Another article on how cross cell replication is working in HPE Data Protector. Jim wrote and posted this nice document.
Category Archives: HowTo
Wie geht das? Knowledgebase, HowTo, RTFM, …
Step by Step: Power On and Live Migrate from 3PAR Snapshots
After I released the article https://www.data-protector.org/wordpress/2015/08/tested-poweronrun-live-migrate-3par-snapshots-smartcache/, some of you asked for a more detailed documentation regarding “Power On and Live Migrate from 3PAR Snapshots or Smartcache”, the new features introduced in Data Protector 9.04. To use Zero Downtime Backup you can refer to the guides ZDBAdmin.pdf and ZDBIntegration.pdf. In addition in the document Integration.pdf on pages 717, 719 and 748 additional information can be found.
A documentation with screenshots for “Power On from 3PAR Snapshots” you can find here: http://www.syncer.de/?p=795. Thanks to Sebastian for the detailed documentation.
TESTED: PowerOn&Run and Live Migrate from 3PAR Snapshots or SmartCache
With Data Protector 9.04 the new features PowerOn & Run and Live Migrate from 3PAR snapshots or SmartCache were introduced. This feature allows to power on and test virtual machines (VMware), backed up using the VE integration. Recently I had the opportunity to implement this feature in a proof of concept and to present the solution to the customer. I have also been able to talk with some partners and customers meanwhile and bottom line is: a successful implementation! Some discussions around these features lead to the question whether Veeam is still required as supplementary backup solution, as Data Protector provides one more time an another attractive recovery option for virtual machines in Enterprise environments.
In the mentioned PoC I had performed the backup of multiple virtual machines into the SmartCache. In addition I backed up some machines using Zero Downtime Backup and implemented it into the customers backup and recovery strategy. PowerOn and Migrate worked perfectly when I used 3PAR snapshots and the SmartCache and I encourage all customers to test these new Data Protector features. If no 3PAR in place, no problem, you can perform the test with the SmartCache as I did in addition.
Speaking perfectly, two problems raised up. In one case, an error message when turning on the virtual machine was displayed: “A question needs to be answered before power on can be completed” and as a result “Error powering on”. In the second case, an error message came: “Share presentation failed” and as a result “Power On Virtual Machine failed”. In both cases, however, Data Protector has shown only a consequence of failure of another cause. Before anyone comes across the same error, follow the explanation of the cause.
“A question needs to be answered before power on can be completed”
Using Virtual Center, the unanswered question was displayed with the comment “Diese virtuelle Maschine wurde möglicherweise verschoben oder kopiert…” (German message). In debug files following was seen:
[ 99] 2015-08-19 14:23:46.463 ("/integ/vep/vepa/Plugins/Vmware/VmwareHelpers/ConfigUtil.cpp $Rev: 49146 $ $Date:: ":293)
[ 99] ===>> (6) ConfigUtil::answerQuestion {
[ 99]
[ 20] [ConfigUtil::answerQuestion] waiting for different question
_vmx1
Diese virtuelle Maschine wurde möglicherweise verschoben oder kopiert. Um bestimmte Verwaltungs- und Netzwerkfunktionen konfigurieren zu können, muss VMware ESX wissen, ob diese virtuelle Maschine verschoben oder kopiert wurde. Wenn Sie es nicht wissen, antworten Sie mit "Ich habe sie kopiert".
[ 20] [ConfigUtil::answerQuestion] Throwing exception: class std::runtime_error
Hold On! German error message in Data Protector? The Virtual Center Server definitely was installed with English Locale, however, the service and user was not set to English Locale. So wondering why I translated this topic for you? The error might appear to you too, especially when not located in native English speaking countries. With the link (Thanks to Sebastian): http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=2121646 the cause was fixed very soon and PowerOn and Migrate worked as expected.
“Share presentation failed”
[Normal] From: RSM@hpcs.customer.local "" Time: 8/10/2015 9:54:59 AM Restore session 2015/08/10-10 started. [Normal] From: RMA@hpcs.customer.local "SmartCache_gw1 [GW 18252:0:17740599057275740933]" Time: 8/10/2015 9:55:38 AM STARTING Media Agent "SmartCache_gw1 [GW 18252:0:17740599057275740933]" [Normal] From: RMA@hpcs.customer.local "SmartCache_gw1 [GW 18252:0:17740599057275740933]" Time: 8/10/2015 9:55:39 AM Loading medium from slot \\hpcs.customer.local\D:\Cache\358d640a_55d57510_0908_13a1 to device SmartCache_gw1 [GW 18252:0:17740599057275740933] [Major] From: RMA@hpcs.customer.local "SmartCache_gw1 [GW 18252:0:17740599057275740933]" Time: 8/10/2015 9:55:45 AM Share presentation failed. [Major] From: OB2BAR_VEAgent@hpcs.customer.local "/DATACENTER" Time: 8/10/2015 9:55:45 AM Received ABORT request from RSM (ERR: Error starting backup/restore (BMA cannot be started or similar).) [Major] From: VEPALIB_VMWARE@hpcs.customer.local "/DATACENTER" Time: 8/10/2015 9:55:45 AM Preparation of replica vm 'virtualmachine' for restore failed ... [Critical] From: VEPALIB_VMWARE@hpcs.customer.local "/DATACENTER" Time: 8/10/2015 9:55:45 AM Power On Virtual Machine failed.
