Hotfix NB_PDE_6.6.0.2_339244.tar provides critical fixes to Symantec Veritas NetBackup (tm) PureDisk 6.6

Article:TECH78125  |  Created: 2009-01-17  |  Updated: 2010-11-24  |  Article URL http://www.symantec.com/docs/TECH78125
Article Type
Technical Solution

Product(s)

Environment

Issue



Hotfix NB_PDE_6.6.0.2_339244.tar provides critical fixes to Symantec Veritas NetBackup (tm) PureDisk 6.6

Solution



Name: NB_PDE_6.6.0.2
Date: December 27, 2009

=========================================================================
This Hotfix provides features and fixes to the Veritas NetBackup
PureDisk Remote Office Edition 6.6 software.
=========================================================================

* PREREQUISITES

* PRODUCT FIXES

* KNOWN ISSUES

* VULNERABILITIES - RESOLVED

* DATALOSS ISSUES - RESOLVED

* PUREDISK DOCUMENTATION CHANGES

* PREINSTALLATION STEPS

* INSTALLATION INSTRUCTIONS

* UPGRADING PDDO AGENTS ON MEDIA SERVER CLIENTS



=============
PREREQUISITES
=============


This Hotfix can be installed on PureDisk 6.6 and PureDisk 6.6.0.1.


Note:


A client agent update job is not created for PDDO agents. Please follow
the procedure "Upgrading PDDO client agents on media servers".



=============
PRODUCT FIXES
=============


This patch contains the following product fixes:

* Etrack 1855576: Upgrading a Windows 32-bit client from PureDisk 6.5.1
 to PureDisk 6.6 a blue-screen is generated at reboot. This fix is also
 included in PureDisk 6.6.0.1.

 If you upgraded an existing 6.5.x PureDisk client on Windows
 XP/2003/2008 to the 6.6 client, do NOT reboot these Windows systems.
 First apply this hotfix and make sure that the client is properly
 upgraded.

 This issue applies to Windows XP, 2003 and 2008 on 32-bit platforms
 only. This issue applies to both the PureDisk backup agent and the PDDO
 agent on a media server . Windows 64-bit platforms are not affected.
 Windows 2000 machines (of any type) are not affected. This issue does
 not occur if PureDisk 6.6 is installed as the base install (i.e. not
 upgraded from 6.5.x).

 More information on this issue is available:
 http://entsupport.symantec.com/docs/334818

 Note:
 

 Windows 2000 machines MUST be rebooted after EACH install or upgrade of
 the PureDisk client.


* Etrack 1883248, 1883247: PDDO optimized duplications

 PDDO optimized duplications to PureDisk Remote Office Edition 6.6 may
 fail. The problem is due to an incorrect database view definition
 resulting in unpredictable query results on the target system. The
 backup fails with a NetBackup Status Code 84, "media write error".

 The NetBackup Administration Console Activity Monitor Detailed Status
 for a failed duplication displays the message: "optimized duplication
 failed, media write error (84)".

* Etrack 1767047: PDDO web service timeouts

 The PDDO plugin can timeout on a web service call to PureDisk server.
 This issue causes a storage leak during image cleanup. This patch
 provides a retry mechanism around the call.

 The following error message is seen in bptm or pdplugin logs:

 PDVFS: [1] pdvfs_lib_log: Webservice operation failure (opcode 28,
 Operation timed out after 120000 milliseconds with 0 bytes received)
 PDVFS: [3] filldir_mb: PdvfsCasMBGetFirst failed: 15

* Etrack 1883254: PDDO data removal locks PDDO backups.

 PDDO backups on NetBackup remain in "connected" state if a PDDO data
 removal job is queued or running on the PureDisk Storage pool. The
 following messages are seen in /Storage/log/pdwfe.log (and rotates)
 when this problem occurs:

 Agent 'PDDO' (id: 3): Could not create PDDO Backup Job: PDDO Data
 Removal queued or running

 This fix provides improved functionality with a new parameter for the
 Workflow Engine:

 File: /etc/puredisk/pdwfe.cfg

 Section: [intervals]

 Parameter: maxstarvation (default: 86400)

 Value: amount of time in seconds the PDDO data removal job needs to be
 queued before locking the creation of PDDO backup jobs.

 Example: A maximum starvation of one day (86400 seconds), means that a
 PDDO backup jobs runs correctly and timely the first 24 hours that a
 PDDO data removal job is queued. After those 24 hours, the PDDO data
 removal receives priority and no PDDO backup jobs are accepted, causing
 the NetBackup jobs to stall.


 Note:
 

 PDDO backup jobs and PDDO data removal jobs do not run simultaneously.
 This fix only provides improvement for the PDDO data removal jobs that
 are queued for a long period.


