Since I have more than a passing interest in databases, I always try to keep appraised of the Oracle module for NetWorker. It therefore surprised me a few days ago to see that v5 of the module had been released in March. I guess my excuse is that March was an insanely busy month for me between work and travel. (Well, that’s my excuse, and I’m sticking to it.)
So yesterday I downloaded v5 of the module (for Linux), and spun it up. This is a version I really, really like.
Now, here’s a few bullet points before I get to the most impressive feature:
- No longer supports Oracle 9i or lower; if you want older, unsupported versions of Oracle you have to use an older version of the module.
- Requires features that exist only in Networker 7.5.x as the underlying client.
- Must have the NetWorker regular client installed and running in order for the module software to correctly install and activate,
- Can work with the 7.4.x NetWorker server with the exception that what I’m about to describe below doesn’t work with a 7.4 server.
- Now has a client configuration wizard that works within NMC and makes Oracle backup configuration a breeze.
Honestly, if you’re about to do a new NetWorker install into a site that has Oracle, skip everything else and install 7.5.1. I.e., this is one of these compelling reasons for 7.5.x.
The Oracle client configuration wizard is integrated into NMC’s wizards. Right-click on a client in the configuration panel, choose “Client Backup Configuration -> New”, and you’re off and running:
Note that you won’t reach this point if you’ve disabled ‘nsrauth’ authentication on the backup server. I had done so on my lab server as a test on Monday, and spent half an hour trying to work out a … rather inexact … error message.
The above step is where things get fun. Note that if you are given these details, you don’t even need to log onto the client to setup an nsrnmo script any longer. This is the start of A Really Good Thing.
Also, I should note, in the above screen shot, because I was using a temporary database installed just for a few tests and I was in a rush, I used the sys account for connecting to the target database. No, you shouldn’t ever do that – create a backup user and use that account, please.
Note that Oracle, and the Oracle Listener, must both be running on the client in order to clear the above step.
After the above, we then start to get into the ‘regular’ client configuration options:
This summary screen shows you what you’re going to get as far as the configuration is concerned – including the RMAN script that has been automatically generated for you:
Confirmation of sweet success:
The finished client in NMC:
Once configured, you’re ready to start backing up straight away. Honestly, it couldn’t be simpler.
As a closing note, I know some other backup products have had Oracle backup wizards for some time, so I’m not claiming EMC is the first with this style of setup, but I do think it’s a great feature to see included now.
I want to do some RMAN backup testing using networker 7.5 and NMO 5.0. I am doing these testing on VMware. I am not sure whether I need to buy some tape drive for my home PC to do backup testing or I can mimic hard drive as tape drive in VMWARE.
Do you have any idea? any suggestion?
Thanks..
You don’t need a tape drive to test RMAN backups. Even though NetWorker configures channels of type SBT_TAPE, these will work equally to a disk backup unit.
The way the RMAN interface works is that it provides a native SBT_DISK type channel, which is used for Oracle backups direct to disk on locally attached storage as regular filesystem copies on the Oracle server.
RMAN then allows 3rd party backup software (such as NetWorker) to hook in and provide an alternate location for the backup through the SBT_TAPE option.
Thanks for response Preston…
Basically I am evaluating nerworker 7.5 and NMO5.0 software. As I don’t have the tape device and want to some fake tape device to networker. I tried to present fake tape drive with one text file name as “F:Tapevolume” and select media type optical but then rman get stuck at following line…
Recovery Manager: Release 10.2.0.1.0 – Production on Tue May 26 22:51:58 2009
Copyright (c) 1982, 2005, Oracle. All rights reserved.
