CMMDB - Path away from it?

I inherited an environment with 2 Cell Managers. One acting as MoM for centralized licensing and CMMDB to make life interesting and the other cell manager at another site is tied to it.

I've had 2 outages recently of the CMMDB cell manager and it's caused all backups at the other site to not happen.  The only benefit I've seen so far in the CMMDB is it makes device management easy.  However, being unable to create backups during outages and the potential that Auto Replication Sync doesn't work in CMMDB for my site to site replication, means I need to get rid of it.

I have about a year of backups in the current IDBs at both sites.  I opened a ticket asking about how to go forward and dropping the CMMDB and got a "that isn't supported" response.  

The only positive thing I can find about getting rid of it is this for Version 5.1 and it's dated January 16, 2004: How to go back to local MMDB from a CMMDB configuration

If ARS worked I'd be able to see all the clients backups that replicated from the other site on my local StoreOnce and could perform some 1 offs to LTO6 to capture everything and if need be roll a brand new IDB at that site independent of the primary site.  However, since ARS is failing, I'm looking for ideas on moving forward and dropping the CMMDB.  

I am leaning towards getting disk images of my cell managers and trying out the 2004 instructions and see what happens.   

Any thoughts or suggestions are appreciated.

 

 

 

Parents
  • Update: Got word from HPE Support about the ARS not working with CMMDB.  Mainly - It won't work as the CMMDB makes the Media Session Manager fail during ARS because it thinks the media is a duplicate.  

    I've still got a ticket open about breaking the CMMDB up and have pointed the service reps to this thread asking for technical feedback on feasibility.  I'd like to move forward with a supported step, but if I can't then I'll backup both Cell Managers and then try and break the CMMDB.   

  • Yesterday I began trying to setup my environment as it should be.  

    My HPE Support Engineer reply regarding getting rid of the CMMDB:  

    Regarding your update:

    1. MCF export of all pools that relate to data that either host should be aware of on their corresponding StoreOnce device
    2. then breaking the CMMDB and either rebuild the devices on the other IDB or have to create a new cell manager
    3. perform an MCF import to repopulate the IDB that needs to be removed from the CMMDB

     That is how the process should be.

     So, starting from there. With Primary site (old CMMDB and Secondary Site (client of CMMDB)  I backed up both Cell Managers, performed a Copy to Catalog (MCF) of every Media in every pool of both CM's (this had to be done on the respecitve CM for where the media was created), removed Foreign Cell Managers as clients fom both cell managers, shut off DP services on both, removed the mom and cmmdb files, brought them back up.  Found that all the devices and media pools were missing in the Secondary Cell Manager.  The Primary site still had all its devices and media pools untouched.  Running omnisv -status showed no CMMDB.  

    On the Secondary, I attempted to Import Catalog from MCF files but ran into issues as it's IDB was saying they were duplicates.  I reset the IDB and then the MCF's were imported mostly fine - one out of over 1000 came up with an issue.

    I then added devices back for the StoreOnce gateways and StoreOnce VTL setup. 

    What does work:

    • Jobs on both Primary and Secondary are working.  
    • ARS now functions from Primary to Secondary and those files can be restoree from the Secondary CM
    • Secondary CM can do a restore from of an item that used the StoreOnce VTL as target 

    What does not work:

    • Secondary CM I'm getting Mount Requests when I try to restore from a job that went to StoreOnce Catalyst target.  
    • When I try to setup ARS to go from Secondary to Primary, the Object Operations Libraries shows all my StoreOnce Catalyst targets as Capable of Replication, but when I go to set the Destination, none of the devices show as being capable of replication.  I now can't replicate without ARS or with ARS to my other StoreOnce device.  

    I have an email out to my Support Engineer with what's going on.    Not quite sure why replication between two different StoreOnce Catalayst devices with the same settings for Block/Segment sizes isn't working.

    Will update when I get some progress (or lack of) on this.

  • Ok, more digging around on my part and I found when I recreated the devices I missnamed my StoreOnce Gateway for my Linux Catalyst store.  Had to correct the devices in the related Backup Specs and Object Operations.  Still no go.  Found an obsolete (backup spec no longer scheduled) Object Operation job that was pointed between a VTL store and a Catalyst Store, deleted that.  All of this took place around the DailyMaintenance/CheckTime which hadn't occured since breaking the CMMDB.  I went back ater those occured and found the Replication option was now available on all my object operation jobs.  Not quite sure if one badly setup Object Operation job along with the missnamed devices (though they were only for 1 of my 4 Catalyst stores) caused all my replication options to be unavailable or if the DailyMaintenace cleared up things.

