Arcserve® Unified Data Protection Version 5.0 Update 4 Release Notes

1.0 Welcome

Welcome to the Arcserve UDP Version 5.0 Update 4 Release Notes. This Release Notes contains important information about this update. Updates are cumulative and can be installed on any previous version of Arcserve UDP.

2.0 Enhancements

The following enhancements or features have been added to Arcserve UDP for this update:

• VMware vSphere 6 is supported, but only when patch ESXi600-201505001 for ESXi 6.0 is applied. For more information, see VMware ESXi 6.0, Patch Release ESXi600-201505001 (2116125).
• Arcserve UDP has added a fix in Update 4 that allows you to use all the latest browsers (including Google Chrome v41, Firefox, and Internet Explorer) to browse the Arcserve UDP Console/RPS/Agent on Windows web pages if HTTPS is used.
• Arcserve UDP has been certified for Exchange Shared and Linked mailbox support.
• JRE support has been upgraded to 1.8 (Windows BMR is using 1.7.)
• VDDK support has been upgraded to 5.5.4.
• Arcserve UDP Version 5.0 Update 4 supports the Auto Update feature:
  • Arcserve UDP Version 5.0 Update 3 users can leverage auto update to upgrade to Arcserve UDP Version 5.0 Update 4 now or they can also manually upgrade.
  • If you are using versions prior to Update 3, you have two ways to upgrade to Update 4:
    • Manually upgrade to Update 3 first, and then check auto update to upgrade to Update 4.
    • Directly perform a manual upgrade to Update 4.
• When creating a new data store, the default value for “Concurrent streams to data store” has been changed from 20 to 4.
• Some security vulnerabilities have been fixed.
• Multiple Test Fixes have been installed. For more information, see Issues Fixed in the Release Notes.

3.0 Issues Fixed

This update fixes the following issues, which could have occurred:

• Includes patch RO79230: WIN-ARCSERVE UDP R5.0 UPDATE 3.
• Fixes a problem where virtual standby always creates a new virtual machine on a Hyper-V platform, if the Hyper-V server is a non- English system. (Test Fix T55Z087)
• Fixes a problem with Korea Exchange databases, when you try to dump mails to a folder, and the restore job crashes. (Test Fix T5LE208)
• Fixes a problem where after recovering a VMware VM, the MAC address of the recovered VM is automatically generated and so it is different from the original one. (Test Fix T5LE197)
• Fixes a recover VM UI time out error message when you give Hyper--V details to submit the recovery job. The recover VM job is not getting submitted when the datastore or copy recovery point sessions is accessed as a local path up to the parent folder level from the Agent UI. (Test Fix T55Z085)
• Fixes a problem where a virtual standby job failed if one or more folders existed between the data center and the ESX server in a VMware managed object hierarchy. (Test Fix T5LE199)
• Fixes the message "Failed to communicate with deduplication %1!s! on server %2!s!. Network error = [%3!s!]." started by a new line when performing the following command: “cmdutil /db /dump /rps Log result”. (Test Fix T5LE200)
• Fixes a problem where a catalog was not generated after the backup finished, when you configured email alerts and the mail server has a problem. (Test Fix T5LE202/T5LE235)
• Fixes a backup failure of the Hyper-V VM if you have 1 MB block size disk attached to the VM. (Test Fix T5LE204/T5LE205)
• Fixes a license problem that all the VMs are showing under one ESXi host under the Hypervisor tab on the console and not the correct ESXi host. (Test Fix T5LE206)
• Fixes a problem that when restoring one or more fully selected mailboxes, it crashes when the memory is not enough. (Test Fix T5LE207)
• Fixes a problem with the BMR media process creator, where it incorrectly shows German and Deutsch on the language selection screen. (Test Fix T5LE209)
• Fixes a problem where the latest browser cannot browse Arcserve UDP Console/RPS/Agent for Windows Web Pages if HTTPS is used, for example Google Chrome v41+, Firefox 36/37+, and the latest version of Internet Explorer. (Test Fix T5LE224)

  For more information, see the KB Article https://arcserve.zendesk.com/hc/en-us/articles/204542275