RMAN> connect target *
2>
3> RUN {
4> ALLOCATE CHANNEL CH1 TYPE ‘SBT_TAPE’;
5> BACKUP
6> FULL
7> FORMAT ‘%d_%U’
8> DATABASE
9> INCLUDE CURRENT CONTROLFILE
10> PLUS ARCHIVELOG
11> ;
12> RELEASE CHANNEL CH1;
13> }
14>
connected to target database: D801 (DBID=1863601699)
using target database control file instead of recovery catalog
allocated channel: CH1
channel CH1: sid=12 devtype=SBT_TAPE
channel CH1: NMO v5.0.0.0
sent command to channel: CH1
Starting backup at 26-MAY-09
current log archived
channel CH1: starting archive log backupset
channel CH1: specifying archive log(s) in backup set
input archive log thread=1 sequence=1110 recid=182 stamp=683675650
input archive log thread=1 sequence=1111 recid=183 stamp=683750275
input archive log thread=1 sequence=1112 recid=184 stamp=683840777
input archive log thread=1 sequence=1113 recid=185 stamp=683848847
input archive log thread=1 sequence=1114 recid=186 stamp=684196996
input archive log thread=1 sequence=1115 recid=187 stamp=685494305
input archive log thread=1 sequence=1116 recid=188 stamp=685494423
input archive log thread=1 sequence=1117 recid=189 stamp=687901787
input archive log thread=1 sequence=1118 recid=190 stamp=687911147
input archive log thread=1 sequence=1119 recid=191 stamp=687911609
input archive log thread=1 sequence=1120 recid=192 stamp=687912724
channel CH1: starting piece 1 at 26-MAY-09
I am getting alert message “waiting for 1 writable volumes to backup pool”…any help appreciated…Thanks so much..
The device type for disk backup units is not optical, but ADV_FILE. (You may have to delete and recreate the device in order to change the type, but you can certainly try just changing the type from optical to ADV_FILE.)
After you’ve changed the type or created the device as ADV_FILE, you need to label and mount a “volume” in the device. This can be done using NMC or from the command line. Based on the RMAN script you’ve come up with, I’d suggest labeling into the “Default” pool.
Please ignore…my bad..I did not mount it..
Thanks Preston so much for your help. It worked after mounting that file.
Hi Preston,
One more problem, I noticed, please let me kow if you have any idea regarding this issue.
We always see level fulls reported regardless of the level we run in RMAN while RMAN catalog shows correctly. Is there any way to take this and show RMAN backup type and level correctly.
C:Documents and SettingsHP_Administrator>mminfo
volume client date size level name
ayushi.001 ayushi 5/26/2009 556 MB full RMAN:D801_1rkg1fe5_1_1
ayushi.001 ayushi 5/26/2009 3119 MB full RMAN:D801_1skg1ffb_1_1
ayushi.001 ayushi 5/26/2009 7169 KB full RMAN:D801_1tkg1fkh_1_1
ayushi.001 ayushi 5/26/2009 257 KB full RMAN:D801_1ukg1fkr_1_1
ayushi.001 ayushi 5/27/2009 559 MB full RMAN:C:oraclermanrman_full.rcv
ayushi.001 ayushi 5/27/2009 3119 MB full RMAN:C:oraclermanrman_full.rcv
ayushi.001 ayushi 5/27/2009 7169 KB full RMAN:C:oraclermanrman_full.rcv
ayushi.001 ayushi 5/27/2009 257 KB full RMAN:C:oraclermanrman_full.rcv
ayushi.001 ayushi 5/27/2009 562 MB full RMAN:D801_23kg3nle_1_1
ayushi.001 ayushi 5/27/2009 3119 MB full RMAN:D801_24kg3nmv_1_1
ayushi.001 ayushi 5/27/2009 7169 KB full RMAN:D801_25kg3ns5_1_1
ayushi.001 ayushi 5/27/2009 257 KB full RMAN:D801_26kg3nsg_1_1
ayushi.001 ayushi 5/27/2009 563 MB full RMAN:incr_D801_27kg3p6t_1_1
ayushi.001 ayushi 5/27/2009 3119 MB full RMAN:incr_D801_28kg3p8f_1_1
ayushi.001 ayushi 5/27/2009 7169 KB full RMAN:incr_D801_29kg3pej_1_1
ayushi.001 ayushi 5/27/2009 257 KB full RMAN:incr_D801_2akg3pes_1_1
ayushi.001 ayushi 5/27/2009 564 MB full RMAN:incr_D801_2bkg3qb6_1_1
ayushi.001 ayushi 5/27/2009 1025 KB full RMAN:incr_D801_2ckg3qcn_1_1
ayushi.001 ayushi 5/27/2009 7169 KB full RMAN:incr_D801_2dkg3qgv_1_1
ayushi.001 ayushi 5/27/2009 257 KB full RMAN:incr_D801_2ekg3qh8_1_1
C:Documents and SettingsHP_Administrator>
The other thing we notice that NetWorker only reports the rman script name as the save set name when nsrnmostart program executed from client side, so all the save sets have the same name. We’re running Oracle 10g with NW 7.5.1 and NMO 5.0. Any solution for this issue??