    It's been running since yesterday afternoon and the ARS function is working now between both sites.  

    Restore Functionality of Data from the Secondary sites StoreOnce Catalyst Stores doesn't appear to be functional.   New data going into the stores is restorable.  During the CMMDB the Secondary sites backups were all going to a tape library.  I am considering the loss of the data on the Secondary Site's StoreOnce as acceptable since I still have all my weekly tape backups and am running full backups.  The site is much smaller than my Primary Site so this is an easy task.  I recycled/exported the tapes in my MSL tape library that had been a pool tied to the Secondary site and began importing them into the Primary site.  I was able to perform a restore from the tape without issue.

     

    Regarding the StoreOnce File Library having issues on the restore

    From the Help:

    Copying the Catalog Media Data to the MCF File

    Limitations
    You can select only Data Protector media.

    Due to the nature of the Data Protector file library where media cannot be exported from one and imported into another library, Copy Catalog to File and Import Catalog from File of such media should be avoided

    Should be avoided, in my case is It doesn't work...

    Now it appears that the CMMDB break is done.  I've run into a separate issue.  

    Priorly with the CMMDB the Secondary Site was able to send data from its local StoreOnce to the Tape Library.  Now that path isn't available.  ARS is functioning, but there isn't an automated Object Operation that allows for the selection of Replication Sessions.  So, I've submitted what is probably the 3rd, Support Enhancement Request for the ability to select those Replication sessions with Object Operations for a Post-Backup (Replication)/Scheduled job.  

    I can go into the Interactive Copy Operation, select Replication Sessions, and then copy them to my tape library.  But IMHO, what's the point of an Enterprise Backup Solution if you have do stuff manually. 

Reply
  • Ok, more digging around on my part and I found when I recreated the devices I missnamed my StoreOnce Gateway for my Linux Catalyst store.  Had to correct the devices in the related Backup Specs and Object Operations.  Still no go.  Found an obsolete (backup spec no longer scheduled) Object Operation job that was pointed between a VTL store and a Catalyst Store, deleted that.  All of this took place around the DailyMaintenance/CheckTime which hadn't occured since breaking the CMMDB.  I went back ater those occured and found the Replication option was now available on all my object operation jobs.  Not quite sure if one badly setup Object Operation job along with the missnamed devices (though they were only for 1 of my 4 Catalyst stores) caused all my replication options to be unavailable or if the DailyMaintenace cleared up things.

    It's been running since yesterday afternoon and the ARS function is working now between both sites.  

    Restore Functionality of Data from the Secondary sites StoreOnce Catalyst Stores doesn't appear to be functional.   New data going into the stores is restorable.  During the CMMDB the Secondary sites backups were all going to a tape library.  I am considering the loss of the data on the Secondary Site's StoreOnce as acceptable since I still have all my weekly tape backups and am running full backups.  The site is much smaller than my Primary Site so this is an easy task.  I recycled/exported the tapes in my MSL tape library that had been a pool tied to the Secondary site and began importing them into the Primary site.  I was able to perform a restore from the tape without issue.

     

    Regarding the StoreOnce File Library having issues on the restore

    From the Help:

    Copying the Catalog Media Data to the MCF File

    Limitations
    You can select only Data Protector media.

    Due to the nature of the Data Protector file library where media cannot be exported from one and imported into another library, Copy Catalog to File and Import Catalog from File of such media should be avoided

    Should be avoided, in my case is It doesn't work...

    Now it appears that the CMMDB break is done.  I've run into a separate issue.  

    Priorly with the CMMDB the Secondary Site was able to send data from its local StoreOnce to the Tape Library.  Now that path isn't available.  ARS is functioning, but there isn't an automated Object Operation that allows for the selection of Replication Sessions.  So, I've submitted what is probably the 3rd, Support Enhancement Request for the ability to select those Replication sessions with Object Operations for a Post-Backup (Replication)/Scheduled job.  

    I can go into the Interactive Copy Operation, select Replication Sessions, and then copy them to my tape library.  But IMHO, what's the point of an Enterprise Backup Solution if you have do stuff manually. 

Children
  • In hindsight, changing the backup specs to use the StoreOnce's VTL option instead of the StoreOnce Catalyst would likely have meant any data stored there would then be easily retrievable since the MCF Export/Import of the File Library (Catalyst Stores) were unrestoreable, giving mount requests and not being able to access the stores for prior data.  

    Any data done after the CMMDB break isn't having an issue with Restore.