95B.log).


Event Type: Error
Event Source: ESE
Event Category: General
Event ID: 490
Date: 3/17/2008