[FIXES] v6 Known Issues and Patches

[Gostev]

Patch #1

Issue 1
Direct SAN Access processing mode does not work for certain SAN makes and models.

Issue 2
Virtual Appliance (Hot Add) processing mode does not work when backup proxy is installed on localized version of Microsoft Windows.

Issue 3
Upgrade related issues affecting jobs created in v5 - poor incremenal backup performance, and large VIB file created on the first incremental run.

B&R 6.0 Patch 1 addressing all issues above this line is available for download starting December 10, 2011 (superseded by Patch 2)

[Gostev]

Issue 6
Guest file system indexing fail on certain VMs with the "Win32 error:The handle is invalid" error.

Issue 8
For v6 backup servers that were upgraded from v5, jobs fail to send email notifications if SMTP server requires authentication.

Issue 9
Reversed incremental backup mode jobs scheduled to run on specific days only will also start on synthetic full backup day (as set under incremental backup mode options in the advanced job settings). Due to the same bug, Replication and VM Copy jobs will always start on Saturday (which is the default day for synthetic full backup).

Issue 10
For jobs writing to CIFS-based backup repositories, backup results are not set in the selected VM attribute upon job completion.

Issue 11
Added support for Cisco Nexus 1000V switch (port groups were not visible in UI).

Issue 13
Added VM heartbeat check for Migration jobs. Original VM files will not be removed until the heartbit is received. This was needed because it was reported that under certain circumstances (we were not able to reproduce this to date), the migration job fails to copy the virtual disk contents correctly, but nevertheless deletes the original.

Issue 15
Under certain circumstances, the job fails with the following error: "Client error: Cannot allocate memory for an array. Array size: [4198400]."

Issue 16
For VMs configured to use VMware DVS (Distributed Virtual Switch), restore and replication into a different vCenter server will fail.

B&R 6.0 Patch 2 addressing all issues above this line (including Patch 1 issues) is available for download starting December 27, 2011 (superseded by Patch 3)

[Gostev]

Issue 5
Application-aware processing fails for SQL server where Veeam configuration database is located.

Issue 12
Unable to specify "0" in any octet of IP address when configuring re-IP rules of replication jobs.

Issue 17
Hot add operation can sometimes cause a very long backup proxy VM stun (VMware issue). The VM stun can be so long that it can cause all jobs using this backup proxy to fail due to timeout with the "Error: Client error: A connection attempt failed because the connected party did not properly respond after a period of time" error (follow this link for more details). To work around this, Veeam B&R will now automatically disable changed block tracking on all backup proxy VMs.

Issue 18
For VMware VMs with multiple disks, CBT may fail on virtual disks other than Hard Disk 1 with the "Soap fault. A specified parameter was not correct" error when the hot add processing mode is used. Follow this link for more details on the issue.

Issue 19
Backup console UI freezes periodically (upon automatic refresh event) when there is a large number of managed objects exist.

Issue 20
Any job processing the same VM longer than 48 hours is forcibly terminated. The new timeout is set to 7 days.

Issue 21
Very large backup jobs start failing with the "Failed to save file content" error after the VBM metadata file reaches 4 MB in size.

Issue 22
Hyper-V CSV ownership change during backup or replication preparation process may take too long with certain storage devices, causing bug check (BSOD) on Hyper-V hosts (Microsoft issue). Thanks to Jeff Dwyer (jldwyer1420) for incredible research and troubleshooting effort with this issue.

Issue 23
Hyper-V virtual disks with the disk file’s extension in capital letters (.VHD) are backed up like regular files, and not like virtual disks.

Issue 24
Copying very large files to Linux or ESX host with service console credentials specified may fail, resulting in copied file never appearing in the target folder despite File Copy jobs reporting success.

Other improvements
Patch 3 brings the following additional changes and enhancement requested by the community members:
• For VMware VMs with multiple disks, hot add now happens once for each processed VM instead of multiple times, before each processed disk, which slows down the VM processing. Thanks to Lucio Mazzi (Moebius) for bringing this to our attention.
• Improved vPower NFS performance by removing some excessive logging. Thanks to Ed Kok (E.Kok) for bringing this to our attention, and all the help and persistence chasing the root cause.
• Improved VMware replication failback performance by enhancing the virtual disk digest calculation process.
• By popular demand, the warning about outdated VMware Tools on the processed VM was changed to an information event.
• Removed confusing CBT warning from the initial run of backup and replication jobs which have backup and replica mapping configured.
• Added support for protecting VMs on Hyper-V hosts with Veeam B&R server installed in the parent partition. Previously, adding such hosts to the console would fail with the "User credentials cannot be used for local connections" error.
• Updated formulas for proxy and network processing stages statistics of the bottleneck analysis functionality. For example, heavy backup target load should no longer affect the network stage percent busy value.

B&R 6.0 Patch 3 addressing all issues above this line (including Patch 1 and Patch 2 issues) is available for download starting February 13, 2012 (KB1442)

[Gostev]

The following is the list of all other commonly encountered known v6 issues and workarounds.

Issue 4
In some cases, backup repositories automatically created by setup during upgrade from v5 may start writing data into C:\Windows\SysWOW64 if backup files used to reside in the root folder of the volume. The workaround is to re-create the backup repository manually.

Issue 7
Job fails with "Unknown network: not VMNetwork neither DVS" error. This is typically caused by an abandoned reference in the VMX file to a network adapter that is either not in use (not connected to any network), or has been previously removed. Clean up the VM configuration, and the issue should go away.

Issue 14
Processing VM with virtual disks located on datastore with ampersand in its name results in "Error: An error occurred while parsing EntityName". Current workaround is to rename the datastore.

[Gostev]

Concurrent tasks
There is one behavior changer that upgrading users should keep in mind. The default VMware backup proxy (local) is pre-configured for 2 concurrent tasks. If you don't deploy additional backup proxies, or increase this number (if server CPU allows), you will not be able to run more than 2 tasks in parallel - while you have been able previously. For newly added proxy servers, the maximum concurrent tasks amount is set automatically according to CPU capabilities by the wizard, however the product setup does not have CPU detection code, and just sets the number of tasks on local proxy to 2.

How to fix broken CBT
Not specifically related to Veeam v6, but good to keep in mind in time of vSphere 5 upgrades. Under certain circumstances, CBT data may corrupt in VMware on individual VMs, causing "Cannot use CBT" warning in the Veeam job log. For some customers this issue appeared on a few VMs with their vSphere 5 upgrade, for other it was triggered by the storage disaster, or just "came from nowhere". The below is the procedure that you can use to fully reset CBT data on a VM.

Code: Select all

1. Power off VM
2. In vSphere Client, select VM, Edit Settings, go to Options tab, select General under Advanced, click Configuration Parameters.
-- Set "ctkEnabled" to false.
-- Set "scsi0:X.ctkEnabled" to false for EACH virtual disk.
3. Power on VM.
4. Power off VM again (VM CBT data will reset in VMware CTK database at this point only).
5. Power on VM.
6. Run any Veeam job on that VM. The job will enable CTK will be automatically.
7. The following job runs will now be able to use CBT data.