* Etrack 1886171: PDDO replication fails with "no such object"

 PDDO replication jobs fail with "Error: 2 : CRReplicate: Could not
 receive DO b8a61d743b47de956e4d1ca9d469d3fc to replicate: No such
 object." This problem happens when PDDO data removal expires a file,
 and PDDO replication runs before the metabase garbage collection.

 To fix this problem, expired or de-referenced files are no longer
 replicated to destination storage pool. As NetBackup is authoritative
 for image expiration on PureDisk, these "expired records" are obsolete
 if they are unknown to NetBackup. PDDO replication should not process
 the files.

* Etrack 1884684: PDDO backups fail when rerouting is in progress.

 A large portion of PDDO backups can fail when the rerouting workflow is
 active on the storage pool. This problem occurs because the PDDO
 backups add a reference to a fingerprint which is not yet rerouted
 correctly.

 Error:

pdvfs_lib_log: Sending binary message to 10.74.176.38:10082: DO_DS_REF_ADD
    2049263392 3 2: 7f08c44fb724ea7d3ee5a5fe58f3ddb8 2
pdvfs_lib_log: Received binary message from 10.74.176.38:10082: STATUS
    2049263392 3 1: 2
PDVFS PdvfsCasGetCrWriteConnection CRFileAddRef failed: n
pdvfs_lib_log: CRShutdown: start

 To fix this problem, a reference was added on the correct content
 router for fingerprinting during rerouting.

* Etrack 1853294: PDDO backup performance can decrease significantly
 due to low stream rate

 PDDO backups only use the first fragment of the previous backup as
 fingerprint cache. If a second fragment is made, the caching is no
 longer possible and more communication occurs between the NetBackup
 media server and PureDisk content routers.

 Fix: Use all fragments of previous image as fingerprint cache.

* Etrack 1883256, 1884725: Performance and stability issue when
 accessing large PDDO images

 For PDDO, especially Exchange granular recovery clients, backups and
 restores can fail on large images (larger than 200GB) with the
 following error messages:

 PDVFS: <ERR> pdvfs_lib_log: data retrieval failed: received -1 instead
 of 55595690 bytes (no such object)

 Error bpbrm(pid=6332) from client X: ERR - Error encountered while
 attempting to get additional files for Microsoft Information Store:\*

 Improvements to caching and to memory usage are implemented to avoid
 these issues.

* Etrack 1839973: Large PureDisk replication jobs can slow down

 PureDisk replication jobs, processing data selections with more than 1
 million files, can have very low performance during the "Import
 replication data selection" job, run on the destination storage pool.

 To fix this problem, the evaluation phase during remote Metabase import
 job for all data selections except PDDO was removed. This phase was
 moved to the data removal workflow.

* Etrack 1884812: Duplicate entries in the metabase engine are not
 protected by data removal

 Metadata in the metabase engine database can contain duplicates for the
 data that is backed up before PureDisk 6.6. In this case, a data loss
 issue can occur if the following conditions are true:

 - Time-base data removal policy is enabled

 - A "metadata filter" is used

 - The duplicate entries do not match the filter expression

 - The duplicate entries do not have a GUID (new in PureDisk 6.6)


* Etrack 1832593: Full disaster recovery backups fails if multiple
 metabase engines exist

 Full disaster recovery backups fail if multiple metabase engines are
 configured in the storage pool and encryption is enabled for the
 backup. The following error message is seen: No agent found for
 metabase engine controller id: X.

* Etrack 1883418: LVM/iSCSI partition not correctly mounted after a
 reboot

 If /Storage, /Storage/data or /Storage/databases is an LVM partition on
 top of iSCSI, these partitions may not be mounted after a reboot.

* Etrack 1912529: Not all backed up drives display in PureDisk WebUI.

 This issue affects the computers that run a Windows backup with
 snapshotting enabled and a data selection which has files and/or
 folders selected on multiple drives. This results in the restore WebUI
 showing only one drive as backed up.

 This issue also occurs with a Full System backup policy on a client
 system with multiple physical drives.

* Etrack 1904336: Path object list on client side can grow very large.

 This issue affects the computers that run Windows backups with
 snapshotting enabled and a data selection which has files and/or
 folders selected on multiple drives. This can result in very large path
 object list files under the PureDisk client installation path.



