Input Connector's getting stuck after a database lock

243
3
Jump to solution
02-01-2019 10:07 AM
NathanKoski
New Contributor III

We have a table on our ORACLE database, named Order_Updates, which Geoevent is querying on a 1 minute interval using the "Poll an ArcGIS Server for Features." The table is inside of a feature service which is hosted on the GeoEvent's ArcGIS Server, so on the localhost, and it deletes the rows after it finishes its query.

The problem is that every night, at 8pm, our database does a Reconcile batch job to move all the updates from that day from Common to Default, which puts a lock on all the tables and features in that database. And every night after 8pm Geoevent is unable to query the table giving this error:

com.esri.ges.transport.featureService.FeatureServiceInboundTransport

Error while retrieving features from service Cust_Order_Data. Error: Error accessing data store at https://localhost:6443/arcgis/. : 400 : Unable to complete operation..

This continues to appear until exactly midnight, which after the server seems to have no issues performing the input connector's query. I can help aid this process by manually restarting the input connector after which the connections are fine. I am thinking the reason why it works after midnight is that midnight is when Geoevent is set to do its daily backup, and something in that is effectively restarting the connector.

I understand that during the 5-10 minutes that the batch job is running, and the lock is placed on the table, that GeoEvent and the feature service will be unable to access it, but why is it that after the batch completes and the lock is removed that it does not automatically restore its connections? Is there some type of behavior with Feature Services that might cause this, or is it a GeoEvent problem?

Any input would be amazing because I cannot find any info online regarding this behavior. Right now, if we disable the batch job, the Connectors remain active, but that is not a long term solution.

Thanks,

Nate

Reply
0 Kudos
1 Solution

Accepted Solutions
Stefan_Jung
Esri Contributor

Hi Nate,

did you check the ArcGIS Server Logs as well? My tip would be that it has something to do with the recyling configuration of the feature service.

The default recycling time is midnight. So a solution could be to set the recycling time for this feature service to 8:15pm and test if this works.

Best,

Stefan

View solution in original post

3 Replies
Stefan_Jung
Esri Contributor

Hi Nate,

did you check the ArcGIS Server Logs as well? My tip would be that it has something to do with the recyling configuration of the feature service.

The default recycling time is midnight. So a solution could be to set the recycling time for this feature service to 8:15pm and test if this works.

Best,

Stefan

View solution in original post

NathanKoski
New Contributor III

Hi Stefan,

Thanks for the reminder to check the ArcGIS Server Log

Error: Failure to access the DBMS server [ORA-01012: not logged on Process ID:*** Session ID: *** Serial number: ***] [GISADM.GEO_OMS_ORDER_UPDATES].

Is the error I receive. GEO_OMS_ORDER_UPDATES is the database table. This appears to only affect the non-spatial table, as I never see an error from the Point feature that is attached to this feature service, which is in the same database.

After 12:00am this stops. I set the recycle time to be 8:15pm, and so will test this tonight.

Thank you for your advice, I will post an update tonight/tomorrow.

Nate 

Reply
0 Kudos
NathanKoski
New Contributor III

This solution worked perfectly. I set the recycle to happen at 8:30pm to ensure that the reconcile job had fully completed, and the gap trap processor, which is used for data check-ins, in the geoevent service shows data stopping at 8pm as excepted, but returning through the input at 8:30pm.

Thanks for your advice!

Nate

Reply
0 Kudos