• Fixes a problem that fails to delete a file that causes the merge job to fail. This problem occurred on some machines with the backup destination in a NAS device. (Test Fix T245948)
• Fixes the problem that when clicking Test Connection in the cloud configuration window displays “Failed to communicate with the cloud” message. (Test Fix T5LE203/RO79422)
• Fixes the problem where volume information is not shown in the restore UI, as information is missed in the adrconfigure.xml file after a host-based backup. The VM parser only treats the “Active” partitions as candidates of the system volume. As a result, it fails to find the system volume and all the system information (Test Fix T5LE212/T5LE214).
• Fixes the problem when in an agentless backup, if you set the command "before the start of a backup" to stop a service (for example, net stop "service xyz"), the command "following the acquisition of the snapshot" is not performed if there were snapshot acquisition errors (net start "service xyz"). (Test Fix T5LE213)
• Fixes the problem where a granular restore catalog fails to be created when an Exchange DB, log, and system files in the mount point folder are for different volumes. (Test Fix T5LE215)
• Fixes the problem that if you set the filter as the last 1 day, the report will show count of jobs which ran in the last 24 hours, not jobs run from 00:00 AM of the last day. (Test Fix T5LE233)
• Fixes the problem where the bandwidth throttle does not work for agentless VM backups. (Test Fix T5LE219)
• Fixes the problem where a makeup job is not working for agentless scheduled backup jobs. (Test Fix T5LE221/T5LE234)
• During a backup, the message "as-hbbu-vmwarebackup DeploySuccess.sig was not found" displays from the VI client and is caused by Arcserve UDP detecting the snapshot method. If the file C:\as-hbbu-vmwarebackup\DeploySuccess.sig exists inside the guest OS, the "VSS" snapshot method is detected; and if not, then the "VMware Tools" snapshot method is assumed. (Test Fix T5LE218)
• Fixes the problem where incremental backups are getting converted to verify backups, when the volume has more than 1 volume GUIDs. (Test Fix T5LE211)
• Fixes a problem that the Arcserve UDP Agent service cannot start during DST. (Test Fix T5LE232)
• Fixes an issue where the file system catalog job crashes and 0 size dump files are generated in the logs folder when there are many deep directories. (Test Fix T5LE227)
• Fixes a problem that the compress attribute for files and folders is changed by the restore job (Test Fix T5LE216)
• Fixes the problem where a virtual standby job fails because it cannot find/locate a VMware data center. (Test Fix T5LE199)
• Fixes an issue that the agentless backup jobs cannot be cancelled when they are in the waiting queue. (Test Fix T5LE236)
• Fixes an issue that an incremental backup of a disk is marked as full, but actually the data is still incremental, which may cause data corruption. (Test Fix T245946)
• Fixes an issue that characters are corrupted on the Japanese version and unreadable in the boot menu when starting a machine which is recovered by BMR. (Test Fix T245958)
• Fixes an issue that a SQL Database’s transaction log cannot be truncated and shrunk when the database is under FULL mode and a full database backup has been performed (Test Fix T36A881)

• Fixes a problem that some unused AVHD/AVHDX files occupied too much disk space after the next snapshot was taken. (Test Fix T5LE240)
• Fixes a problem that when two parallel CRP jobs for two different VMs (under the same proxy) are running, then the second CRP job can create new session instead of overwriting the previous session. (Test Fix T5LE226)
• Fixes an issue that the agentless backup jobs show the wrong backup type (Full/Verfy/Incremental) when they are in the waiting queue. (Test Fix T5LE229)
• Implements an enhancement that an agentless backup job will abort when all virtual disks of a VM are excluded from the backup due to certain reasons (for example, the virtual disk is an independent disk). (Test Fix T5LE228)
• Fixes a problem on Windows 7/8, if Bitlocker is enabled, after rebooting, a verify backup will run instead of an incremental backup. Sophos Safe Guard has the same problem as Windows Bitlocker. The impact only affects Arcserve UDP Agent, when full volume encryption is enabled on one or more volumes, such as Windows Bitlocker or Sopho Safe Guard. (Test Fix T5LE225)

  The Changed Block Tracking (CBT) driver (ARCFlashVolDrv.sys) is sitting below the Bitlocker full volume encryption driver (fvevol.sys). When saving CBT data to disk when the machine is shutting down, the data is saved as not encrypted because the Bitlocker driver is not loaded. While the system starts up, the driver tries to read data from the CBT data file, and it is not able to get the correct data.

  The resolution is to move the CBT driver to above the Bitlocker driver in the driver stack, so when the CBT driver saves data to disk it will be encrypted by the Bitlocker driver properly. After you apply the fix, restart your computer and run a full backup.

  To Hot Apply the fix:

  1. Open a command line window, and switch to folder “C:\Program Files\CA\arcserve Unified Data Protection\Engine\BIN\DRIVER”.

