DX NetOps

  • 1.  How to sync the modeling catalog

    Posted Aug 14, 2014 04:06 AM

    I'm trying to add extra Object ID's to spectrum, but i see that there is a conflict.

    The attribute id of an existing object id differ on the landscapes.

     

    Spectrum say to sync the modeling catalog with the master, but how can i sync the catalog?

    I can't find it in the documentation

     

    Thanks

    Regards



  • 2.  Re: How to sync the modeling catalog

    Posted Aug 14, 2014 05:51 AM

    Hello Frank,

     

    Not sure it will answer the question.

    No matter how you want to add OID with spectrum ( using Model Type Editor or Mib Tools), this action should be done on the MLS server (I guess it is automatic with Mib tools with map attributes,but documentation to be read here for confirmation)

     

    To sync you catalog between SpectroServer, you have to save the catalog and load it :

    either you  can use you last backup and load only the catalog ( #SS> ../SS-Tools/SSdbload -c saved_zungiped_file )

    Or you can save only the catalog ( #SS> ../SS-Tools/SSdbsave -c filename) from where you have the correct one and apply it on the others by loading the catalog.

     

    Cheers,

    Virginie



  • 3.  Re: How to sync the modeling catalog
    Best Answer

    Posted Aug 14, 2014 10:24 AM

    Virginie is right on the money with how to sync the catalog.  However, there are a couple of steps left out.  You need to backup the models on the SpectroSERVER before you load the catalog, then overlay those saved models after you load the catalog.

     

    - make the catalog changes on the master system (we use our MLS)

    - backup catalog on the master system (

    --         stop SpectroSERVER.

    --         cd ${SPECROOT}/SS

    --         ../SS-Tools/SSdbsave -c {filename}

    --          then start up the MLS SpectroSERVER again)

    - copy this catalog to all your SpectroSERVERs

    - on each SpectroSERVER load this MLS catalog

    --       Stop SpectroSERVER

    --       cd ${SPECROOT}/SS

    --       ../SS-Tools/SSdbsave -m {models save filename}     # save the existing models

    --       ../SS-Tools/SSdbload -il {catalog backup}                  # load master catalog

    --       ../SS-Tools/SSdbload -l {models save filename}        # load the existing models save

    --      Start the SpectroSERVER.

    --     Run online backup to sync this database with the secondary

     

    As you see this is disruptive.  For us it is also very time consuming to run around 54 spectroSERVERs to keep them all in sync.  Also, what we have noticed, is that you can use MibTools to update SpectroSERVERs with new attributes, but it has several issues.

    - There is no way to document what changes you have made to the catalogs.  You kinda just push the button and it magically pushes out to the SpectroSERVERS, but there is no way to document that for CM or documentation purposes.

    -  If a SpectroSERVER in your environment gets a catalog that is different and doesn't have the attribute that was pushed out via mibtools, then you are hosed.  MibTools won't push the change out to the systems that are different, or even let you fix the difference.  It either tells you that the attribute doesn't exist, or is out of sync (I forget which).

     

    So, what we now do is we go to our lab box, update the lab box with the current "master catalog".

    Go into MTE with the -log flag.  The -log flag will save your changes via an xml file.

    We then use MTE to update the catalog on the lab box.

    Then we take that XML that was generated by MTE back to the master server.

    Bring down the master server and update the catalog with dbtool and that xml.

    Then we run around all our servers, stopping and updating the catalogs.

    This way we also have a document, the xml file, that documents what has changed in our environment.

     

    We would love to see this functionality in Mibtools.  It would make our life much easier and would help promote loading in the pollable mib objects into Spectrum and using Spectrum for more than just a trap processor.  After all if MibTools can update all the Spectrum servers by adding attributes already, why can't we get it to sync the landscapes and document the changes?

     

    I had put in an enhancement request for this a long time ago, which has probably been lost.  There is an "idea" out there but I can't find it for part of this functionality.  So I guess I'll submit an idea for this entire scenario later today.