Post reply

Check for availability before removing snapshots

  • November 9, 2009 at 9:25 am

    #213756

    For a client of ours, we have configured a Server Of Last Resort using a snapshot of their mirrored production database. The creation of this snapshot uses a simple two step process which is executed every 30 mins:

    - Delete present snapshots

    - Create new snapshots

    This weekend we encountered a situation whereby the the principle database was unavailable for more than 30 minutes and the site was running from the snapshot until the script to synchronise the snapshots re, leaving us in a situation with no new snapshots and no source from where to create them.

    How can I add an additional step to verify if the principle server is available before proceeding to erase the present snapshots?

    Regards,

    AtlasAdmin

  • November 9, 2009 at 12:28 pm

    #1076860

    I'm a little confused by your post - are you talking about database snapshots rather than snapshot replication or some other copy? If so, how can you run off the snapshot if the principal is down, and why wouldn't you run off the mirror?

  • November 10, 2009 at 3:32 am

    #1077143

    Thanks for the reply Matt,

    the Technical Architect for this solution is no longer at the company, but my understanding of the DR process is the following:

    In case of DR situation occurs the failover process is the following:

    London --> Frankfurt SOLR ---> Frankfurt ---> Frankfurt SOLR.

    An SQL 2005 cluster in London acts as the principle site and an SQL 2005 cluster in Frankfurt acts as the mirrored site. There is also a process to create snapshots of the principle databases and these snapshots are run from the Frankfurt SQL nodes. It is here where we require an additional LOGIC step to verify that the London site is back online before removing

    Regards,

    Paul Lambert

  • November 10, 2009 at 8:20 am

    #1077297

    I think you may be referring to replication snapshots or a database backup/restore solution. I would take one of these approaches:

    a) try and run a query on the principal via linked server, continue with the process if you get a result set

    b) do the create first, continue with the deletes if it is successful

    B is probably the simpler solution.

    Matt.

  • November 10, 2009 at 8:26 am

    #1077305

    Could you please specify what kind of snapshots are you talking about? Thanks in advance



    My blog

  • November 10, 2009 at 9:09 am

    #1077336

    Sorry, I should have specified earlier, these are database backup/restore snapshots. My understand is they are readonly DBs while the main site is down.

    Regards,

    AtlasAdmin

  • November 10, 2009 at 9:16 am

    #1077343

    Are you using snapshot replication?



    My blog

Viewing 7 posts - 1 through 6 (of 6 total)

You must be logged in to reply to this topic. Login to reply