3par Prep Guide

1

Abstract This document is intended for use as a planning tool for HP customers. It contains the HP recommendations for HP 3PAR OS upgrade process. This document is divided into 3 sections. Refer to each section and follow the checklist given below and mark the task status as complete and NA if the recommendation is not applicable: Pre-upgrade Validation Checks to be done during the upgrade Post upgrade tasks NOTE: HP recommends updating host drivers to a supported level of the targeted HP 3PAR OS upgrade version before attempting an online update. This information is available on the HP Single Point of Connectivity (HP SPOCK) website: www.hp.com/storage/spock HP recommends checking the compatibility information for HP CloudSystem Matrix versions. See The HP CloudSystem Matrix Compatibility website: HP CloudSystem Matrix Compatibility

These documents are provided to assist you in setting up host servers with HP-supported configuration information. In addition to this planning document, the implementation guides are available on the HP Business Support Center (BSC) website: http://www.hp.com/go/bsc

HP 3PAR OS Upgrade Preparation Guide February - 2015

2

Pre-upgrade validation

Applies to Host OS/Array

Task (Click on hyperlink) Impact

All 3PAR OS Versions Windows

2008/12 Update MS 2008/2012 MPIO hotfix & persona check, non-PP VLUNs check

Absence of MPIO hotfix may lead to loss

of access to 3PAR LUN

All 3PAR OS Versions

Red Hat,

Oracle, Linux &

SUSE OS Verify multipath.conf as per IG

Loss of access to 3PAR LUN after 3PAR

Node reboot

All 3PAR OS Versions VMware MPIO Path Policy set to Round-Robin Loss of access to 3PAR Luns on ESX server during the upgrade

Upgrade to 3PAR OS 3.1.3.x

VMware Set auto_failover for PP RCOPY groups

Loss of access to 3PAR Lun from the ESX hosts(both primary and secondary)

All 3PAR OS versions NetApp Dedicated ports for NetApp Controllers

NetApp hosts needs the host port to be dedicated for Ontap only

All 3PAR OS Versions HP insight Control

Advisory Supportability

All 3PAR OS Versions HP Cloud System

Validate compatibility with 3PAR OS Cloud system would break if 3PAR OS is upgraded to an non-compatible version

All 3PAR OS Versions VERITAS DMP Set VERITAS DMP Restore Interval Loss of access to 3PAR LUNs during the OS upgrade

All 3PAR OS Versions Oracle RAC Verify timeout values Loss of access to 3PAR LUN after 3PAR Node reboot

All 3PAR OS Versions Solaris Verify minimum patch levels Loss of access to 3PAR LUNs during the OS upgrade

All 3PAR OS Versions Exanet ExaStore Clustered NAS

Execute scripts I/O quiescence during the HP 3PAR OS upgrade

312.MU5 and higher Host based applications Update host based applications

Host Based applications might cease to work if not updated

All 3PAR OS Versions HP-3PAR Array Remote Copy support Remote Copy failure

3.1.2 and higher HP-3PAR Array Persistent ports

Avoids host outage

3.1.3 and higher HP-3PAR Array Stop and start Quorum witness

Host outage if VV involved is part of PP Rcopy groups

All 3PAR OS Versions HP-3PAR Array Validate 3PAR Array Health If a pre-check fails, upgrade cannot be completed

Upgrades from 3.1.2.x and higher

HP-3PAR Array HP Proliant Model Service Processor

Post reboot, HP Proliant Model SP might not come back

Upgrade to 3PAR OS 3.1.3.GA and higher

HP-3PAR Array Set speed to auto on admin and RCIP ports

MC and CLI sessions will stop working causing array to be unmanageable

Upgrade from 3PAR OS 3.1.1 to 3.1.2

HP-3PAR Array Validate showpd -path PDs go into degraded state in the middle of the upgrade causing upgrade to fail

Upgrade from 3.1.x. to 3.1.2.x and above

HP-3PAR Array Stop AO, DO and active tasks Upgrade will not proceed if any active tasks are detected

3

Checks to be done during the upgrade



All 3PAR OS Versions AIX Monitor paths in AIX

Risk of losing access to 3par LUN

All 3PAR OS Versions iscsi hosts Ensure continuous I/Os during upgrade

Paths will not be reconnect post Node

upgrade of 3PAR

All 3PAR OS Versions Solaris Monitor paths in Solaris

Risk of losing access to 3par LUN

3.1.2.x and Higher HP-UX HP-UX Path rescan

Host path shows "No_Hw"

Post upgrade tasks



All 3PAR OS Versions HP 3PAR Array Change persona Unexpected behavior of LUNs exported from 3PAR

All 3PAR OS Versions HP 3PAR Array Update GUI and CLI clients Applications might not work with older 3PAR OS

All 3PAR OS Versions HP-3PAR Array Capture existing policies Existing AO policies will be lost

4

NOTE: Failure to follow the below recommendations from Microsoft may lead to host losing access to storage.

1. Recommended Hotfixes by Microsoft for Windows 2008 & 2012 servers For Windows Server 2008 using MS MPIO or Windows 2008 VM running on ESX using a HBA in pass-through mode or an iSCSI LUN, following Microsoft hotfixes must be applied before updating the HP 3PAR Storage System: Windows Server 2008, SP1, SP2, R2 and R2 SP1 require KB2754704 installed to resolve issue with MPIO path failover. Read MS Article: http://support.microsoft.com/kb/2754704 for more details. Windows Server 2012 and Windows Server 2008 R2, R2 SP1requires KB2821052 installed to resolve issue with MPIO path failover. Read: http://support.microsoft.com/kb/2821052 for more details.

2. Load Balancing Policy should be Round-Robin For Windows 2008 that are using the Microsoft native MPIO driver, it is required the load-balancing policy be set to Round-Robin".

