Subscribe

Thoroughly confused with SnapVault, Ops Manager, and the NMC

Hi all,

I'm trying to implement Snapvault to our 2050's (secondary) from our primaries (3050's and OSSV clients). I've tried a number of ways to create the relationships and am just not happy with any method yet. I have a couple of questions if you can bare with me...

I first tried to use Proection Manager via the Netapp Management Console. I created Datasets, protection policies and schedules. I put various OSSV clients in to a dataset and then applied a protection policy and a schedule. Essentially it worked but I'm just not happy with Protection Manager. It creates hideous snapshot names and the snapvault relationships seem to be recreated every night rather than performing a simple backup.

My biggest gripe, however, is that nothing ties in to Ops Manager. When logging on to the Ops Manager website datasets are non-existent and even schedules found there have nothing to do with those set in Protection Manager.

I was going to leave Protection Manager for now (possibly importing the relationships later) and use Ops Manager to create the SV relationship. With Ops Manager, at least, it seems to have schedules that you can apply to new SV relationships. Only thing is...when I try and create a relationship the schedule option is greyed out and I can't see anyway of applying a schedule to a relationship once it's been created.

I love the command line, so it really doesn't bother me having to use it. But what I'm trying to get is an easy to manage interface for other sys admins that aren't command line lovers. I am going to have a load of Vols on the 3050's to backup and 50+ OSSV clients (and I want to have a vol purely for System States and C drives to allow for good de-dupe ratios). Having to do EVERYTHING by the command line puts even me off.

Everything seems totally disparate from the command line, to Ops Manager, to Protection Manager. I feel like I'm missing a part of the puzzle and if someone could  explain it suddenly everything would make sense!

Re: Thoroughly confused with SnapVault, Ops Manager, and the NMC

Argh! I knew this would happen...as I soon as I post it I find the answer to one of my questions.

I've figured out how to apply Ops Manager schedules to an existing volume (containing SV secondary snapshots) but I'm still confused as to why it doesn't reflect what you do in Protection Manager.

Re: Thoroughly confused with SnapVault, Ops Manager, and the NMC

Okay, I'm well aware of looking like a complete loon having a conversation with myself, but I'll risk it...

I seem to be making progress with Ops Manager, but the schedule I've applied to the volume does not reflect what the command line says if I run snapvault snap sched.

Via the web interface I've created a schedule for daily 8am snapvault snapshots. However, this is what the command line shows:

filer> snapvault snap sched
create OSSV_Test_System  0@-

Is this normal?

Re: Thoroughly confused with SnapVault, Ops Manager, and the NMC

Hi David,

I'll jump in to rescue you from this conversation with yourself.

I don't use protection manager today but have a good understanding of how the product works.

Protection Manager makes all its policy actions via API calls to the Filer.

For example OM server clock reaches a certain time and Protection Manager wakes up and tells a device to perform a snapvault operation. So the OM server is technically the scheduler in this setup as it does not apply a schedule to the Filer for the Filer to autonomously perform the operation, it rather instructs that filer to perform the operation now.

I agree that this makes for a confusing data protection experience.

Richard

Re: Thoroughly confused with SnapVault, Ops Manager, and the NMC

Hi David,

OpsMgr does not push the Snapvault or OSSV schedule to the filer.

OpsMgr maintains the schedule and issues the updates itself at the appropriate time.

For SnapMirror, OpsMgr does put the schedule on the secondary filer.

OpsMgr write the SnapMirror schedule into the /etc/snapmirror.conf file on the secondary filer.

Re: Thoroughly confused with SnapVault, Ops Manager, and the NMC

Thanks for your posts guys. That certainly helps clear things up a little.


My two questions now would be...

  1. If OM is the scheduler should I use snapvault snap unsched to remove any schedules that exist on the filers?
    e.g. one Snapvault secondary volume has this schedule (which seems to have been created automagically)...
    create OSSV_Test_System  0@-

  2. How does the scheduler within PM compare to OM? Does it work on the same principle as OM? ...does PM act as the scheduler to call the SV commands? What's confusing is that schedules/template schedules don't match up at all. Also OM and PM seem to be totally unaware of each other and each others' schedules. Am I correct to assume this?

I'm wondering out loud here: how many people actually use PM? Seems to be more hasstle than its worth!

Re: Thoroughly confused with SnapVault, Ops Manager, and the NMC

Hi David --

What version of ProtMgr were you using?  3.8.1 was just released the other day and I strongly recommend upgrading to it.

We should not be re-creating relationships every night.  They should get created once and updated whenever you schedule them.  If they're being re-created, there's a problem and you should contact NGS.  Most likely, we think your destination aggregate is over-full and we'll need to figure out why.

