Forum Discussion

Andrew_Spear_68's avatar
Andrew_Spear_68
Icon for Nimbostratus rankNimbostratus
Apr 05, 2005

SharePoint monitoring

We are using BigIP to monitor a Microsoft SharePoint cluster. The monitor simply hits a particular SharePoint site and checks for a snippet of HTML in the return. Here's my problem - we have many individual sites in SharePoint and any of which can be deleted by the user (or automatically by SharePoint if there is no activity). Recently the site that BigIP was monitoring was deleted, and thus all traffic to our cluster was rejected - even though all the other sites were up and available. So, I am wondering if there is a way for BigIP to monitor multiple sites, and only reject traffic to a server if ALL of the site monitors fail for that server. Please excuse my ignorance, I am not the BigIP guy around here and know VERY little about it.

 

 

Thanks-

 

Andrew

3 Replies

  • bknotwell_12713's avatar
    bknotwell_12713
    Historic F5 Account
    This isn't really a rules or iControl question but typical usage would be to do something like the following (nothing sharepoint specific below):

     

     

    o create a monitoring only meta-site

     

    o monitor it

     

     

    While monitor instances can be created with monitors logically and'ed together, there's no capability to logically or them together.

     

     

    Since this isn't really on-topic for devcentral, I'd suggest contacting our support organization on this issue.
  • bknotwell_12713's avatar
    bknotwell_12713
    Historic F5 Account
    H'mmm, come to think of it, this could be iControl-ish after all.

     

     

    You could write an external monitor that did something like the following (assumption: sites are identically represented across the breadth of the cluster):

     

     

    o configure a class containing the list of sites to monitor for a specific pool

     

    o create an external monitor/pool with the class name as an argument

     

    o write an ECV

     

     

    The ECV would look something like the following:

     

    o query the class via iControl

     

    o check each of the specified sites sequentially with the monitor

     

    printing something and exiting at your first success

     

     

    Needless to say, you wouldn't want to do this too often and you need to set your monitor interval and timeouts high enough to allow N queries. . .where N == number of entries in the class.

     

     

    This is all dreamt up in the last 5-10 minutes so it's likely this is a horrible idea.

     

  • Jay_Bruce_41121's avatar
    Jay_Bruce_41121
    Historic F5 Account
    Just for completeness sake, note that we can monitor applications

     

    with BIG-IP v9.x. In this version you would create a pool for

     

    each application. Each pool would be monitored by a different monitor

     

    which monitors the specific application. An iRule is used to vector

     

    off to the different pools based on the header. If a monitor reports

     

    a pool down, the corresponding application will not get traffic, but

     

    the other pools will remain up, receiving traffic.