The procedure for checking or changing load-balancing policy can be found at http://technet.microsoft.com/en-us/library/ee619752%28WS.10%29.aspx See section "To configure the load-balancing policy setting for a LUN" You must restart the computer after applying the above listed hotfix. 3. ONLY for 3.1.3.x and higher Upgrades - Persona 15 for Windows 2008 and Windows 2012 Server: New Host Persona 15 for Windows Hosts Needs Set after upgrading to HP 3PAR OS 3.1.3.x and higher to facilitate appropriate LUN behavior. Click here for instructions to change persona on a windows host.

4. For OS upgrades to 3.1.1.x and 3.1.2.x (not applicable for 3.1.3.x and higher. See Host persona notes in post upgrade section for 3.1.3.x upgrades) On the first Windows Server 2012 or Windows Server 2008 reboot following an HP 3PAR array firmware upgrade (whether a major upgrade or an MU update within the same release family), the Windows server will mark the HP 3PAR LUNs offline but the data remains intact. This behavior is seen only in the following cases: HP 3PAR LUNs seen by Windows standalone servers. HP 3PAR LUNs that are used in Microsoft Failover Clustering which are not configured as shared storage on the Windows failover cluster. This behavior (LUNs being marked offline) is not seen if HP 3PAR LUNs are configured as shared storage in a Microsoft Failover Cluster.

Windows Server 2008 and 2012

5

NOTE: When the HP 3PAR LUNs are marked offline, the Windows server administrator must follow these steps so that applications can access the HP 3PAR LUNs: a. Click Computer Management Disk Management. b. Right-click each of the HP 3PAR LUNs. c. Set the LUN online. To prevent this issue, execution of Microsoft KB2849097 on every Windows Server 2008/2012 host connected to a HP 3PAR array is recommended. This must be executed on a host at the following times: a. Prior to performing a 3par OS upgrade. b. Each time new HP 3PAR LUNs are exported to a host. KB2849097 is a Microsoft PowerShell script designed to modify the Partmgr Attributes registry value that is located at HKLM\System\CurrentControlSet\Enum\SCSI\\\DeviceParameters\Partmgr. The value is responsible for the state of HP 3PAR LUNs following an array firmware upgrade. The script sets the value to 0 essentially changing its policy to online. NOTE: The following procedure will ensure proper execution of KB2849097, which will prevent the HP 3PAR LUNs from being marked offline when the Windows server is rebooted following an array firmware upgrade. a. Windows Server 2008/2012 requires the PowerShell execution policy to be changed to RemoteSigned to allow execution of external scripts. This must be done before the script is executed. To change the PowerShell execution policy, open the PowerShell console and issue the following command: Set-ExecutionPolicy RemoteSigned You might be prompted to confirm this action by pressing y. b. The next step is to save the script as a .ps1 file to a convenient location and execute it by issuing the following command in a PowerShell console window: C:\ps_script.ps1 The above command assumes that the script has been saved to C: under the name ps_script.ps1. c. The Administrator will then be prompted to provide a Vendor String, which is used to distinguish between different vendor types. The script will only modify those devices whose Vendor String matches the one that has been entered into the prompt. Enter 3PAR in the prompt to allow the script to be executed on all HP 3PAR LUNs currently presented to the host as shown in the output below: Enter Vendor String: 3PAR d. The script will then iterate through all HP 3PAR LUNs currently present on the host and set the Attributes registry value to 0. In order to verify that the Attributes value for all HP 3PAR LUNs were properly modified, issue the following command: Get-ItemProperty path

"HKLM:\SYSTEM\CurrentControlSet\Enum\SCSI\Disk*Ven_3PARdata*\*\Device

Parameters\Partmgr" -Name Attributes

The Attributes value should be set to 0 as shown in the example below: TOP

6

5. Remote Copy Groups Using Non-Peer Persistent VLUNs and Windows 2012 Host Clusters DESCRIPTION

After upgrading from HP 3PAR OS 3.1.2.x to 3.1.3 or higher, the passive node (seeing the read-only copy of the RC

group) will hang during a reboot. Furthermore, if only the boot disk is presented (no clustered disks are

presented), the system boots correctly. If one of the clustered disks is added, the passive node will hang during

the reboot. Then when adding the clustered disks while the passive node is up and running, both Device Manager and Disk Manager hang during a re-scan.

The results to the active node are assumed to be the same if it is accessing a read-only copy of an RC group.

This situation can occur in the following environment:

Cluster nodes are Windows 2012 R1 Standard Edition

Windows Failover Cluster with CLX 4.0

HP 3PAR arrays involved in RC Groups are running HP 3PAR OS v3.1.3

Boot from SAN

Clustered nodes are configured as Persona 15

Peer persistent VLUNs are not active/presented to clustered nodes

RESOLUTION

Make sure that the Windows hosts are accessing non-peer persistent VLUNs! DO NOT use the following steps

on peer persistent VLUNs!

These issues may be avoided or corrected by performing the following steps. Follow these ONLY on Non-Peer

Persistent VLUNs as Peer Persistent VLUNs need auto_failover to be set prior to the upgrade.

To prevent this issue from occurring prior to upgrading to 3.1.3 or higher:

Perform the following on primary groups on both arrays:

setrcopygroup pol no_auto_failover

The policy, path_management, does not exist prior to 3.1.3.

7

To correct this issue if it occurs after upgrading to 3.1.3 or higher:

Perform the following on primary groups on both arrays:

For the non-peer persistent group, the following policies need to be removed:

auto_failover

path_management

Remove the policies using:

setrcopygroup pol no_auto_failover

setrcopygroup pol no_path_management

6. Configuring the HP 3PAR StoreServ Storage Host Persona and Ports (HP 3PAR OS 3.1.x) For an HP 3PAR StoreServ Storage running HP 3PAR OS 3.1.x, each HP 3PAR StoreServ Storage port connecting to

a host HBA port through a fabric or direct connection must be set to the correct host persona.

