Fixed Issues - 11.3.6 SP4 CUM1

This article details the issues that are fixed in CA Workload Automation AE Release 11.3.6 SP4 CUM1.
wcc114sp5p
This article details the issues that are fixed in CA Workload Automation AE Release 11.3.6 SP4 CUM1.
JOBS IN OFF HOLD BOXES RUN WITH WRONG NUMBER AFTER PR#4408
Problem Number: 4430
When you run a multi-level box after placing one of the inner level boxes on hold, the scheduler executes all jobs except the box that is in the ON_HOLD status. When you take the box off hold, the scheduler executed the box but the jobs in the box retained the previous run number instead of receiving a new one.
SCHEDULER CORE ON SHUTDOWN IN LIBMGRCOMM PINGING DOWN AGENTS
Problem Number: 4437
As the scheduler shuts down, it might try to initiate a ping to an agent that is in the Missing state. Through an internal race condition, the scheduler might try to access an invalid memory address resulting in a core.
SCHED PROCESSES EVENTS OUT OF ORDER WITH DUAL EVENT SERVERS
Problem Number: 4441
In dual event server mode, when you send events manually and specify the highest event priority during a period of time when the scheduler has many job events in the ujo_event table, the scheduler fetched job events out of order.
DECOUPLE SCHED_SCALE WITH NUMBER OF AGENT LISTENER THREAD
Problem Number: 4500
Starting in CA Workload Automation AE Release 11.3.6 SP3, the scheduler's agent listener thread is multi-threaded. The number of agent listener threads were controlled by the SCHED_SCALE environment variable. When SCHED_SCALE is set to 64 or to a non-default value, it impacted the horizontal latency (average processing time required for all the events).
To address this issue, a new environment variable named AGENTLISTENER_THREAD_CEILING=
n
is introduced.
For more information about the AGENTLISTENER_THREAD_CEILING environment variable, see Environment Variables -- Network Communication.
WITH NOLOCK - IGNORE WHEN JOB_NAME CONTAINS TABLENAMES
Problem Number: 4454
On Windows, if any one of the following table names: ujo_event, ujo_proc_event, ujo_job_status, ujo_job_runs, is part of the job name, the query optimization logic added the With NOLOCK clause in the wrong position in the SELECT queries that are used in the product.
SCHED CREATES DUPLICATE EVENTS WHEN AGENT COMM TIMES OUT
Problem Number: 4455
When the scheduler is running with a multi-threaded agent listener and the agent resends job status messages for which it did not receive an acknowledgement, the scheduler did not detect that the message is a duplicate and it generated a repeating job event.
JOBS RUN WITH THE OLD CONNECTION PROFILE
Problem Number: 4464
When the job definition is updated to use a new connection profile (than the original connection profile that you specified while defining the job initially), the job still continued to use the original connection profile. This problem is applicable to the CA WA Advanced Integration for Hadoop job types (HDFS, OOZIE, HIVE, SQOOP, and PIG).
IMPROVED SCHEDULER PERFORMANCE FOR FETCHING EVENTS
Problem Number: 4465
The scheduler always sleeps for a minimum of half a second for each event fetching cycle to avoid executing an intensive database stored procedure too many times in the same second. During periods of high workflow activity where many events accumulate in the database, the scheduler may spend an extended period of time on