Orphan Assignments in the Database and Cube

[Linda]

I have a mass of orphaned assignments in the Project Server 2003 database and
in the cube. Managed time periods had been turned off and projects were
deleted and then re-imported to the server. Then actuals were "overwritten"
to correct assignment data and managers ignored the warning messages that
there was remaining work on resources that were no longer active in the
server. One project plan has duplicate assignments with duplicated actuals
so I know a major clean-out is necessary.

I want to save all the project files with their actuals locally. Then I
want to delete all the project plans from the server. But, I'm not sure what
order I should follow for the next steps.

How do I make sure the database and cubes are cleared of all tasks and
asisgnments before re-importing the project plans back to the server? I can
force a rebuild of the binary record in the database but is that all I need
to do?

And--must I reactivate the inactive resources that are in these project
plans so that the data can be reposted to the database?

I'm taking control of this server and its projects until project managers
are properly trained regarding assignments.

 

[Joel P.]

Ouch, sounds like you are taking drastic measures in deleting and
re-importing plans with Actuals. That aside, these are euphemistically known
as "orphaned tasks", and can be deleted safely by running a database script.
The good folks at ProjectServerExperts (who are indeed the true experts)
published a brief on this. Please see
http://www.projectserverexperts.com/Shared Documents/deletecancelledtasks.htm

Warm regards,

Joel Plaut

 

[Joel P.]

Ouch, sounds like you are taking drastic measures in deleting and
re-importing plans with Actuals. That aside, these are euphemistically known
as "orphaned tasks", and can be deleted safely by running a database script.
The good folks at ProjectServerExperts (who are indeed the true experts)
published a brief on this. Please see:
http://www.projectserverexperts.com/Shared Documents/deletecancelledtasks.htm

Warm regards,

Joel Plaut

 

[Linda]

Yes, it sounds drastic but the only data I can depend on is in the project
plans themselves. I'm finding no duplicate projects just duplicate task
actuals for 2 different projects. I originally thought all 15 projects
contained duplicate actuals. So I am not going to remove all projects from
the server as originally planned.

I delete cancelled tasks nightly using a stored proc. My problem is that
the duplicate tasks show up in the database and subsequently the portfolio
analyzer cube not on the user's timesheets.

My DBA is going to help me clean up the binary files of the two offending
projects I saved offline. I'm hoping that this will correct the duplicate
records in the database so I can rerun the cube. But I'm not sure if it will
work.

I don't know if republishing assignments when I bring the project plan back
online will introduce more duplicates because I don't know if updating the
binary file will eliminate the duplicate task actuals.

This problem was introduced by project managers trying to move ticket items
(and their actuals) between software release project plans. Management is
changing the contents of releases very frequently lately due to shifting
priorities. This makes it difficult to move tickets to a different release
correctly when actuals are already reported on a ticket. By the way we do
not have a need to lock down timesheets.

If we don't move the tickets, progress on our releases (we work 5 or 6 at
the same time) doesn't appear accurately in our portfolio views. If we move
tickets we cause more problems by not moving them correctly.

Since I can't fix the shifting priorities I'll probably search for a
third-party tool to report at the task level so I can create a custom
"release number" field for each ticket. Then I can get my progress reports
through that field.