  2. Run command “ARCFlashVolDrvINSTALL.exe –p –output=.\fix_fve.log”

• Fixes a problem that the replication network throttling does not work when the replication jobs are launched at the same time, for the multiple nodes protected by the single plan with a very small replication bandwidth setting. (Test Fix T5LE238)
• Fixes vulnerabilities that unauthorized users can download and delete files which are on the Arcserve UDP Console’s local disk. (Test Fix T5LE230/T5LE231)
• Fixes a problem where a 0 size dump file is created after each backup, when there are many deep paths on the volume.(Test Fix T5LE239)
• Fixes a problem that if the Exchange 2013 database is too large, the catalog job will crash when there is not enough memory for the job to load the database. (Test Fix T5LE241)
• Fixes a problem that restoring more than two mail objects or restoring the folder or mailbox to the original/alternate location for Exchange 2013 crashes. (Test Fix T5LE217)
• Fixes a problem that Recovery Point Check fails due to unknown problem. (Test Fix T5LE244)
• Fixes the problem that if a SSD is used for a deduplicated hash database, when it is full, the subsequent merges will also fail. Therefore, there is no opportunity to purge the data to reclaim the space. (Test Fix T5LE242/T5LE243)
• Fixes a problem that if a purging thread is running to update hash keys, when the hash roles receive the ‘exit’ signal, it is possible that it could not complete the ongoing operation within the 1 minute timeout period, if the system is very busy. As a result, the incorrect hash key number might be recorded into the hash database header file. Then it is possible that the hash role encounters a fatal error during the next startup.(Test Fix T5LE242/T5LE243)
• Fixes a problem that when rebuilding the hash database tool, it does not handle the high frequency blocks well. When the reference number is more than 65535, it might generate the incorrect reference number. After importing the new hash database, it might cause this kind of blocks to be incorrectly deleted, even though some recovery points are still linking to them.(Test Fix T5LE242/T5LE243)
• Fixes an issue where after BMR, the server cannot boot with various errors. When checking files under Windows\system32, some files are corrupted. This usually happens when Windows updates is downloading/installing along with the Arcserve UDP Agent backup job. After the fix, a full or verify backup needs to be performed manually.
• Moves the tool location from C:\ to C:\Program Files\Arcserve, which is deployed to the VM when using VSS snapshot.
• The Apache HTTP Server used by Arcserve UDP makes use of OpenSSL 1.0.1j that is affected by vulnerability CVE-2015-1637, also known as “Factoring attack on RSA-EXPORT Keyes” or FREAK vulnerability. For a detailed description of this vulnerability, see https://web.nvd.nist.gov/view/vuln/detail?vulnId=CVE-2015-1637 and https://technet.microsoft.com/library/security/ms15-031.

  To avoid this vulnerability, Arcserve UDP has disabled the use of vulnerable ciphers and switched to using the following safe ciphers:

  • ECDHE-RSA-AES128-GCM-SHA256
  • ECDHE-RSA-AES256-GCM-SHA384
  • ECDHE-RSA-AES128-SHA256
  • DHE-RSA-AES256-SHA256
• Fixes a problem that some Completed Jobs display if Group Jobs by Plan is checked when selecting the In-Process panel in the dashboard view.
• Fixes the subsequent replication failure that occurs when the previous replication failed or was cancelled, and the backup job changed to a new data store as the backup destination (which is the replication source).
• Fixes the problem when a merge job tries to replace the original index file with a new index file at some phase of the job. If the merge job happens to fail at that time, the index file will be in the intermediate state. If you try to purge the specific node at that time, the purge job will fail. It also will cause all recovery points of the node to fall into an abnormal state.The fix imports a rollback mechanism for purging the node job. Once the purge node job detects this kind of recovery point, it will first roll back these recovery points to a normal state and then execute the purge operation. Note that if one node has already run into the issue, the fix will not help.
• Adds a switch to fix the problem where you want to keep the file system catalog for faster restore file/folder browsing, but do not want to use the catalog to do a file/folder restore, since it is slow when restoring a small file from a large volume. To enable this fix, you need to enable a switch in the Windows registry. If you want to restore a few small files, you can set the following registry to force the restore job to mount a recovery point and restore files from the mounted volume, instead of using the file system catalog.

  [HKEY_LOCAL_MACHINE\SOFTWARE\CA\ARCserve Unified Data Protection\Engine\AFRestoreDll]

  "RestoreCtrl"=dword:00000001

• Fixes the problem with Linux nodes, where if you shut down the target machine when the restore file job is running, the restore file job hangs.
• Fixes a problem that the managed capacity license sometimes shows more than the manage capacity report.
• Fixes a problem where some system files are not released after the backup job is completed.
• Fixes an issue that for uEFI systems, after BMR the order of partitions are not the same as the original system.
• Fixes an issue that when inserting a driver into the original OS, it may fail during BMR recovery, and may cause the server not to boot after the BMR.
• Fixes a problem where the tape backup job of a Arcserve UDP session fails to release resources and subsequent merge jobs may fail, unless the data store is restarted. A system error 32 is returned.
• Fixes a problem where the GUI hangs, when importing nodes from ESX, and you select nodes with a filter, unselect them by removing the filter, and then click Add to List.
• Fixes the problem where a merge on the RPS job monitor is not removed in the dashboard even after the merge job is finished.
• Fixes a problem where VSB did not support 3rd party virtual distributed switch, such as CISCO VDS 1000.
• Fixes a problem where VSB can not boot for a CHS environment.
• Fixes a problem that the translation in German could be misleading in the Replication Task name.
• Fixes a problem that a makeup job will run every time the Arcserve UDP Agent service is rebooted, if the start time for the scheduled backup time is set to a future time.
• Fixes a problem on the date that DST starts, where the recovery points in the last time range of the days before DST start date could not be displayed.
• Fixes a problem that search in the AD restore UI is case sensitive, and should be case insensitive.
• Fixes the problem of failing to restore the user mailbox for Microsoft Exchange2007, when the mailbox user name is different from the display name.
• Fixes the problem where replication is easy to fail, if a command takes a long time to finish due to an unstable network over the WAN and low performance of disk, cpu, or memory.
• Fixes a problem where the following dashboards may not display data in date order.
  • Raw Data and Actual Data Storage
  • Restorable Data and Actual Data Storage
  • Actual Data Storage
• Fixes a problem that fails to purge a node in the deduplication data store, where the node has previously been merged.
• Fixes a problem where with a certain disk/volume configuration, the backup fails at reading data from the volume shadow copy, with the following Windows system error 27:

  "The drive cannot find the sector requested."

