Ah… 4 december the day SMO 3.0 was released to the public. I couldn’t help it, I had to free up a machine and start testing it. Netapp announced some major updates in this major release.

  • built-in scheduler
  • built-in support for snapvault/snapmirror
  • new GUI

I expected a lot from this release because, face it, SMO 2.2 was lacking in professionalism. It looked ugly, felt clumsy but it did its job. Some people still are doubting the added value of a tool like SMO, we’ll see if netapp can change their mind with 3.0.

 

Testsystem

My tests are done on Oracle Enterprise linux 5.2 with latest patches. I’ll snapshot a 11.1.0.7 database. The SMO repository will also exist inside a 11.1.0.7 database.All mounts are done over NFS.

Installing

Right out of the box snapdrive and snapmanager install flawlessly. What could catch you off guard is that snapdrive 4.1 comes with https enabled out of the box. So if your filer doesn’t support https, you need to change the snapdrive.conf file so it uses http to connect.

GUI

Oh i like it already, the JRE is part of the startup process this time and gets downloaded and installed automatically. Still not working with java 6 though. It even works in firefox, finally :)

Ok, first startup… looks better but still just a sterile look. On unix though, my subwindows are just empty. Could be a problem with my machine, cause i get the same effect with dbca from oracle. so let’s give SMO the benefit of the doubt. Ok, starting my virtual machine to test in windows. All seems fine. Time to create the repository and let’s use the wizard from the gui for once ( normally i’m command line minded, but in this case the gui needs to be tested thoroughly ;) )

ORA-17002 : Ouch, IO exception while connecting with jdbc to the database. Only 1 entry on metalink and talks about hostresolving on RAC, not the case here and hostresolving is set correctly.Seems my tests are temporary to a halt while i figure this out. I hope the 11G repository is not the problem. Tune in later for more ;)

Back after a short intermezzo with a workaround. On command line I have no issues connecting with jdbc and creating the repository

smo repository create -repository -dbname SMO -host ls-alfresco -login -username SMO -port 1521 Enter password for database connection SMO@ls-alfresco:1521/SMO: *** [ INFO] SMO-20019: Set password for repository “SMO@SMO/ls-alfresco:1521″ in user credentials for “root”. [ INFO] SMO-09202: Creating new schema as SMO on jdbc:oracle:thin:@//ls-alfresco:1521/SMO. [ INFO] SMO-09205: Schema generation complete. [ INFO] SMO-09209: Performing repository version INSERT. [ INFO] SMO-09210: Repository created with version: 62 [ INFO] SMO-13048: Repository Create Operation Status: SUCCESS [ INFO] SMO-13049: Elapsed Time: 0:00:09.084

It also works with a colleague on windows, so it seems there is an issue isolated to my machine. I reran netca, tested the connection and now it worked. Later I ran into a problem with SOAP when i tried to create a profile. I should warn you to checkwhich version of java your OS uses to launch the java application. Make sure 1.6 is either not installed or not enabled in the java control panel. My problems were fixed after disabling java 1.6.

The gui now exists out of 3 parts. Monitoring, Scheduling and backups/clones. They are easy accessible using the tabs on top. The monitoring tab looks like the old GUI screen but now we see a reports section in the sidebar where we can see all operations in a certain time range ( last hour, last month, … ). The scheduling tab lets you.. well… schedule things and the backup/clone tab shows a detailled list of backups and clones.

Profiles

I’m glad to see SMO is now working with retention classes and can be configured for both duration and count. One of the major changes is that 1 profile now can have multiple retentions. The defaults are adjustable in smo.conf. No more creating 4 profiles for every database for hourly, daily, weekly, monthly.Also new is that smo checks your database for a valid restore profile and if your database can be restored using fast restore. It not only says if its possible or not, you also get a reason why. For example :

–[ WARN] SMO-05071: Database profile ALFRESCO is not eligible for fast restore:  Restore Plan: Preview: The following files will be restored completely via: storage side file restore /alfresco/DB/oradata/alfresco01.dbf /alfresco/DB/oradata/nagios01.dbf /alfresco/DB/oradata/sysaux01.dbf /alfresco/DB/oradata/system01.dbf /alfresco/DB/oradata/undotbs01.dbf /alfresco/DB/oradata/users01.dbf

 

Analysis: The following reasons prevent certain files from being restored completely via: storage side file system restore Files in file system /alfresco not part of the restore scope will be reverted. Files not in restore scope: /alfresco/DB/flash_recovery/ALFRESCO/archivelog/2008_12_05/o1_mf_1_548_4ml7jmjs_.arc /alfresco/DB/oradata/control02.ctl /alfresco/DB/oradata/redo02b.rdo Files to restore: /alfresco/DB/oradata/alfresco01.dbf /alfresco/DB/oradata/nagios01.dbf /alfresco/DB/oradata/sysaux01.dbf /alfresco/DB/oradata/system01.dbf /alfresco/DB/oradata/undotbs01.dbf /alfresco/DB/oradata/users01.dbf

So let’s bring our database over to the netapp so we can test further.

 

This concludes part 1 :)