-- Pete

Re: Thoroughly confused with SnapVault, Ops Manager, and the NMC

Hi David --

Now I'm the one doing all the talking :-).

You can leave the snapvault snap sched around.  We create that because SnapVault needs it to be there when we create the archival snapshot.  For the most part, you can ignore it.

Regarding schedules, we think the ProtMgr schedules are more flexible than the ones in ONTAP.  We also let you set one schedule and we'll take care of creating the snapshots on both the source and destination.  And if you set the schedule using ProtMgr, you don't need to modify all SnapVault relationships should you decide to change the schedule.

-- Pete

Re: Thoroughly confused with SnapVault, Ops Manager, and the NMC

Thanks Pete. I'm was so close to closing this thread and walking away happy, but - guess what...?

So I've now created a bunch of SV relationships via the Ops Manager interface. Obviously I had to create a seperate relationship per local drive for each Wintel box and another for the System State. I was happy enough. The jobs showed in the Ops Manager web interface and also on the command line.

I was about to import these relationships in to Protection Manager and noticed that a load were not showing.

This is the output from the command line...

filer01> snapvault status
Snapvault is ON.
Source                       Destination                                              State          Lag        Status
wintel01:D:/                 filer01:/vol/OSSV_Test_Data/wintel01_D__                 Snapvaulted    22:50:59   Idle
wintel01:E:/                 filer01:/vol/OSSV_Test_Data/wintel01_E__                 Snapvaulted    22:50:50   Idle
wintel02:D:/                 filer01:/vol/OSSV_Test_Data/wintel02_D__                 Snapvaulted    22:50:37   Idle
wintel02:E:/                 filer01:/vol/OSSV_Test_Data/wintel02_E__                 Snapvaulted    22:50:29   Idle
wintel03:D:\                 filer01:/vol/OSSV_Test_Data/wintel03_D__                 Snapvaulted    22:50:13   Idle
enterprisevault:D:\          filer01:/vol/OS_EnterpriseVault/enterprisevault_D__      Snapvaulted    04:26:36   Idle
wintel01:C:\                 filer01:/vol/OS_SystemState/wintel01_C__                 Snapvaulted    04:17:30   Idle
wintel01:SystemState         filer01:/vol/OS_SystemState/wintel01_SystemState         Snapvaulted    26:56:51   Idle
wintel02:C:\                 filer01:/vol/OS_SystemState/wintel02_C__                 Snapvaulted    04:17:17   Idle
wintel02:SystemState         filer01:/vol/OS_SystemState/wintel02_SystemState         Snapvaulted    26:54:09   Idle
wintel03:C:\                 filer01:/vol/OS_SystemState/wintel03_C__                 Snapvaulted    04:17:07   Idle
wintel03:SystemState         filer01:/vol/OS_SystemState/wintel03_SystemState         Snapvaulted    26:53:57   Idle
enterprisevault:SystemState  filer01:/vol/OS_SystemState/enterprisevault_SystemState  Snapvaulted    26:53:32   Idle
wintel04:SystemState         filer01:/vol/OS_SystemState/wintel04_SystemState         Snapvaulted    26:53:27   Idle

But within Ops Manager and Protection Manager (using the Netapp Management Console) I can only see the following jobs when I go to Data > Unprotected Data > External Relationships...


wintel01Smiley Very Happy:/

wintel01:E:/

wintel02Smiley Very Happy:/

wintel02:E:/

wintel03:E/

wintel04:SystemState

Aargh! As I type wintel04:SystemState has disappeared off the list of available external relationships, yet it is still visible on the command line of the filer!

I don't get it. Either I've seriously under estimated how difficult this is to set up or something is going wrong!

Do you only have a certain amount of time after manually creating relationships to import them in to Protection Manager? I would have thought they'd always be visible......

By the way we're using DFM 3.8.1 (upgraded from 3.8)

Re: Thoroughly confused with SnapVault, Ops Manager, and the NMC

Aargh! As I type wintel04:SystemState has disappeared off the list of available external relationships, yet it is still visible on the command line of the filer!

I don't get it. Either I've seriously under estimated how difficult this is to set up or something is going wrong!

Do you only have a certain amount of time after manually creating relationships to import them in to Protection Manager? I would have thought they'd always be visible......

No, that's odd.  Try running "dfbm primary dir list" from the CLI (or look for the relationships in the Backup tab from OpsMgr).  I'm guessing we're somehow losing the relationship out of our database, but I don't know why that would be.  You could also scan .../DFM/logs/dfmmonitor.log to see if any errors are showing up.

No, there's no timer -- unprotected data stays that way forever unless you do something about it.  What's in that table should match what's in our database, which should match the OpsMgr GUI and the CLI.