none
Upgrading to FIM 2001 R2 RRS feed

  • Question

  • We're upgrading our PREPROD environment from ILM to FIM 2010 R2.  We've gone via FIM 2010, having started the direct upgrade and then read Tim Macaulay's post on this thread (we didn't spot this in the documentation).

    It's  taking a long time.  We started yesterday at 11.47.  By 12.15, we were seeing:

    Action 12:15:50: ConfigDB. Converting representation of references

    and we're still seeing them.  At the start, we were seeing them at 1-3 second intervals.  Now, after over 600, we're up to about 3 1/2 minutes.  Does this just refer to the main DN of directory objects or is it every DN (seeAlso, manager, etc.)?  Does having Reference DN attributes like manager impact this worse than just having a main DN?

    The machine is a VM.  It has 4 vCPUs and 32 GB of RAM.  It's currently using about 26% of the CPU, spread across all four cores; and 13 GB of the memory (the data file MicrosoftIdentityIntegrationServer.mdf is about 8.2 GB).

    First, if we continue to see them, do we just assume all is well for now?

    Second, is there any way to interpret progress from these?

    Third, is there anything we can do to speed things up?

    We have three directories which are entirely generated from other directories and databases (so we could remove them before the upgrade and put them back afterwards).  Each probably has 80,000 entries most of which will have the manager attribute populated.  Would it be quicker to export the MA config, delete the MAs, do the upgrade, import the MAs and Full Import, Full Sync them?  Is this the sort of thing anyone else has tried?

    Sunday, September 30, 2012 7:00 AM

Answers

  • R2 handles references differently in the database and it can take an extended time to upgrade. Just leave it running. You can monitor what it's doing with the SQL Profiler but I don't know of any way to tell how far it is through the job.

    On the idea about deleting the data - could work. I wouldn't delete the whole MA, just clear the connector spaces.


    http://www.wapshere.com/missmiis

    • Proposed as answer by Carol Wapshere Monday, October 15, 2012 1:03 AM
    • Marked as answer by SSG31415926 Monday, October 15, 2012 9:35 AM
    Monday, October 1, 2012 12:03 AM
  • Thanks.  It finished as predicted and took 25 hours.

    And it ran a lot faster in PROD - about 9 hours.  Apparently, there are two levels of 'GOLD' discs in our VM infrastructure! 

    For info, it used around 28 GB of RAM (on a 32GB machine) and 90%+ of 3 vCPUs (we dropped it by one since they weren't being used on PREPROD).  Not sure why it was so different but no time to investigate and no plans to repeat. :-)

    • Marked as answer by SSG31415926 Monday, October 15, 2012 9:35 AM
    Monday, October 15, 2012 9:35 AM

All replies

  • R2 handles references differently in the database and it can take an extended time to upgrade. Just leave it running. You can monitor what it's doing with the SQL Profiler but I don't know of any way to tell how far it is through the job.

    On the idea about deleting the data - could work. I wouldn't delete the whole MA, just clear the connector spaces.


    http://www.wapshere.com/missmiis

    • Proposed as answer by Carol Wapshere Monday, October 15, 2012 1:03 AM
    • Marked as answer by SSG31415926 Monday, October 15, 2012 9:35 AM
    Monday, October 1, 2012 12:03 AM
  • Thanks.  It finished as predicted and took 25 hours.

    And it ran a lot faster in PROD - about 9 hours.  Apparently, there are two levels of 'GOLD' discs in our VM infrastructure! 

    For info, it used around 28 GB of RAM (on a 32GB machine) and 90%+ of 3 vCPUs (we dropped it by one since they weren't being used on PREPROD).  Not sure why it was so different but no time to investigate and no plans to repeat. :-)

    • Marked as answer by SSG31415926 Monday, October 15, 2012 9:35 AM
    Monday, October 15, 2012 9:35 AM