Schneider Electric GIS Tech Support

Morning all,

From time-to-time we will have a design that reports conflicts but no conflictes are detected when reconcile is run manually. Has anyone else run into this situation?

Background:
We have a program that runs nightly which attempts to reconcile all versions in our Oracle SDE database. The program is coded to detect that a conflict exists and write a line to our Reconcile Report and then move to the next version.

Each morning I open those designs that report conflict and manually reconcile them. On occasion when I run reconcile myself no conflicts are found. I save and close the design. Many times this same design will appear again in the report.

We have tried to determine why this sort of behavior would be taking place and have not been able to discover a direct cause.

I decided to check with the rest of you to see if this issue only exists at our location or if others are experiencing it too.

I look forward to hearing from you either way.

Thanks,
george

I've run into similar situations. Maybe when you do the manual reconcile you are using different defaults (column vs. row/pre edit vs. conflict) than your batch program. So then your manual reconcile could be filtering out conflicts that your batch program doesn't.

One possibility:
If you have custom conflict filters developed, make sure they are configured on the machine that is doing the batch reconcile. I also discovered with ours that when ESRI and MM changed (around ArcGIS 9.2?) to requiring code to manually initiate licenses that my custom conflict filters were failing. I needed to make sure the reconcile app was checking out an ArcFM license.

I've seen this behavior also since upgrading to ArcGIS 9.3. I'm not sure that is the cause but it's one clue. Since the upgrade to 9.3 I have a version that is the child of another version, not DEFAULT. It has conflicts when batch reconciled, but reconciles without conflicts when run in ArcMap. The report from my conflict filter shows that the shape length for two features in the version have conflicts and it fixes them by taking the conflict version, but when I check afterwards, the shape length is still slightly different from it's parent which it should have just copied. I know with network features, changing the shape of a connected feature will cause adjacent features to adjust, but then I should be seeing conflicts with those features as well but I'm not. I have not had time to fully analyze this - this is just what I've noticed so far.

Kevin, that sounds like a pretty serious problem. We're not on 9.3 yet but when I start testing it I'll see if we get the same behavior.

Please post a follow up if you discover anything more about this. Thanks.