NOTE: When deploying HP Virtual Connect Direct-attach FC Storage for HP 3PAR StoreServ Storage, where the HP

3PAR StoreServ Storage ports are cabled directly to the uplink ports on the HP Virtual Connect FlexFabric 10

Gb/24-port Module for c-Class Blade System, follow the steps for configuring the HP 3PAR StoreServ Storage

ports for a fabric connection.

Configuring Ports on the HP 3PAR StoreServ Storage for a Direct Connection:

To configure HP 3PAR StoreServ Storage ports for a direct connection to the Windows Server 2012/2008 host on

HP 3PAR OS 3.1.x, complete the following steps:

To set up the HP 3PAR StoreServ Storage ports for a direct connection, issue the following set of commands with

the appropriate parameters for each direct connect port, one port at a time:

A. Validate the current port mode:

showport c

(Topology should be set as loop for direct attached hosts. If it is set to fabric, follow procedure below:

B. controlport offline

C. controlport config host -ct loop where -ct loop specifies a direct

connection.

D. controlport rst

E. Validate that ports is online (showport c N:S:P)

F. Validate if the host paths are online before working on the next port.

Use showhost d to see active paths from the array.

TOP

8

NOTE: Failure to follow the below recommendations may lead to host losing access to storage.

Observe the following host configuration guidelines when planning an online HP 3PAR OS upgrade

for Red Hat, Oracle, Linux, or SUSE OS: 1. We have seen a few instances where the multipath.conf file is not configured per the guide. This may cause incorrect detection of path removal during a node reboot or incorrect path failover. For the Red Hat hosts, please ensure that the system is configured per the current 3PAR/HP Implementation Guide. It can be downloaded from: http://h20000.www2.hp.com/bizsupport/TechSupport/DocumentIndex.jsp?lang=en&cc=us&taskId=101&prodClassId=-1&contentType=SupportManual&docIndexId=64255&prodTypeId=18964&prodSeriesId=5044394 2. The SCSI timeout value for RHEL 4 and SUSE 9 must be changed from the default of 30 seconds to 60 seconds. To change the timeout value, see the HP 3PAR host OS implementation guide specific to the OS. This needs to be performed for both FC- and iSCSI-connected hosts. This change is not required in RHEL 5 and 6. WARNING: If not set to 60 seconds, the SCSI timeout will result in host disks being taken offline during InServ Storage Server rolling upgrades. Furthermore, Remote Copy requires the SCSI timeout value of 60 seconds, otherwise remote copy operations will become stale with a node reboot. 3. QLogic drivers for SUSE 9, 10 and Red Hat 4, 5 should be running with driver parameter PORT-DOWN retry

count (qlport_down_retry) set to 1. To set the value, see the HP 3PAR host OS implementation guide specific to

the OS.

NOTE: If the HP 3PAR StoreServ Storage array is running HP 3PAR OS 3.1.1 or later you must modify the HBA

parameter by setting qlport_down_retry to 10. If the HP 3PAR StoreServ Storage array is running an HP 3PAR OS

version earlier than 3.1.1, set qlport_down_retry to 1.

5. For hosts using Oracle Clusterware, make the following required changes or adjustments before performing an

upgrade:

For 10gR1 10.1.0.x and 10gR2 10.2.0.1 (without bug 4896338 patched), set css miscount to 195:

#set css miscount=195

For 10gR2 10.2.0.1 (with bug 4896338 patched) and later, 11gR1, and 11gR2, css disktimeout must not be

lowered from its default value of 200.

For all Linux hosts with Oracle Clusterware, install and adjust Linux Hangcheck tick parameters to 30 seconds.

Red Hat, Oracle, Linux, and SUSE OS

9

1. VMware ESX/ESXi server - Active/Active, with a path policy of Most Recently Used (MRU), does not maintain or re-instate balancing of I/O load after a failover/failback multipath event. This could leave I/O in an unbalanced state, which may yield significant I/O performance issues. Implementation of an MRU path policy is not recommended by HP. As of ESX 4.0, a third path policy choice of Round Robin is available. This is the recommended path selection policy on ESX systems where it is available. The VMware vCenter Site Recovery Manager and the HP 3PAR Adapter for VMware Site Replicator require newer versions to be compatible with HP 3PAR OS 3.1.1 or later. For details, see VMwares Site Recovery Manager Storage Partner Compatibility Matrix: http://www.vmware.com/pdf/srm_storage_partners.pdf

2. Host persona 11 is the HP-recommended host persona for all supported versions of VMware ESX with HP 3PAR OS 3.1.2 and later. After HP 3PAR OS 3.1.3 and its MUs, host persona 6 will not be supported for any VMware ESX version with HP 3PAR storage systems. For information regarding HP 3PAR host persona support for all supported VMware ESX versions, see Table 4: Host Persona Details Table (page 12).

Click here to view VMware ESX Implementation Guide

3. Upgrades to HP 3PAR OS 3.1.3 & Higher when Persona 11 is used with Remote copy (Sync) - UPDATE: Peer Persistence is ONLY supported using ESX hosts. When Remote Copy is configured with RC groups (Sync Mode) that are in 3.1.2 MU2 and higher system, VVs are exported from both Primary/Secondary RC systems to the same ESX host. Volume paths for a given volume are Active only on the array where the Primary copy of the volume resides. Other volume paths are marked Standby. Primary volume is in read/write (RW) mode, secondary volume is in read/only (RO) mode. This can be viewed on the ESX host path management. Peer Persistence has an existing setting called "auto_failover to support Automatic Transparent Failover (ATF) in high availability (HA) configuration that are in synchronous replication. If this setting is NOT turned on before Upgrading to 3.1.3.x and higher, the paths for the PP RC Groups will change from standby to active. ESX host will no longer be able to write to LUNs as these paths on the secondary array are marked as read-only (RO). 1. Check if the system is licensed for Peer Persistence by issuing the Cli Command: Cli%showlicense. If Peer Persistence license is not found, this recommendation can be ignored. 2. To check if this is a Peer Persistent configuration, the same WWN is exported from both primary/secondary site and exported LUN numbers are the same for those LUNs. NOTE: If you know this is a Peer Persistent configuration and the WWN and LUN numbers are NOT the same, they need to be configured as such or Peer Persistence will not function when a failover is initiated and the LUNs will not be accessible after the upgrade without considerable support work 3. To check if auto_failover is set for all RC Groups: a) Use the "showrcopy -d" output to verify. You will find "auto_failover" under the options column in the output:

VMware

10

Group Information

Name ID Target Domain Status Role Mode LocalUserCpg

LocalSnapCpg RmUserCpg RmSnapCpg Options

BCK-CLU-01-DP_VMGRE 284 3PAR01-DC2 - Started Primary Sync

auto_failover,path_management

LocalVV ID RemoteVV ID SyncStatus Resync_ss Sync_ss

VV_iter R_iter S_iter LastSyncTime

BCK-CLU-01-DP_VMGRE 1038 BCK-CLU-01-DP_VMGRE_DR 980 Synced none none

NA NA NA NA

b) Setting auto_failover can be done using the following command: Cli%setrcopygroup pol auto_failover This command can be issued only on a group that is in the source role. The policy is also assigned to the target group when issued on the source. 3. Verify Hosts in this Peer Persistent configuration are only ESX hosts. Windows hosts are not supported. 4. Verify the ESX host's persona is set to 11 TOP

Environment

HP 3PAR StoreServ 10000 Storage



1. NetApp hosts needs the host port to be dedicated for Ontap only, sharing the port with a non-ONTAP host is not

supported.

Validate if you are using a Netapp V-series controller in your environment .In case of a Netapp V-series controller,

kindly note that storage array target ports cannot be shared with other hosts or even with other V-series

controllers. Ensure that storage array target ports should be dedicated to the V-Series controller.

2. Check for any ONTAP hosts in the host configuration using Cli command showhost d

host.xxx.xxxxxxx - ONTAP-legacy 500xxxxxxxxxx1x 4:2:3




If the 3PAR OS is below 3.1.2, shared RCFC ports and host ports on the same HBA are not supported.

Ontap Hosts shares port with Non-Ontap Host(s)

11

Solution:

As of HP 3PAR OS 3.1.2, the HBAs on each HP 3PAR StoreServ 7000 Storage system can use shared RCFC and host

cable connections to connect the systems through the FC SAN. This configuration is only supported on HP 3PAR

StoreServ 7000 Storage systems running HP 3PAR OS 3.1.2 or later.

For other 3PAR Storage models (V, F & T class), NetApp hosts needs the host port to be dedicated for Ontap only,

sharing the port with a non-ONTAP host is not supported.

The command controlport config can be used to set the specified connection mode and type on FC ports.

It also allows setting the unique node WWN option for the port. When unique_nwwn is enabled, the port presents

a unique node name on the connection and this is needed by certain initiators such as ONTAP. Please refer to the

HP 3PAR Command Line Interface Reference for further details. Click here to view the HP 3PAR Command Line

Interface Reference manual.

HP Insight Control for VMware vCenter Server does not support cluster provisioning operations for HP 3PAR

Storage arrays running firmware 3.1.2 MU23PAR CLI Client Recommendation. Customer Advisory below:

HP Insight Control for VMware vCenter Advisory

If a cloud system matrix is deployed in the environment, it is important that the HP 3PAR OS to which the 3PAR

Array will be upgraded is compatible with the CSM version.

To validate the compatibility, refer link below:

http://www.hp.com/go/insightmanagement/docs

Under the Getting started section, look for the document called HP Insight Management Support Matrix

(including HP Insight Control and HP Matrix OE support matrix). Once that is opened, go to the section Supported

Storage Products and you will see a chart listing the supported versions of 3PAR OS with MOE and Insight Control.

NOTE: The following recommendations are from Symantec. Please contact the vendor for further support on this. The default value of the VxDMP link restore (dmp_restore_interval tunable) interval must temporarily be changed to a value of 1 second before performing an online update of the HP 3PAR OS for the following VxDMP versions:

VERITAS DMP

HP Insight Control for VMware vCenter

HP Cloud systems/Converged systems Interoperability with 3PAR OS

12

All Linux/Solaris/AIX VxDMP versions older than 5.0MP1 All HP-UX VxDMP versions older than 5.0MP3 Changing the link restore interval causes the VxDMP software to react to the return of missing paths more quickly than it normally would with default settings. Failure to perform the change to the shorter restore interval for the noted VxDMP versions above, may cause a loss of host connectivity to HP 3PAR volumes during the upgrade. For instructions, see Changing the VERITAS DMP Link Restore Interval (below). After the online update, reset the link restore interval to the previously configured value. For instructions, see Restoring the VERITAS DMP Link Restore Interval (below). As of VxDMP 5.0MP1 for Linux/Solaris/AIX and 5.0MP3 HP-UX, Veritas software was modified to circumvent the path loss issue related to the link restore interval during an HP 3PAR OS upgrade and no changes to the restore interval are necessary. TOP

Changing the VERITAS DMP Link Restore Interval: The following procedure may be required if you are performing an online update to the HP 3PAR OS and VERITAS DMP is being used for the multipathing solution on the host. The procedure must be repeated for each host server that is connected to the HP 3PAR StoreServ Storage and using VERITAS DMP. Use the following procedure to gather the current settings and change the link restore interval: Display the current setting of the link restore interval as follows. Record the interval so that it can be restored later. 1. Display the current setting of the link restore interval as follows. Record the interval so that it can be restored later. # vxdmpadm stat restored (note the current setting) 2. Stop the daemon. # vxdmpadm stop restore 3. Restart the daemon with the interval set to 1. # vxdmpadm start restore interval=1 4. Verify that the daemon is running with the interval set to 1. # vxdmpadm stat restored NOTE: The restore daemon reverts to the default value of 300 if or when a host reboot occurs. Restoring the VERITAS DMP Link Restore Interval If the VERITAS DMP link restore interval has been changed to 1 second for HP 3PAR OS online update, use the following commands to change the link restore interval to the previous setting on storage server hosts after completing an online update. The following procedure must be repeated for each host server that is connected to HP 3PAR Storage and using VERITAS DMP. Use the following procedure to restore the link restore interval to its original setting: 1. Stop the daemon. # vxdmpadm stop restore 2. Restart the daemon with the interval set to what it was previously. # vxdmpadm start restore interval=