To have the level reported correctly, you need to set the level both within the RMAN script and the schedule for the client.
As to the name of the saveset on server initiated backups vs client initiated backups – I am not sure how you’d force a particular saveset name from the client. (You might want to see whether the nsrnmo command will accept a -N argument, which allows the specification of the saveset name.)
(It’s actually been a little while since I’ve run up multi-level Oracle backups, and my Oracle test environment is currently shutdown while I work on other activities at the moment, so I’m afraid I can’t currently 100% verify these responses.)
Preston
Save set names for NMO backups always begin with RMAN: and I believe the RMAN backup script is the one who formats whatever goes after the colon.
(Actually under older versions of the module the saveset name, as defined in the client resource, was usually the name of the RMAN script to be invoked.)
I have Oracle databases that run within packages in a ServiceGuard (active/passive) cluster.
If you follow the NW/NMO documentation you need a nsrnmo backup script + rman script per instance on each node – with multiple nodes and multiple instances it is a nightmare keeping all nodes sync’ed.
Is this stil valid for v5, has any of you had a look at what is build and where in the Oracle clients?
I’m on NMO 4.5 + NW7.5.1 and my solution (that I think will be good for v5 as well if the above is true) is that I custom build a unique nsrnmo script where I can pass parameters (very similar to the save command) and thus only need to put one single script on each node + the rman script under the package mount point, and then send any specific instance parameters to it.
You still need an RMAN script per client instance, but using the wizard I’ve not needed to construct an nsrnmo script on the tests that I’ve done, which is at least one improvement. (And obviously, using the wizard, you have the option of not having to craft the RMAN script yourself…)
I don’t believe that the wizard/integration functionality in v5 of the module is going to be all things to all people – some environments are going to find the wizard perhaps a bit limiting – but that’s the nature of wizards/automated setup utilities.
I guess the other answer is that just by being on the v5 module, you don’t have to use the wizard – I’ve also done tests using a more conventional backup configuration (with both the nsrnmo and RMAN scripts manually crafted), so going to v5 of the module doesn’t mean you have to abandon/completely change your current setup. This means you can upgrade to v5 of the module then generate some sample instances using the wizard to see if you’re personally satisfied with what it offers and whether it provides you configuration simplification.
I know it’s a bit of a sitting-on-the-fence answer, but anything to do with wizards end up being a bit like that – what works for some won’t appeal to others, etc. It is however very gratifying to generate a configuration with such little effort…
Preston,
Thank you for exposing our new configuration wizard on your BLOG. One feature that is worth mentioning…once you have moved through the wizard panels and selected the options you want, we allow you to edit the generated RMAN script within the wizard itself. We provide this function to support some of the more obscure options of RMAN that customers may be using.
We also provide an Oracle recovery wizard and the ability to modify an existing configuration through the wizard.
Cheers,
Skip
NetWorker Usability
Hi Preston.
To be able to create the saveset by browsing is neat. But what if you have environment that is segmented , that is many subnets? We get this when using this feature “failed to contact using UDP ping” when creating new clients with ther wizard. Is there a way to use the wizard in a firewalled environment / distributed environment?
Johannes
I’ve not used this particular feature in a firewalled environment. However, have you tried in NMC to turn off the option “RPC ping via UDP when connecting to NetWorker”? If you go into Setup -> Setup -> System Options, you’ll find it as the last option in the dialog box. It’s meant to be used for firewalled environments, and perhaps it applies to clients as well as the server.
Johannes
You need to take a look at NetWorker’s admin guide. It has an appendix on firewall configuration.
There is also a white paper on EMC website about it.