============
KNOWN ISSUES
============


The following are the known issues associated with this patch:

* ET 1922432: Disaster recovery restore fails after the PureDisk
 6.6.0.2 patch is applied because a newer version of the Java Runtime
 Environment is installed. The error message is as follows:

+--------------------------------------------------------+
|              Reinstalling Configuration                |
+--------------------------------------------------------+
+ Checking the content of 'topology.ini'...
+ Performing installation environment sanity check ...
While checking rpm dependencies, the following problems were found:
   -   package jre-1.6.0_17-fcs (which is newer than jre-1.6.0_14-fcs) is
already installed

 Workaround: Uninstall the Java Runtime Environment package on all nodes
 before running the DR_Restore_all.sh script using the following command:

 rpm -e jre-1.6.0_17-fcs


* ET 1922902: At the end of a Disaster Recovery restore of PureDisk 6.6.0.2,
 the following message appears on screen:

 ./DR_Restore_all.sh: line 14: 17828 Segmentation fault      
 $PHPBIN /opt/pdinstall/lib/DRRestoreAll.php

 This message does not impact the restore or system functionality and can
 be ignored.

==========================
VULNERABILITIES - RESOLVED
==========================


The following are resolved vulnerabilities in PureDisk 6.6.0.2:

* OpenSSL package update (ET1895970)

 The TLS/SSLv3 protocol as implemented in openssl before this update was
 not able to associate data to a renegotiated connection. This allowed
 man-in-the-middle attackers to inject HTTP requests in an HTTPS session
 without being noticed. For example Apache's mod_ssl was vulnerable to
 this kind of attack because it uses openssl. Please note that this
 update disables renegotiation and may cause problems in some cases.
 (CVE-2009-3555: CVSS v2 Base Score: 6.4)

* Java Runtime Environment package update (ET1877357)

 Multiple security vulnerabilities have been found in JRE 1.5 and 1.6.
 The latest JRE version update addresses these issues.

* Clear-text password between metabase server and metabase engine
 encrypted (ET1861632)

 The communication between the metabase server and the metabase engine
 postgres isn't SSL protected. Passwords pass over the wire as plain
 text. Since all the databases use the same password, every database is
 compromised.



==========================
DATALOSS ISSUES - RESOLVED
==========================

The following data loss issues are resolved in this Hotfix:

* After a client backup, files and folders that were deleted off the
 client are added to the PureDisk metabase as deleted entries. These
 entries are then used to purge data during the data removal process.
 For Windows agents that use PureDisk version 6.6 and 6.6.0.1, files and
 folders that are still present on the client system can incorrectly be
 added as deleted entries in the metabase. (Etracks 1895349 and 1901224)
 More information about this issue is available.
 http://entsupport.symantec.com/docs/338287

 Unless you upgrade to 6.6.0.2 without having run any backups at 6.6,
 you should contact technical support for additional cleanup work if you
 meet the criteria in the techalert.

* Duplicate entries in the metabase engine are not protected by data
 removal (Etrack 1884812)

 Metadata in the metabase engine database can contain duplicates for the
 data that is backed up before PureDisk 6.6. In this case, a data loss
 issue can occur if the following conditions are true:

 - Time-base data removal policy is enabled

 - A "metadata filter" is used

 - The duplicate entries do not match the filter expression

 - The duplicate entries do not have a GUID (new in PureDisk 6.6)


 More information about this issue is available.
 http://entsupport.symantec.com/docs/337448

==============================
PUREDISK DOCUMENTATION CHANGES
==============================


The following information has moved from the PureDisk Storage Pool
Installation Guide to the PureDisk Administrator's Guide:

* Configuring external directory service authentication

* Configuring single-port communication

* Stopping and starting processes


Please review the PureDisk Administrator's Guide for the relocated
information.


=====================
PREINSTALLATION STEPS
=====================


This section describes the preinstallation steps for the Hotfix.

Downloading and extracting the patch

1. Use scp to copy the tar file, NB_PDE_6.6.0.2_339244.tar to the /root
  directory of the PureDisk node that hosts the storage pool authority.

2. Log on as root to the node that hosts the storage pool authority.

3. Type the following command to verify the integrity of the
  Hotfix:

  md5sum /root/NB_PDE_6.6.0.2_339244.tar

  This command computes the md5 checksum of the Hotfix. The md5
  checksum of the Hotfix must match '8c14acaad3defb531543032eec0f8bd7'.
  http://entsupport.symantec.com/docs/339244.

  If you obtain a different checksum, the Hotfix was corrupted
  during download. Try to download the Hotfix again.

4. Type the following command to extract the README file:

  tar -C / -xf /root/NB_PDE_6.6.0.2_339244.tar ./NB_PDE_6.6.0.2



=========================
INSTALLATION INSTRUCTIONS
=========================


This section describes how to install the software.


Note:


Windows 2000 users should be aware that Windows 2000 servers must be
rebooted after upgrading their severs using this hotfix. Please plan the
deployment of this package appropriately to eliminate any unscheduled
down time in your environment.