• Fixes the problem where no mail alerts are received for missed backup jobs, even though they are configured.

For all issues fixed in this release, see the Arcserve UDP Version 5.0 Release Notes.

4.0 Known Issues

The following issues could exist in this update:

• When updating from Arcserve UDP Version 5.0 or Arcserve UDP Version 5.0 Update 1 to Arcserve UDP Version 5.0 Update 4, a system reboot is required. When updating from Arcserve UDP Version 5.0 Update 2 or Arcserve UDP Version 5.0 Update 3 to Arcserve UDP Version 5.0 Update 4, a system reboot is not required.
• Arcserve has been tracking, with growing concern, an issue with a VMware bug that has not been resolved. This defect might cause backup products to back up a corrupted snapshot. The impact of this problem is such that it could render the backups completely unusable for data retrieval to varying degrees of data loss. It was found that agentless backup of VMware systems using Arcserve UDP encountered this problem.

  Symptom:

  VMware has a bug when quiescing a VM so that its snapshot contains corrupted data. The backup reads data from the snapshot so that the data backed up also becomes corrupted. For more information about this issue, see the following VMware KB article: Application quiescing with Windows 2008 R2 SP1 and Windows 2012 with vSphere Data Protection, VMware Data Recovery, and third-party backup software (2044169).

  Note: This problem can occur with all VMware ESXi versions and on a VM with guest OS Windows 2008 R2 SP1 and Windows 2012. The data corruption problem cannot be detected by the software because VMware does not return an error in this case. This may lead to the situation that you are not aware of the problem until you try to restore some data.

  Solution:

  Perform the following methods provided in this update to detect and resolve the problem:

  • Recovery Point Check - Mount the volumes from the recovery point and verify data consistency by running the Microsoft tool CHKDSK at the end of the backup job. For more information, see the Arcserve UDP Solutions Guide. Since running CHKDSK may take a long time, it is suggested to enable this option for weekly or monthly backup jobs.
  • VMware Snapshot Quiescing Method - In case the Recovery Point Check detects a problem, perform the steps described in "How to check if you may be affected by this bug" in the Arcserve KB article. If it is confirmed that the problem is caused by this VMware known issue, you can use the Microsoft VSS inside the VM option to resolve the problem before VMware can solve the issue. For more information, see the Arcserve UDP Solutions Guide.
• A machine may fail to boot after performing a BMR, when the machine has a 4 KB disk and the source is Windows 8.1 or Windows 2012 R2.

  Symptom:

  When the source machine is a Windows 8.1 or Windows 2012 R2 system and you perform a BMR on a machine with a 4 KB disk, it may fail to boot the machine after BMR with error message: system_thread_excption_not_handled (WppRecorder.sys).

  Solution:

  Boot to the recovery console (you should automatically get there after a few Blue Screen crashes).

  1. Launch the command prompt and run the following command:

     c:\windows\system32\compact.exe /U c:\windows\system32\drivers\*.sys

  2. Reboot the machine.
  3. As soon as you successfully boot the machine, disable NTFS compression system-wide so that the CBS Scavenger does not reintroduce the issue again:

     fsutil behavior set DisableCompression 1

  4. Reboot again (so the DisableCompression setting takes effect). Files that are compressed will stay compressed (and readable). However, CBS Scavenger will be unable to compress the files and cause the issue.
• When migrating recovery points from a shared folder to a datastore using the Arcserve UDP Console, the migration job monitor/job history is not available from the Nodes list view on the resources tab.

  Symptom:

  A migration job does not show in the Nodes list view on the resources tab.

  Solution:

  Access the RPS view or jobs tab to check the job monitor.

• The Arcserve UDP Recovery Point View does not support the volumes which are created by Map network drive.

  Symptom:

  Cannot browse volumes in the Arcserve UDP Recovery Point View.

  Solution:

  Use a UNC path directly instead of mapping to a network drive.

• For host-based backup Exchange Granular Restore, when the Windows version of the backup proxy host is an earlier version than the virtual machine, the subsequent Exchange catalog job may fail due to a Windows update. When a virtual machine is backed up with the Generate Exchange Catalogs option enabled for the plan, the catalog job fails with following activity log message:

  The Windows version of proxy [<Proxy Name>] is an earlier version than virtual machine [<Virtual Machine Name>]. As a result, the subsequent Exchange catalog job may fail, and you will need to install the related Windows update package to resolve the problem."

  Symptom:

  The Exchange catalog job will use the Exchange binaries from the virtual machine. If the Windows version of the proxy machine is an earlier version than the virtual machine, the Exchange binaries do not work well, and as a result, the catalog job will fail.