13

3. Verify the daemon is running with the interval set to the prior setting. # vxdmpadm stat restored

1. Please ensure that Solaris host matches the minimum patch requirements for various versions of Solaris and other associated drivers as listed in the Solaris Implementation guide available at: http://h20000.www2.hp.com/bizsupport/TechSupport/DocumentIndex.jsp?lang=en&cc=us&taskId=101&prodClassId=-1&contentType=SupportManual&docIndexId=64255&prodTypeId=18964&prodSeriesId=5044394 2. Solaris using VxDMP 5.0MP3 RP1HF3: dmp_fast_recovery turned off as a minimum. However, 5.0MP3 RP2HF1 does not require this tunable turned off. Please contact Oracle for further Support on this. 3. Solaris 10 MU3 with an iSCSI configuration is not supported for online updates.

There are special procedures that will need to be performed on an Exanet ExaStore v4.0.x Clustered NAS system when an HP 3PAR OS update needs to take place. In the standard ExaStore configuration, NAS clients that connect to the ExaStore might experience excessive periods of I/O quiescence during the HP 3PAR OS upgrade. Contact Exanet for details on obtaining this script as well as specific usage instructions. An overview of the process is included for quick reference: The procedure includes running two scripts: 1. Pre-upgrade script, which performs the following:

Modifies appropriate configuration values.

Generates the post-upgrade script so that the system can be returned to the standard configuration after the HP 3PAR OS software update is completed. 2. Post-upgrade script, which sets all appropriate values that were modified in the Pre-upgrade script back to their original values. Installing 3.1.2.MU5, 3.1.3.x and 3.2.1 version of HP 3PAR OS requires that all affected software components in the storage environment be upgraded or installed in a specific order as detailed in the following list. Any listed host applications accessing the HP 3PAR storage system must be upgraded to the specified software levels in order for the storage environment to take full advantage of the new SSL certificate management functionality.

Sun Solaris

Exanet ExaStore Clustered NAS System

HP 3PAR Host Based Software

14

Depending on the number of host applications in the storage environment, the full installation process might take a considerable amount of time to complete. Compatibility matrices for all HP 3PAR Host based applications can be found in the link below: http://h20272.www2.hp.com/Pages/spock2Html.aspx?htmlFile=sw_array_3par.html These software can be be obtained from the following link:

http://www.hp.com/go/hpsoftwareupdatesupport

You will be required to create a HP Passport login account and provide the contractual support agreement (SAID). The host application upgrades listed in step 1 below can be completed weeks ahead of the remaining upgrade/install steps, if needed.

1. Upgrade the following applications, in any order desired, to the revision level indicated:

Recovery Manager for Exchange (RME)

Recovery Manager for SQL (RMS)

Recovery Manager for HyperV (RMH)

Recovery Manager for VMware (RMV)

Recovery Manager for Oracle (RMO)

VSS Provider

HP 3PAR Cluster Extension for Windows (CLX)

Host Explorer

HP Storage Plug-in for SAP Landscape

HP StoreFront Analytics Pack for VMware vCOPS

MetroCluster/Serviceguard

2. Upgrade the HP 3PAR Management Console

3. Upgrade the CLI and SNMP to respective upgraded OS version

4. Upgrade the HP 3PAR StoreServ 7000 Storage SmartStart Software (SS)

TOP

15

If you have this array in remote copy relationship with another 3PAR array, the allowed delta between major

versions of the HP 3PAR OS release is limited to adjacent major release levelsfor instance, from HP 3PAR OS

3.1.2 to 3.1.3. Please see table below for supported OS levels between an RCOPY Pair:

We only support replication for 5 weeks if there is a difference of 1 major OS version between the two arrays (e.g. 3.1.3 to 3.2.1 is supported). We do not support replication if there is a difference of more than 1 OS major version (e.g. 3.1.2 to 3.2.1 replication is not supported)

Note: Upgrade the remote copy pair to the same OS level within a period of 5 weeks

If SAN/Fabric does not support NPIV, Persistent ports functionality on 3PAR needs to be disabled to avoid any path loss issues during the 3PAR OS upgrade. This is applicable for OS upgrades from 3.1.2.x and above. This can be done using the following command: setsys PortFailoverEnabled no

Before performing an online upgrade to HP 3PAR OS 3.1.3 MU1, stop HP 3PAR Quorum Witness, and restart it upon completion of the upgrade. To stop and start a quorum witness, follow these steps:

Start-and-stop Quorum witness steps

HP 3PAR Remote Copy Support

Persistent Ports

16

1. Check the status of the quorum with the showrcopy -qw command. For example

2. Before upgrading to HP 3PAR OS 3.1.3 MU1, stop the quorum witness on both arrays with the setrcopytarget witness stop command. For example:

3. Verify that the quorum witness is in the Not-started state with the showrcopy-qw command. For example

If the HP 3PAR Quorum Witness was updated to the latest version and a new virtual machine installed, remove the old HP 3PAR Quorum Witness IP address, and add the new IP address to your configuration. See HP-3PAR Upgrade preplanning guide Page 17. 5. After the upgrade to HP 3PAR OS 3.1.3, restart the quorum witness on both arrays with the setrcopytarget witness start command. For example:

Verify that the quorum witness started with the showrcopy-qw command. For example:

17

Run the following commands from InServ CLI to validate the status of the array.

