• State Not Answered
  • Replies 2 replies
  • Subscribers 2 subscribers
  • Views 67 views
  • Users 0 members are here
Options
Related

How much PI DA Collective functionality is server-side?

Community User

From what I understand, PI Collective Manager runs on client machines, and tells the client which secondary archive to request data from if the primary doesn't respond. For PI Interfaces, the PI Buffering Manager tells the interface node to buffer data to all archives in the collective. Since the data archives in a collective also don't sync, is there any configuration needed on the archives themselves to tell them that they are part of a collective with other archives?

  • 0

    I think you have a misunderstanding. PI Collective Manager is an admin tool, and it can be run from an admin's PC or from a server. Just because you run PI Collective Manager from your own PC does not mean your PC retains any special connection or configuration once the app is closed.

    Create and Manage Data Archive Collectives

     

    So, no, PI Collective Manager app does not tell the client running it which secondary archive to request data from if the primary does not respond. Again, there is no client-specific configuration at play here. PCM lets you create a collective, and the first Data Archive added becomes the primary. As you add one or more additional Data Archives, they become secondaries.

    What you perform through PCM is pushed into the member servers configuration. There are PI SDK and AF SDK commands to let any Data Archive know whether it belongs to a collective, and if so, what role it plays.

    Also, its more of a connection manager's responsibility on a client to determine connection behavior. For PI System Explorer, this is done by Tools -> Options -> (tab) Server Options. This is where you would Prefer Primary or Require Primary.

     

    And there are some things that you do external to PCM that is collective aware. Imagine running PI SMT on a client PC that does not have PI Collective Manager on it. Let's say you connect to the primary member of a collective to become your active Data Archive. And let's say you add new PI Points to that member. That exact Point definition, including the Point ID, is replicated to all members of the collective. So while archive data is not replicated, point definitions are. To emphasize again: in this example, PI Collective Manager is not a factor, and the primary member server is away of its sibling members.

    BTW ... each member server's archive will create a "Pt Created" value for each new tag, but the timestamps may not be exactly equal. If the tag replication takes 4-seconds, then the "Pt Created" will be 4-seconds off.

     

     

  • 0 in reply to Rick_Davin_3.0

    Thank you. Is the collective configuration stored separately on each archive server's PI Module Database, or somewhere else? Also are there occasionally issues where this configuration is out of sync between the respective collective members?