  Solution:

  Install the following Windows update package to resolve the problem:

  http://support.microsoft.com/kb/2533623

• After recovering a clustered VM to the original Hyper-V cluster, the network adapter could not connect to the virtual switch with a warning message "Unable to connect the network adapter <<adapter name>> to the virtual switch." in the activity log.

  Symptom:

  Failover happens for the clustered VM before taking a snapshot when backing up a VM. This causes the Hyper-V host recorded in the backup session to be inconsistent with the VM configuration.

  Solution:

  You can manually connect the network adapter to a virtual switch on its Hyper-V host or use the Restore to an alternative location option to recover the VM with which you can set the restore configuration for the VM.

• In Arcserve UDP Agent (Linux), a new plan is created on the Console when remigrating a Linux Backup Server to the Console.

  Symptom:

  Suppose you migrate a Linux Backup Server to the Arcserve UDP Console. If you release the Linux Backup Server and then migrate it again, you will see a new plan is automatically created on the Console.

  Example:

  You have a Linux Backup Server1 that manages Linux_Node1. You have created a backup job named New Plan. You also have a Arcserve UDP Console that manages Linux_Node2 and Linux Backup Server2 and the plan is also named New Plan.

  You migrate Linux Backup Server1 to Console using the d2dreg command. After the migration, the backup job name changes to New Plan_<Linux Backup Server1>. Deploy New Plan_<Linux Backup Server1>.

  Now you add Linux_Node1 to New Plan and add Linux Backup Server1 as the server and deploy the plan. New Plan now protects Linux_Node1 and Linux_Node2 and the server is Linux Backup Server1.

  Now you release Linux Backup Server1 from Console. Delete New Plan and New Plan_< Linux Backup Server1> from Console. When you migrate Linux Backup Server1 again to Console, two plans get migrated: New Plan and New Plan_<Linux Backup Server1>.

  Solution:

  Delete the new plan. The new plan does not have any nodes, therefore, it does not affect any backup schedule.

• In Hyper-V 2012 or later, the virtual machine stays in the “Backing up” status although the agent-less host-based backup job of this virtual machine has already finished.

  Symptom:

  Although the backup job of one virtual machine has already finished, the virtual machine's status is still “Backup up” in the Hyper-V Manager. Therefore, if another backup job for this VM starts at this time, it will fail with error “The Hyper-V VSS writer has encountered an error when processing this virtual machine”. In addition, you cannot perform some operations, such as power on/off for the VM at that time in Hyper-V manager. And if the VM is Hyper-V cluster, you cannot perform live migration for it.

  This problem happens during the following situations:

  • There are several backup jobs starting at the same time or at times close to each other (within 1 minute).

  • One or more backup jobs finished, but there is still at least one backup job which is still in progress.

  Root of the problem:

  To avoid bringing unnecessary workload to the Hyper-V host, instead of taking one VSS snapshot for each VM, Arcserve UDP tries to take one VSS snapshot for all VMs if their backup jobs start at the same time or at times close to each other. After the VSS snapshot is taken, all the VMs inside this VSS snapshot instance will be “locked” (in Backing up status). Because Arcserve UDP cannot release the snapshot until all backup jobs are finished, even if the backup job of a VM already completed, that VM is still “locked”. Due to the limitation of VSS snapshot that only one snapshot can be taken for one VM at one time, if another backup job of the same VM starts at this time, it will fail with the error “The Hyper-V VSS writer has encountered an error when processing this virtual machine”. In addition, some operations (such as power on/off) are disabled in Hyper-V Manager and, if the VM is in a Hyper-V cluster, live migration is also not allowed. This does not happen in Hyper-V 2008R2 because Hyper-V 2008R2 has different behavior on the VSS snapshot mechanism.

  Solution:

  While the VM is “locked”, you can still use the guest OS as normal. Therefore, this has no impact on the usage/availability of the guest OS. However, if you have concerns and want to avoid this situation, you can do either of the following:

  • Use the Hyper-V Snapshot Separation option on the resources tab of agent-less host-based backup. For more information, see the Arcserve UDP Solutions Guide.
  • Use different plans to protect VMs which have a different size. In other words, include the VM with a similar size in one plan so that those backup jobs take a similar amount of time and, at the same time, set different schedules for those plans.
• Sometimes, the UI may respond slowly when accessing Arcserve UDP Agent (Windows) from certain versions of Internet Explorer 10 or 11.

  Symptom:

  The UI may respond slowly when accessing Arcserve UDP Agent (Windows) from Internet Explorer 10 or 11, if you are using any of the specific versions of Internet Explorer 10.0.9200.17XXX and Internet Explorer 11.0.9600.17XXX.

  If this issue occurs, all other subsequent requests will wait for response for 5 minutes.

  This issue only happens when using any of the versions of Internet Explorer 10 and 11 mentioned. All other browsers do not have this issue.

  Solution:

  Perform one of the following temporary solutions:

  • Close Internet Explorer which responses slowly. Open any other browser, such as Chrome or Firefox to log in.
  • Clean the cache of Internet Explorer and then refresh the page.