Note: Cli% symbol before a command means that it needs to be executed from 3PAR CLI

Cli%showversion to verify the current InForm OS level on the array

Cli%checkupgrade to verify if the system is ready to undergo an ONLINE upgrade. Sample output below:

== checkupgrade ==

The current system configuration is valid for proceeding with an online upgrade.

--> The system can undergo an online upgrade.

Return Code = 0

Cli%checkhealth -svc -detail performs a full HealthCheck on the system, if there are any issues noticed (such as failed drive), please notify HP support to fix it before scheduling

the upgrade.

Cli%statcpu -t -d 15 -iter 1 (Applicable only for ONLINE upgrades)

CPU idle time should be more than 50 % during the InForm OS upgrade. As one node in a node pair

reboots at a time during the upgrade, the other node has to handle the entire load during that time. If a

system is already running low on resources, we do not recommend performing an OS upgrade. We

therefore strongly recommend the InformOS upgrade is scheduled during off-peak hours.

Cli%statport -d 15 -iter 1 -ni (Applicable only for ONLINE upgrades)

The I/O load limit recommended for host ports during OS upgrade is as shown below. We therefore

strongly recommend the InformOS upgrade is scheduled during off-peak hours. Port Connection

parameters can be found using the below commands from the StoreServ CLI

Check the host and Partner port connections with showport:

Cli%showport

Then, get the configured rates of the primary and partner host ports both with showport par. We will use the configured rate for the subsequent port I/O load review:

Cli%showport -par

Average I/O and KB/s on each port can be found using the below command from the StoreServ CLI:

Validate 3PAR Array Health

18

Cli%statport -d 15 -iter 1 ni

For each host port, the average I/O - KBytes per sec - should not be greater than:

160,000 KBytes per sec for 4 GB connections

360,000 KBytes per sec for 8 GB connections

The below-mentioned workaround is applicable only to HP Proliant Service Processors, where a reboot is required

(for example applying patch P001 on SP OS 4.3.0, P003 on SP OS 4.2.0, and P011 on SP OS 4.1.0.GA-97 or any

other equivalent patches released for addressing the Bash ShellShock vulnerability) or where the SP has to be

rebooted as a part of troubleshooting. Please contact the 3PAR support team for assistance with procedure

below.

Follow the below-mentioned steps before applying the above-mentioned patches or before issuing the reboot

command on HP Proliant Service Processors (Engage the assigned deployment engineer to follow steps below)

Step 1: Login as root

Step 2: Create a copy of the grub configuration file by using the command cp

/etc/grub.conf /root/grub.conf.orig

Step 3: Create a copy of the inittab file by using the command cp /etc/inittab /root/inittab.orig

Step 4: Edit the current grub.conf file by using the command vim /etc/grub.conf

Step 5: Delete the entries for console within the grub.conf file as shown in the picture below and save the

grub.conf file by using EscKey followed by :wq and exit from the grub.conf file

HP Proliant Model Service Processor

19

Step 6: Edit the current inittab file by using the command vim /etc/inittab

Step 7: Comment out the entry for agetty serial redirection by placing a # symbol at the beginning of the line

shown below so that the line should read as - #co:2345:respawn:/sbin/agetty 57600 /dev/ttyS0 vt100

20

After performing the task involving the SP reboot, follow the steps below to restore the original configuration

Step 1: Once the SP is rebooted, login as root again

Step 2: Copy the original grub.conf file back using the command cp /root/grub.conf.orig /etc/grub.conf

Step 3: Verify the console entries are back in the grub.conf file using the command - cat /etc/grub.conf

Step 4: Copy the original inittab file back using the command cp /root/inittab.orig /etc/inittab

Step 5: Verify the agetty serial redirection entries are back in the inittab file using the command - cat

/etc/inittab

21

Prior to upgrade to HP 3PAR OS 3.1.3 and higher, network speed on the admin ports on the 3PAR array needs to

be set to auto negotiate. This is to avoid the netc process on the 3PAR nodes to hang. This may prevent successful completion of the upgrade or after upgrade prevent netc from starting on some nodes and subsequently cause frequent netc dumps until the issue is addressed.

Follow the workaround to set the speed on admin ports and RCIP ports to auto negotiate

A) Run command below to set auto negotiate before upgrade to 3.1.3.x

1) cli% shownet

IP Address Netmask/PrefixLen Nodes Active Speed Duplex AutoNeg Status

10.0.121.221 255.255.248.0 01 0 100 Full No Active

Default route: 10.0.120.1

NTP server: None

DNS server: None

2) cli% setnet speep auto

Note: On successfully setting auto speed it returns: Interface speed set. If not, keep retrying

few times and/or consult HP-3PAR.

3) cli% shownet

IP Address Netmask/PrefixLen Nodes Active Speed Duplex AutoNeg Status

10.0.121.221 255.255.248.0 01 0 100 Full Yes Active

Default route: 10.0.120.1

NTP server: None

DNS server: None

B) Setting auto speed negotiate on rcip ports

1. Get list of rcip ports:

cli% showrctransport -rcip

N:S:P State HwAddr IPAddress PeerIPAddress Netmask Gateway MTU Rate Duplex

0:9:1 ready 0002AC80041D 10.97.13.32 10.97.13.33 255.255.255.0 10.97.13.1 1500 1Gbps Full

1:9:1 ready 0002AC802ADF 10.98.13.32 10.98.13.33 255.255.255.0 10.98.13.1 1500 1Gbps Full

2. Setting auto speed for rcip

Usage: controlport rcip speed auto

Example:

cli% controlport rcip speed auto 0:9:1

Remote Copy interface change successful

TOP

Set network speed to auto

22


Remote Copy interface change successful


2:9:1 is not configured for remote copy. Click on "Adaptive Optimization" tab-->clcik "Change" next to the InServ to be upgraded

(you need to do it for all Polcy Ids) --> Change "Configuration Active " to False

Validate showpd -path

AO, DO recommendations

23

B) On Node AO:

The data for AO is now maintained on the StoreServ Storage in an internal database. AO configured from external

System Reporter will not work for 3.1.2 systems, region mover log will generate the following error: No license

for Adaptive Optimization.

