AnsweredAssumed Answered

GeoEvent Error - update a feature - Operation rolled back + ZK Persistence error

Question asked by ascullyCOT on Apr 30, 2018
Latest reply on Nov 8, 2019 by nelsm

We have a fairly straight-forward GeoEvent service that recently stopped working.  Essentially the service takes an input table that has geometry stored as a string, and writes to a polygon feature class (using a field calculator on the geometry string field).  

These polygons are then used as geofences in a separate GeoEvent service.


However something has caused this to stop working - I cannot get the service to write to the feature class any more.  No configuration changes were made, nor schema changes to any table/FC involved. 

I've tried different versions of the output feature service (new services w/ new, empty feature class) - multiple restarts of the service and full server reboots (it's dev environment) - no change.


 - We are running Enterprise 10.5, and I have installed the patch for GEP (GeoEvent-10.5-Patch1) to not avail.

 - The service WILL write to json output successfully, so it must be related to the output (Update a Feature).

 - The Monitor shows that the correct # of records are sent to the output, but nothing gets written to the feature service.

 - There are 2 errors I see in the logs: 

      1) Error Code 400 - Error while writing to feature service xxxx. Error: Unable to complete operation > Operation rolled back...

      2) com.esri.ges.fabric.internal.ZKPersistenceUtility > KeeperErrorCode = ConnectionLoss for /geoevent/config/clusters/default/arcgisServerConnections....(much more here that I can't intelligently decipher).  I've checked to Default connection in the data store, and all appears ok there.  


I'm stumped after 2 days of trying to track this down.  Likely will open a ticket, but figured I'd ask wiser more experienced heads than myself first.