After some debugging I found that port 111 was in use ( nestat -ano). A look at the process list for the given PID indicated the responsible process – ONC / RPC Portmapper. The process (from QLogic’s OneCommand) occupied the port 111 and thus prevented a start of the NFS service. Solution: Disable the service ONC / RPC portmapper, start NFS service, and immediately the presentation to the ESX server is working.
By the way: to purge the powered on machines the command omnidbutilwas enhanced. With command omnidbutil -purge_expired_poweron_vms -daily you can cleanup the environment.
Five Minutes to run your Data Protector Installation Server on Linux
In order to install the Data Protector client software on Linux based systems and on HP-UX, you need to have a server with the installation server package installed. With an installation server you are able to distribute the software onto new clients or to upgrade existing clients (push installation). I use a virtual machine with a basic installation and no graphical interface in these cases. The installation server is supported on RedHat, CentOS, Oracle Enterprise Linux and SUSE Linux Enterprise Server (details and version information can be found in matrix: Platform_Integrtn_SptMtx.pdf). With the documentation below I show you how to install or upgrade the Data Protector Linux Installation Server. This includes the removal of an older major version and will not take more than five minutes in total.
Remove older Data Protector version:
cp /opt/omni/.omnirc /tmp/.omnirc.sav rpm -qa | grep -i ob2 > /tmp/removeOB2 for PACKAGE in `cat /tmp/removeOB2`; do rpm -e $PACKAGE --nodeps --noscripts; done; cp /etc/services /tmp/services.bak cat /etc/services | sed -e "s/omni/\#omni/" > /tmp/services.new cp /tmp/services.new /etc/services rm /etc/xinetd.d/omni rm -r /etc/opt/omni rm -r /opt/omni rm -r /var/opt/omni rm /tmp/removeOB2 rm /tmp/services.new
Comment: This is not the recommended way to uninstall Data Protector, but the fastest method. Keep in mind that Data Protector site specific patches or other hotfixes may not have been installed with the “OB2” string in the package description.
Install new Data Protector version (Installation Server only):
mkdir /dpinstall (*) mkdir /dpinstall/dp9 mkdir /dpinstall/dp9pb copy ISO file to /dpinstall (**) copy patch bundle file to /dpinstall (***) mount -o loop /dpinstall/TD586-15021.iso /dpinstall/dp9 /dpinstall/dp9/LOCAL_INSTALL/omnisetup.sh -IS umount /dpinstall/dp9 tar -xvf DPLNXBDL_00902.tar -C /dpinstall/dp9pb/ cd /dpinstall/dp9pb ./omnisetup.sh -bundleadd b902 cd / rm -r /dpinstall cp /tmp/.omnirc.sav /opt/omni/.omnirc (****)
* or any other folder with sufficient space
** e.g. TD586-15021.iso – Data Protector 9.00 Linux – use winscp to copy the file
*** e.g. DPLNXBDL_00902.tar – Data Protector 9.02 Linux – use winscp to copy the file
**** if file exists – see remove older version of Data Protector
In the example it is assumed that a patch bundle needs to be installed in addition. If no patch bundle is available the steps 3, 5, 9, 10 and 11 are not required. After the installation it is recommend to check the content of the file .omnirc; at least the variable OB2_SSH_ENABLED=1 in file /opt/omni/.omnirc is required in order to distribute new clients using Data Protector GUI. If the file does not exist you can create it from template – cp /opt/omni/.omnirc.TMPL /opt/omni/.omnirc. If additional updates are available copy the patches onto the server and decompress using command tar. A rpm package and text file will be generated. Within the text files the command to install the patch can be found. Please keep in mind, Data Protector patches will be installed using a special order; for details see https://www.data-protector.org/wordpress/2013/06/basics-installation-order-patches/.
Register Installation Server in Data Protector Cell:
omnicc -export_is <installation server> (*) omnicc -import_is <installation server> (**)
* This step is required only if the server was already an installation server
** Use the name of the server (FQDN) for “installation server”
If this installation server was already used in the past to install the Data Protector client software, you now can use it to upgrade your existing client using the Data Protector GUI. If this installation server is your first installation server you need to configure SSH based authentication first.
Create public/private rsa key pair:
ssh-keygen (*) ssh <client - short name> (**) ssh <client - FQDN name> (**) ssh <client - IP address> (**) Add public key (***)
* Answer the questions to generate the “key pair”
** You can cancel the login once you accepted the host key for the client
*** Use ssh to login into the new client and add the public key from the installation server (cat /root/.ssh/id_rsa.pub) into the file /root/.ssh/authorized_keys
It is assumed that defaults will be used when using the command ssh-keygen. Furthermore it is assumed that all defaults were used for the SSH configuration and paths.
EADR and Tape Block Size
In older articles I often wrote about Bare Metal Recovery, a feature included in Data Protector without any additional costs and called Enhanced Automated Disaster Recovery (EADR) – see https://www.data-protector.org/wordpress/?s=eadr. To recover the Cell Server (see link) I always told to use tape devices with a block size of 64K. With Data Protector 9.0X and Windows 2008 / Windows 2012 this is no longer required, the limitation is no longer listed in the Data Protector handbooks. Sebastian Köhler (see http://www.syncer.de) has tested EADR for a Cell Server with a higher block size. He made the tests with SAS / FC libraries, StoreOnce and file libraries. So, you no longer need separate tape drives to backup your Cell Server, this makes the administration more easier.
And an additional comment from Sebastian, the parameter -anyobj for the command omniiso is back and available with Data Protector 9.0X. For additional details, please review the article https://www.data-protector.org/wordpress/2014/04/eadr-cell-server-dp-8-1x-windows-2012-r2/.