  Perform the following permanent solution:

  • Add the following two lines to the file “\Arcserve Unified Data Protection\Engine\Apache\conf\httpd.conf”. Then restart the Arcserve UDP RPS Port Sharing Service.
    • AcceptFilter http none
    • AcceptFilter https none

• If the appliance is joined on the domain, then the DHCP server on the appliance needs to be authorized. If your domain account does not have the rights to authorize the DHCP server, you will see a warning message in the Arcserve UDP configuration command window.

  Solution:

  Contact your domain administrator for help to authorize the DHCP server before using the Arcserve UDP Linux Backup server.

• The Windows operating system does not respond.

  Symptom:

  There is a dead lock in the file system while taking a snapshot when the granular restore catalog job is running.

  The following conditions may trigger the problem:

  • The cache file of the mounting driver used to store the data is on the volume which is protected by the Arcserve UDP Engine.
  • If you perform a backup when the granular restore catalog job is running.

  Solution:

  Configure the following registry key to move the cache file to another location which is not protected by the Arcserve UDP Engine:

  HKEY_LOCAL_MACHINE\SOFTWARE\CA\ARCserve Unified Data Protection\Engine\AFStorHBAMgmt\CacheFilePath

  If the "AFStorHBAMgmt" key does not exists, you have to create a new one.

  1. Under the "AFStorHBAMgmt" key folder, create a string value REG_SZ for "CacheFilePath".
  2. Set the value to a folder on a different volume that has sufficient free space.
     • If the destination is on a local machine and no registry path is specifically set, then the destination session path is used to store the cache path.

       Example: If your backup destination is local drive E:, the backup snapshot will be created on Drive E: into the backup destination path.

     • If the destination is a remote/local path and if the registry is set as HKEY_LOCAL_MACHINE\SOFTWARE\CA\ARCserve Unified Data Protection\Engine\AFStorHBAMgmt\CacheFilePath at some path

       Example: E:\temp then E:\temp is used as cache store path.

     • If the destination is remote and you do not set any registry path, then by default (Installation folder)/Bin/Temp is used to store the cache path.

  Note: This issue happens when the I/O load is high, and the following conditions exist:

  • A granular restore catalog job is running for a large database. The recovery point will be mounted as writable mode, the data written, and the mounted volume will be save to the Arcserve UDP install folder by default.
  • The backup is taking a snapshot of the volume where Arcserve UDP is installed.

  Taking a snapshot will notify the file system driver to flush incomplete data to the volume, and will hold off all new write operations to all volumes for a short time. This will create a dead lock where the mounted volume is flushing data to the file, and at the same time the writer operation is holding off on the volume where Arcserve UDP is installed.

• During granular recovery of the Exchange mail objects, if the Exchange server has too many databases and you try to scroll the slide bar several times to search and select the desired database, some of the database names may not display in the Mailbox Database column.

  Solution:

  Reopen the Restore wizard and try selecting the database again.

  Symptom:

• For Exchange2007, if mail is sent by a user having “on behalf of” privileges for a shared mailbox, after the restore, the mail’s "from" field only displays the shared mailbox's name.
• When you restore files from RHEL 5.x or CentOS 5.x recovery points using RHEL 6.x or CentOS 6.x Arcserve D2D Server, some SELinux (Security-Enhanced Linux) attributes may not get restored and the restore job result may be “failed” if you cancel it.

For all known issues in this release, see the Arcserve UDP Version 5.0 Release Notes.

5.0 Considerations/Limitations

The following considerations could exist in this update:

• Updates are cumulative and can be installed on any previous version of Arcserve UDP.
• When you upgrade to Arcserve UDP Agent (Linux) from Version 5.0, Version 5.0 Update 1, or Version 5.0 Update 2 to Version 5.0 Update 4, the next scheduled backup is automatically converted to a Verify backup.
• The CentOS-6.6. file is governed by the GPL terms located at http://mirror.centos.org/centos/6.6/os/x86_64/GPL. For a period of three (3) years, you may receive upon request a complete machine-readable copy of the corresponding source code.
• The following applies, if you are using Arcserve UDP Version 5.0 Update 2 and would like to upgrade to Update 4. In Arcserve UDP Version 5.0 Update 2, when an agentless backup is performed on a Hyper-V VM, the VM is placed in an offline "saved state" while the snapshot is taken for the VM (in case the VM does not support an online backup). After upgrading to Arcserve UDP Version 5.0 Update 4, the default behavior while in this saved state is changed. As a result you should consider the following for Hyper-V agentless backups:
  • In addition to the new option to select the Hyper-V Snapshot Method, Arcserve UDP Version 5.0 Update 4 also changes the default behavior for when an application consistent snapshot is denied, for both existing Arcserve UDP agentless plans as well as newly added ones.
  • This change to the default behavior while in the saved state is designed to help you make an informed decision about when to perform the backup, based upon either having a short downtime (few seconds to 2 minutes) or a possible lack of data consistency.
  • If after the upgrade, backup jobs are failing that were previously not failing, and you want to change the saved state behavior to the way it was in Update 2, you can modify the plan and allow for saved state backups.
  • Review the VSS errors in the event log of the Hyper-V host machine, to fully understand and resolve the circumstances and enable fully VSS-supported online snapshots.

