Archieving and Deleting Process.

Certified Associate Developer

Which one is preffered Archieving or Deleting a process model in Data management .

Any use case for the both ? 

  Discussion posts and replies are publicly visible

  • 0
    Certified Lead Developer
    in reply to Chris

    That race condition is a fun one.  I bet that was fun to figure out (assuming you know about it the hard way).

  • Those are always "fun"!  If I recall, I learned about it the easy way a while back (from Eduardo during his tenure with Appian).  We hadn't run many subs as asynchronous in the legacy Portal/Apps environment, but we do so more now with Tempo and shorter lived processes - and if they are set to delete, do so after 1 day.  No one needs more issues Slight smile

  • 0
    Certified Lead Developer
    in reply to Chris
    We've archived millions of processes in the past (early theory), while never having to unarchive 1.  They just eat server resources,

    I'm annoyed that it's a strictly one-or-the-other configuration decision still.  We should be able to archive after 3 days, and delete after 30, for example.  It would help if the "new" feature announced last year where we will allegedly be able to view archived processes right in the process monitor, would ever actually roll out.  While expressing my annoyance about this at AW, someone hinted that archival in general might be "going away" soon, though I have no idea what this would really look like at the end of the day.

  • Agree, would be nice if we could control a combo of archive/deletion.

    Maybe (hoping out loud), archiving would some day be replaced with something such as full, automatic (and adjustable) logging of Process History to an area in the Primary Data Source with built-in reporting capability..  Something you could manage with a much lower footprint of storage use, with full auditing capability..  Stick with me...then have the ability to generate a Monitor instance from the data source, without needing the "process instance" unarchived into memory.....

    Are we getting warm yet?!

  • 0
    Certified Lead Developer
    in reply to Mike Schmitt

    Or you could always search Splunk for a given error message to see when it first started spiking, then find the coinciding release date one or two days before that to determine when Dev injected that bug.

    Wound up being how I usually solved the mystery.