Installing the patch

1. Make sure that no PureDisk jobs are currently running or are
  scheduled to be run.

2. Log out from the Web UI.

3. (Conditional) Freeze the PureDisk service groups for the clustered
  PureDisk server.

  Perform this step if the storage pool is installed with VCS cluster
  software. Use the Cluster Manager Java Console, and freeze all the
  service groups.

  For information about how to freeze and unfreeze clustered storage
  pools, see the Veritas NetBackup PureDisk Storage Pool Installation
  Guide.

4. Type the following command to unpack the Hotfix software:

  tar -C / -xf /root/NB_PDE_6.6.0.2_339244.tar ./opt

5. Type the following command to run and install the Hotfix:

  /opt/pdinstall/apply-NB_PDE_6.6.0.2.sh

  If the topology.ini file is encrypted, the software prompts you for
  the password to decrypt this file.

  The Hotfix automatically pushes the software to all nodes in
  the storage pool and to all clients.

  At the end of a successful installation, the software prompts you to
  encrypt the topology.ini file.

6. (Conditional) Unfreeze the PureDisk service groups for the
  clustered PureDisk server.

  Perform this step if the storage pool is installed with VCS cluster
  software.

  The Veritas Cluster Server (VCS) software might detect some faults
  during the upgrade process. If any upgrade actions generate a VCS
  fault, use the Cluster Manager Java Console to clear the fault and
  probe that resource group before you unfreeze the cluster.

7. Monitor the client agent update jobs

  Take the following actions to monitor the client agent upgrade jobs:

  * Invoke the Web UI.

  * Click Monitor > Jobs.

  * Set the "View jobs by" dropdown to "Policy types".

  * Select the "Agent Update" workflow under "Miscellaneous
    Workflows".


  If one of the upgrade jobs fails, PureDisk deactivates those client
  agents. You need to activate these agents again and to select the
  client to upgrade to. Follow the steps below:

  * Invoke the Web UI.

  * Click Manage > Agents.

  * In the left pane, select the deactivated agent you wish to
    reactivate. Select the storage pool level if you wish to activate
    all agents in the storage pool.

  * In the right pane, select "Activate Agent(s)".

  * A new agent update job is scheduled for those agents not
    upgraded yet.

  * Monitor these upgrade jobs again. Start them over if they fail
    or timeout.


8. (Conditional) Upgrade the PDDO agent software on PDDO media server
  clients.

  Perform the procedure called "UPGRADING PDDO AGENTS ON MEDIA SERVER
  CLIENTS", which follows. The release update includes upgraded PDDO
  agents for Linux, Solaris, and Windows clients.



=============================================
UPGRADING PDDO AGENTS ON MEDIA SERVER CLIENTS
=============================================


This section describes how to upgrade PDDO agents on media server clients.


Note:


The upgrade software removes the old PDDO agent software automatically.
For more information about how to install the PDDO agent, see the Veritas
NetBackup PureDisk Deduplication Option Guide.


Upgrading PDDO

1. Make sure that the storage pool software for the PDDO clients has
  been upgraded.

2. Refresh the PureDisk landing page.

  The landing page can be found on:

  https://<your-puredisk-spa>/

  For <your-puredisk-spa>, type the hostname or the IP address of your
  PureDisk storage pool authority.

  Be aware that this Hotfix includes both the PureDisk 6.6 base
  release agents and revised agents for Linux, Solaris, and Windows
  clients. The revised agents include updates for PDDO agent
  installation only. A later step in this procedure instructs you to
  download the revised agents if you want to upgrade the PDDO agent on a
  client.

3. Log on to each PDDO client.

4. From the PDDO media server client, download the new PDDO agent
  software.

  The landing page includes more than one software version of the PDDO
  agent, so take care to download the most recent agent.

5. (Conditional) Freeze the PureDisk service groups for the clustered
  PDDO server.

6. Stop all NetBackup services on the NetBackup media server.

7. Install the agent software that is included with this
  Hotfix.

  * On Windows systems, double-click the PureDisk agent icon. That
    starts the Windows Installation Wizard .

  * On Linux or Solaris systems, you can use either the attended or
    the unattended installation method. The installer prompts you to
    confirm the upgrade. For example, type the following command to
    upgrade the Solaris 10 agent:


  # sh pdagent-Solaris_10_sparc-6.6.0.38844.run

8. (Conditional) Unfreeze the PureDisk service groups for the
  clustered PDDO server.

9. Start all the NetBackup services on the NetBackup media server.




Attachments

NB_PDE_6.6.0.2_339244.tar (165.1 MBytes)


Legacy ID



339244


Article URL http://www.symantec.com/docs/TECH78125


Terms of use for this information are found in Legal Notices