The following limitations could exist in this update:

• Catalog, file restore, BMR, and mounting functions will be affected for Windows agentless VM nodes when the Guest Operating System setting on the Virtual Machine Properties dialog is incorrectly set to non-Windows (for example, Linux or Other). After the agentless backup completes successfully, the catalog job script will not be created and the catalog job will not be launched, even though in the agentless backup plan the enable catalog option has been configured properly. To resolve this problem, set the Guest Operating System setting to Windows.

  Note: This limitation exists only for VMware VMs, which has the option to set the OS type. For Hyper-V, this problem does not exist.

• Microsoft SQL Server 2014 AAG support with BMR is not available at this time.
• Virtual Standby supports Hyper-V clusters with limitations. You can perform a Virtual Standby of any full node backup, for example, Agent based or Agentless (VMware or Hyper-V). Virtual Standby supports both. You can also perform a Virtual Standby to a Hyper-V cluster with a CSV with a limitation that the standby VM will not be a failover resource.
• Exchange Granular Restore does not support restoring the mailbox which is stored in the parent domain but it's account is in the child domain. The restore job would fail.
• If more than two mail objects are in the same conversation, after restoring them, they are not displayed in the same conversation when viewed in the Sent folder. Everything works fine when viewed in the Inbox folder.
• Microsoft Exchange Granular Restore Browse UI does not display any mail under the mailbox for host-based backup sessions if the protected VM is Microsoft Exchange Server 2007 and the Backup Proxy Server is Windows 2012 or Windows 2012 R2. To resolve this problem, use a 64 bit Windows 2008 R2 or lower Windows Operating System as the Backup Proxy Server.
• When a Linux backup server runs in the Arcserve UDP appliance, a PXE-based BMR is not supported.
• For an agentless backup for a hyper-V VM with replication enabled:
  • An Arcserve UDP agentless backup supports backing up the VM from the Master (primary) site.
  • An Arcserve UDP agentless back up from the Replica site has no guarantee of data consistency and accuracy, and the completion of the back up.
• The Arcserve UDP 7000 series Appliance Wizard cannot be launched when the system boots up if you log in to the appliance with a domain user account.

  Symptom:

  When you log in to the Arcserve UDP appliance with a domain account, the appliance wizard is not immediately launched. A User Account Control dialog displays asking you to confirm allowing the program to make changes to the computer.

  Solution:

  From the User Account Control dialog, click Yes and the wizard will be launched.

• Virtual standby does not support a cluster enabled ESX server. If the destination ESX host is down and the standby VM is moved to other ESX nodes, the job fails since it could not connect to the original ESX server.
• Arcserve UDP Version 5.0 Update 4 does not support a host-based backup Exchange catalog job if the Exchange database path, system files path, or log path is in a mount point.
• After upgrading to Arcserve UDP Version 5.0 Update 4, the first Hyper-V VM backup will include the full content of the disk (even if it is an Incremental Backup), and may also display the following log message:

  "The Change Block Tracking(CBT) feature has been upgraded. As a result, redundant data may be backed up."

• The following limitation is caused by Hyper-V, and not Arcserve UDP. Any agentless solution will have this limitation.

  From Windows 2012, there is a new “shared VHD” feature of Hyper-V, which allows you to share the VHD(X) file among multiple VMs. (For more information about this feature, refer to the following blog: Shared VHDX Files - My Favorite New Feature in Windows Server 2012 R2). However, Hyper-V VSS Writer does not support this new feature. As a result, a Hyper-V agentless backup to a VM containing this “shared VHD” feature will fail. In order to back up this type of VM, you will need to run an agent backup by installing Arcserve UDP Agent for Windows on the VM.

For all considerations or limitations in this release, see the Arcserve UDP Version 5.0 Release Notes.

6.0 Installation

The following sections provide information about installation prerequisites and installation instructions.

6.1 Prerequisite Tasks

Consider the following prerequisite tasks before installing this update:

• Before you install this update, verify that Arcserve UDP Version 5.0, Arcserve UDP Version 5.0 Update 1, Arcserve UDP Version 5.0 Update 2, or Arcserve UDP Version 5.0 Update 3 is installed.
• Apply patch ESXi600-201505001 for ESXi 6.0 to support VMware vSphere 6. For more information, see VMware ESXi 6.0, Patch Release ESXi600-201505001 (2116125).

6.2 Installation Instructions

When installing the Arcserve UDP update or the Arcserve UDP Agent (Windows) update, it is important to maintain optimal performance between the Console, the Recovery Point Server (RPS), and the Agents. As a result, when the update is installed in an environment that contains both a Console and an Agent, you must always install the update on the Console first, and then on the RPS, and finally on the Agent. (For the Agent that is installed on the Console or the RPS, the update will be automatically installed on that Agent at the same time).