All existing system reporter based AO configurations will need to be manually recreated through CLI or IMC.

Please refer to the attached procedure -On Node SR & AO - Inform OS 3.1.2 Upgrade Process- to port the existing

System Reporter based AO configuration to 3.1.2 node based AO configuration. Please let us know in advance if

any help is needed from us in migrating the AO configuration after the upgrade

To stop Remote Copy on both primary and DR sites:

stoprcopygroup [option]

stoprcopy [options]

Stopping a Remote Copy group while the group is currently syncing will delay a return of the command.

To stop DO - please cancel all tune tasks (e.g. tunevv)

During HP 3PAR OS upgrade

HP 3PAR OS online automatic upgrades with AIX Direct Connect attached hosts are not supported by HP for any HP 3PAR OS upgrades. Online maintenance upgrades can be successfully performed by means of a node-by-node (or advanced) upgrade method. Using this method will cause the system to wait for user input between each node reboot needed during an online HP 3PAR OS upgrade. This pause time in the upgrade process is used to allow all of the I/O paths to the AIX direct-connect hosts, for all of the presented LUNs, to recover (fail back). To display the paths that must be monitored for recovery, use the AIX lspath s failed command on the attached host: # lspath s failed

Failed hdisk4 fscsi0 Failed hdisk5 fscsi0 ... # After all the paths have recovered, continue with the next node of the HP 3PAR OS. # lspath s failed

IBM AIX

24

iSCSI hosts must have active I/O on all paths during the upgrade. The upgrade process reboots nodes and checks that all hosts have reconnected after each node is rebooted. The upgrade will not advance until all hosts re-establish connection. Since an idle iSCSI host will not reconnect, the workaround is to have I/O on all paths during the upgrade.

As of HP 3PAR OS 3.1.1 GA, online updates can be successfully performed by means of a node-by-node (aka Advanced) update and pausing between HP 3PAR StoreServ Storage node updates. Wait for all I/O paths for all Sun Cluster nodes to recover during each pause between HP 3PAR StoreServ Storage node updates. Refer to I/O Path Monitoring for Solaris page in the Solaris section of the HP 3PAR Operating System Pre-Planning Guide. Link below: HP 3PAR OS Upgrade Pre-Planning Guide

The following HP 3PAR OS online upgrades have been determined to cause HP-UX servers to report device discovery problems until the upgrade is complete:

HP 3PAR OS 3.1.1 (including all MUs) to 3.1.2 MU1, 3.1.2 MU2, or 3.1.2 MU3

HP 3PAR OS 3.1.2 GA to 3.1.2 MU1, 3.1.2 MU2, or 3.1.2 MU3

Issue If a rescan of the HP-UX I/O system hardware is run during the HP 3PAR OS online upgrade process, ioscan might report device paths with a Software State of NO_HW. This is due to a defect that causes an

unexpected SCSI inquiry response to LUN 0 for HP-UX hosts during the HP 3PAR OS online upgrade. It was verified that I/O is not affected and paths remain accessible as long as the server is not restarted during the online upgrade process. When the entire online upgrade process is complete, the SCSI inquiry response to LUN 0 is normal and a rescan reports device paths with a Software State of CLAIMED

iSCSI Configuration

Sun Solaris

HP-UX Troubleshooting

25

If an HP-UX server is restarted during an HP 3PAR OS online upgrade, 1 or more previous known LUN paths might not be discovered by the HP-UX server. The HP-UX operating systems will not continue device discovery when a LUN 0 SCSI Inquiry response returns an unexpected value. Once the upgrade is complete, manual intervention by a system administrator is required for the restarted HP-UX server to rediscover HP 3PAR devices as described in the next section

Resolution NOTE: This issue was resolved with HP 3PAR OS 3.1.3.GA. If you are upgrading to any of the affected HP 3PAR OS versions listed above, follow the steps listed below. During the HP 3PAR OS Online Upgrade process, a rescan of the HP-UX11.31 I/O system hardware reports device paths with a Software State of NO_HW. Perform the following steps to check the health of

the data LUN paths and confirm they are reported as online.

1. Ensure that the HP-UX 11.31 server was not restarted. 2. Scan the HP-UX system hardware and identify the hardware paths for any devices that report a Software State as NO_HW.

3. Verify the health of the data LUN path using the hardware paths from the previous step.

If the LUN paths display as online, I/O is not affected and paths remain accessible as long as the server is not restarted during the online upgrade process.

NOTE: HP-UX OS version 11i v2 (11.23) and earlier do not provide the health check utility, however it has been verified that even though the devices are in the NO_HW state, the online upgrade will resume

26

without I/O interruption. If an HP-UX host is restarted during the online upgrade operation, the following steps must be performed to re-discover 3PAR devices: 1. Ensure all InServ nodes successfully completed the online upgrade and show the correct HP 3PAR OS version

as follows:

2. Scan the HP-UX system hardware for usable I/O system devices: #ioscan --fn

3. If you are using LVM/PVLinks, scan the physical volumes for LVM volume groups and activate them: #vgscan -v --a

Run the following command for each volume group using 3PAR LUNs: #vgchange -a y

4. If you are using VxVM, rebuild the volume device nodes and start the volumes: #vxdctl enable

#vxvol --g start

Verify that the volumes are online:

#vxprint --ht --g

27

Post HP 3PAR OS upgrade

ONLY for 3.1.3.x and higher Upgrades - Persona 15 for Windows 2008 and Windows 2012 Server: New Host Persona 15 for Windows Hosts Needs Set after upgrading to HP 3PAR OS 3.1.3.x and higher to facilitate appropriate LUN behavior. Details: HP 3PAR Windows Server 2012 and Windows Server 2008 Implementation Guide, http://h20628.www2.hp.com/km-ext/kmcsdirect/emr_na-c03290621-14.pdf , See page 8. Previously with HP3PAR OS 3.1.1 and 3.1.2, KB2849097 was required to present the LUNs appropriately. The new Persona 15 presents LUNs in the manner Microsoft Windows expects. Action: After upgrading to HP 3PAR OS 3.1.3.x or higher , HP requires host persona 15 to be used for Windows Server 2008, 2008 R2, 2012, or 2012 R2 to prevent this issue from occurring. Changing to host persona 15 is an online procedure. Please note that if HP Host Explorer application is in use, its mandatory to update to the latest version

before changing host persona 15.Running older version of Host explorer with host persona set as 15 can lead

to an array outage.

Compatibility information for HP 3PAR including Host Explorer is available at: http://h20272.www2.hp.com/Pages/spock2Html.aspx?htmlFile=sw_array_3par.html

Download latest version of host explorer, please visit the following HP URL: https://h20392.www2.hp.com/portal/swdepot/displayProductInfo.do?productNumber=HP3PARHE

Use the HP 3PAR CLI sethost command to change the persona and verify using the showhost command following the upgrade. Cli% sethost -persona 15 windowshost

Cli% showhost

Id Name Persona -WWN/iSCSI_Name- Port

0 windowshost WindowsServer 10000000C9606724 0:5:1

10000000C9606724 1:5:1

TOP

Host Persona

28

1 HP recommends host persona 13 for HP-UX 11i v3 with HP 3PAR OS 3.1.3 MU1 and later. Changing from host persona 7 to host persona 13 can be performed with the host online.

29

2 Use of generic persona allows the host to run Host Explorer on the host. It will automatically create LUN 254 as a SCSI enclosure device for Host Explorer use. 3 Starting with HP 3PAR OS 3.1.3 MU1, host persona 2, which has asymmetric LUN unit access (ALUA), is supported. HP recommends host persona 2 for RHEL 5.8, RHEL 6.1, SLES 10.4 , SLES 11.3 and later with HP 3PAR OS 3.1.3 MU1 and later. Changing from host persona 1 to host persona 2 is an offline process. This change applies to hosts connecting to the array through FC, FCoE, or iSCSI. For more details, see the HP 3PAR host OS implementation guide specific to this OS. 4 HP recommends host persona 2 for RHEL 5.8, RHEL 6.1, SLES 10.4 , SLES 11,3 and later with HP 3PAR OS 3.1.3 MU1 and later. Changing from host persona 1 to host persona 2 is an offline process. 5 VMware ESX 3.5 is not supported in HP 3PAR OS 3.1.1 or later. 6 Host persona 6 will not be supported for any version of VMware ESX/ESXi with future HP 3PAR OS versions after HP 3PAR OS 3.1.3 MU1 and its MUs. HP recommends that customers migrate their VMware ESX configurations on HP 3PAR to Host Persona 11 with HP 3PAR OS 3.1.2 or later. 7 VMware ESX 3.5 is not supported in HP 3PAR OS 3.1.1 or later. 8 As of VMware ESX 4.0, a change of the SATP rules on hosts to enable active/active round robin by default is recommended. See the HP 3PAR VMware ESX Implementation Guide for instructions about changing the VMware ESX/ESXi SATP rules. To obtain this document, see the HP Storage Information Library: http://www.hp.com/go/storage/docs 9 HP recommends host persona 11 for all supported VMware ESX versions with HP 3PAR OS 3.1.2 and later. Changing the persona from 6 to 11 for existing Persona 6 hosts is an offline process. 10 Enabling host persona 2 will enable A/A multipath on LUNs by default. This change will require a restart. Do not change the host persona unless you can restart the Windows Server 2008. If you do not restart the server, the host will lose the mapping of the LUNs. 11 From HP 3PAR OS 3.1.3 MU1, HP requires host persona 15 for Windows Server2012 R2, Windows Server 2012, and Windows Server 2008 R2 and non-R2 hosts. However, host persona 2 is automatically assigned following an online HP 3PAR OS upgrade from OS version 3.1.2. After such an upgrade, you must change host persona 2 to host persona 15. Before changing host personas, see Windows Server OS Guidelines (page 26) for instructions to prevent loss of LUN mapping. 12 Microsoft enables A/A multipathing by default at Windows Server 2008 R2. 13 For Windows Server 2008 R2, configurations of persona 6 and 1 need to be changed manually to persona 2 after an upgrade to HP 3PAR OS 3.1.2. The Windows Server host needs to be shut down when the persona change is performed and then brought back online in order to properly configure the ALUA paths support in persona 2.

.

1. If you are running HP/3PAR software that uses 3PAR CLI, (i.e.: System Reporter, Cluster Extension Software), please ensure that you upgrade the CLI of the hosts to latest CLI client.

You can download the latest CLI and IMC at the following FTP drop box: ftp://imc_cli:[email protected]/ For CLI, Go to 'CLI-SNMP-> (choose the CLI installer for your OS)' For IMC, go to IMC folder

2. Post OS upgrade to 312.MU5,313.x and 321.x, upgrade the following applications to the revision level indicated in any order desired:

EVA to 3PAR Online Support Storage Replication Adapter for VMware depending on your environment. See Storage Replication Adapter for VMware Support Matrix (page 9) for more information. SCVMM UI add-in System Reporter (SR)

Updating GUI/CLI Clients.

30

HP Metrocluster / Serviceguard To download 3par software (e.g. System Reporter), please visit the following HP URL:

http://www.hp.com/go/hpsoftwareupdatesupport

You will be required to create a HP Passport login account and provide the contractual support agreement (SAID).

TOP

Documents

3par Prep Guide