Created from redmine issue http://projects.theforeman.org/issues/4870

Upstream bug assigned to paji

Last year we had a very hard time with foreign key changes in sam.  Is this really slotted for GA? or for future?

Suresh,
I did think about the pain we had last year having written a few of those nightmarish upgrade migration scripts and many of foreign key complaints were on upgrades. And most of those FK issues were because the code would remove one association and forget to remove the child associations. For example if I removed an environment and forgot to remove every place environments were FK'ed we would get a breakage. It was probably harder and more annoying for SAM because many of those tables in the DB were Katello specific and never used by SAM. But atleast we know that there was no dirty data in the DB related to this.

Now for the other side of this issue, if this bug had been merged earlier we would have not enncountered https://bugzilla.redhat.com/show_bug.cgi?id=1113684 where the satellite had repositories that basically pointed to non existent products (due to botched bulk product delete code). The FK'ed code would ve failed earlier and said "bulk delete of product failed"  and left data at a consistent state and not affected a totally different page.

Also we need to consider that this update adds only like 15 foreign keys. There are already like over 100 foreign keys being used in the present product. 

My vote is in favour of including it in the GA,

(In reply to Partha Aji from comment #5)
> Suresh,
> I did think about the pain we had last year having written a few of those
> nightmarish upgrade migration scripts and many of foreign key complaints
> were on upgrades. And most of those FK issues were because the code would
> remove one association and forget to remove the child associations. For
> example if I removed an environment and forgot to remove every place
> environments were FK'ed we would get a breakage. It was probably harder and
> more annoying for SAM because many of those tables in the DB were Katello
> specific and never used by SAM. But atleast we know that there was no dirty
> data in the DB related to this.
> 
> Now for the other side of this issue, if this bug had been merged earlier we
> would have not enncountered
> https://bugzilla.redhat.com/show_bug.cgi?id=1113684 where the satellite had
> repositories that basically pointed to non existent products (due to botched
> bulk product delete code). The FK'ed code would ve failed earlier and said
> "bulk delete of product failed"  and left data at a consistent state and not
> affected a totally different page.
> 
> Also we need to consider that this update adds only like 15 foreign keys.
> There are already like over 100 foreign keys being used in the present
> product. 
> 
> My vote is in favour of including it in the GA,

@Partha: Ok, it makes sense to include this in GA.  Thank you for the explanation.

Moving to POST since upstream bug http://projects.theforeman.org/issues/4870 has been closed

This was delivered with Satellite 6.0 which was released on 10 September 2014.