Before you upgrade to Arcserve UDP Version 5.0 Update 4, review the following backward compatibility considerations in your environment:

• Arcserve UDP works best when all components (RPS + Agents) managed by the same Console are on the same update level as that Console. For that reason, administrators should plan an upgrade of all components, otherwise, it could delay a change to a plan when it is required later on. Arcserve UDP agents and managed RPS upgrades can be started remotely or scheduled from the Arcserve UDP Console, without having to log into the production server node itself.
• The Console of Arcserve UDP Version 5.0 Update 4 will continue to monitor and/or start backups and restores with Arcserve UDP agents on a previous version of Arcserve UDP.
• Freshly adding an agent or an RPS with a previous version of Arcserve UDP to the Console will automatically ask you to upgrade o Update 4.
• After an RPS has been upgraded to Update 4, it can continue to receive backups from an Arcserve UDP agent of a previous version.
• Any change in the Arcserve UDP Plan or in the Agent Node credentials will prompt for a remote-upgrade of the Arcserve UDP agent. An upgrade from Arcserve UDP Version 5.0 Update 2 or Update 3 to Update 4 should not require a Windows Reboot.
• Replication between an RPS on Update 4 and another one on Update 3 is supported, as is the other way around. Although it is encouraged to perform an upgrade of every RPS to the latest version at your earliest convenience.

Note: If you have an environment with both Arcserve UDP and Arcserve Backup installed, and you want to install Arcserve UDP Version 5.0 Update 4, you need to also install the corresponding Arcserve Backup patch RO75131 at the same time.

Note: For appliance users who upgrade to Arcserve UDP Version 5.0 Update 4, the IE browser will show a warning when you launch the Arcserve UDP Console, which is caused by the certification upgrade in Update 4. You can choose Continue to this website (not recommended) to launch the Arcserve UDP Console. You can also resolve this permanently by installing the certificate: View certificates -> Install Certificates -> select “Local Machine” -> select “Place all certificates in the following store” -> select “Trusted Root Certification Authorities”.

From the Arcserve Support site, under the Product Support Information heading, click Patches to access Arcserve UDP 5.0 Solutions and Patches for the latest Arcserve UDP Version 5.0 Update 4 download information.

Installing the Update Manually

For an upgrade installation package for Windows servers and workstations that have a previous version of Arcserve UDP already installed, download and install the update manually using the following links to the installation files:

• Arcserve Unified Data Protection:

  http://downloads.arcserve.com/arcserve_family_products/arcserveudp/v5u4/FullUpdates/r5.0/Arcserve_Unified_Data_Protection_5.0_Update_4.exe

• Arcserve Unified Data Protection Windows Agent:

  http://downloads.arcserve.com/arcserve_family_products/arcserveudp/v5u4/EngineUpdates/r5.0/Arcserve_Unified_Data_Protection_Agent_5.0_Update_4.exe

For a fresh installation package for Windows servers and workstations that do not have Arcserve UDP installed yet, perform the following tasks:

Download and install Arcserve UDP Version 5.0 Update 4:

• Arcserve Unified Data Protection:

  http://downloads.arcserve.com/arcserve_family_products/arcserveudp/v5u4/Remaster_With_Update4/r5.0/Arcserve_Unified_Data_Protection_with_update_4.exe

• Arcserve Unified Data Protection Windows Agent:

  http://downloads.arcserve.com/arcserve_family_products/arcserveudp/v5u4/Remaster_With_Update4/r5.0/Arcserve_Unified_Data_Protection_Agent_Windows_with_update_4.exe

For a Linux installation, download and install the update manually using the following links to the installation files:

• Arcserve Unified Data Protection Linux Agent:

  http://downloads.arcserve.com/arcserve_family_products/arcserveudp/v5u4/GA/511a91c3/arcserve_Unified_Data_Protection_Agent_Linux_Update_4.bin

• Arcserve Unified Data Protection Linux Agent Restore Utility:

  http://downloads.arcserve.com/arcserve_family_products/arcserveudp/v5u4/GA/511a91c3/arcserve_Unified_Data_Protection_restore_utility_Update_4.iso

7.0 Contact Arcserve Support

The Arcserve Support team offers a rich set of resources for resolving your technical issues and provides easy access to important product information.

https://www.arcserve.com/support

With Arcserve Support:

• You can get in direct touch with the same library of information that is shared internally by our Arcserve Support experts. This site provides you with access to our knowledge-base (KB) documents. From here you easily search for and find the product-related KB articles which contain field-tested solutions for many top issues and common problems.
• You can use our Live Chat link to instantly launch a real-time conversation between you and the Arcserve Support team. With Live Chat, you can get immediate answers to your concerns and questions, while still maintaining access to the product.
• You can participate in the Arcserve Global User Community to ask and answer questions, share tips and tricks, discuss best practices and participate in conversations with your peers.
• You can open a support ticket. By opening a support ticket online, you can expect a callback from one of our experts in the product area you are inquiring about.
• You can access other helpful resources appropriate for your Arcserve product.

Copyright © 2015 Arcserve. All rights reserved. All trademarks, trade names, service marks, and logos referenced herein belong to their respective companies.