E1: RTE: RTE Events Out Of Sequence On WAS Cluster
(Doc ID 2678818.1)
Last updated on DECEMBER 02, 2021
Applies to:
JD Edwards EnterpriseOne Tools - Version 9.1 and laterInformation in this document applies to any platform.
Symptoms
When using WebSphere Cluster for RTE server, when one of the nodes is stopped for maintenance or patching, for example, RTEs are no longer picked up from F90710 table.
When checking the F90708, the sequence number is greater than in F90710 table.
We have introduced the new table F90708 while supporting RTE Clustering with TR 9.1. This is applicable to standalone RTE Server as well now.
Typically in a clustered RTE Server environment, say there are Servers S1 and S2 part of the cluster C1. All these clustered servers(S1 and S2) are capable of polling the table F90710 for any new events coming in. There could be possibilities like S1 and S2 tried to poll the table same time or while querying the table, got some duplicate records. To avoid this F90708 is introduced. It is a single record single column table to hold the last record picked up by the RTE server for processing. For a first time initialization the value is set to 0.
So now S1 attempted to poll F90710. First it checks for the last sequence number of message in F90708, found it to be 0 so asks for record 1-1000. It got and update F90708 record with value 1000. During operating on F90708, lock is obtained so that no other servers can operate on this table.
Now S2 comes for polling. It checks F90708 and finds that the record is 1000 and asks for records 1001-2000 and updates F90708 with value 2000.
Again S1 comes and It checks F90708 and finds that the record is 2000 and asks for records 2001-3000 and updates F90708 with value 3000.
Lock is obtained by any server querying F90708 and updating table with last sequence number. So there is no question of duplication.
Typically in a clustered RTE Server environment, say there are Servers S1 and S2 part of the cluster C1. All these clustered servers(S1 and S2) are capable of polling the table F90710 for any new events coming in. There could be possibilities like S1 and S2 tried to poll the table same time or while querying the table, got some duplicate records. To avoid this F90708 is introduced. It is a single record single column table to hold the last record picked up by the RTE server for processing. For a first time initialization the value is set to 0.
So now S1 attempted to poll F90710. First it checks for the last sequence number of message in F90708, found it to be 0 so asks for record 1-1000. It got and update F90708 record with value 1000. During operating on F90708, lock is obtained so that no other servers can operate on this table.
Now S2 comes for polling. It checks F90708 and finds that the record is 1000 and asks for records 1001-2000 and updates F90708 with value 2000.
Again S1 comes and It checks F90708 and finds that the record is 2000 and asks for records 2001-3000 and updates F90708 with value 3000.
Lock is obtained by any server querying F90708 and updating table with last sequence number. So there is no question of duplication.
Changes
Cause
To view full details, sign in with your My Oracle Support account. |
|
Don't have a My Oracle Support account? Click to get started! |
In this Document
| Symptoms |
| Changes |
| Cause |
| Solution |