Problem description

SAS^® Real-Time Decision Manager 6.1 (SAS Decision Services 5.6) introduces new counts monitoring functionality for activated campaigns and diagrams. The new SAS Decision Services Monitor Service writes the counts data to a database. This data is used to return counts for display in the campaign and Diagram nodes in SAS Customer Intelligence Studio. In the initial release of the monitoring service, there is no automatic archiving or deletion of records. Because records are not periodically cleared, the monitoring service might fill its temporary space or database. When the temporary space or database is filled, SAS Customer Intelligence Studio is unable to open activated campaigns and diagrams. Campaigns and diagrams that have not yet been activated are unaffected by this issue.

Error messages

SAS Customer Intelligence Studio returns the generic error message:

The full error is written to the SAS Customer Intelligence core log:

2013-10-24 08:40:03,523 ERROR
Session[41a93e32544d55ca:55e89289:141e97a95c8:-3cb3] User[psimms1]
SID[41a93e32544d55ca:1ac68eaf:141e6b09481:-6152]
com.sas.analytics.crm.cm.ejb.CampaignManagerBean Error loading campaign.
java.lang.RuntimeException: PreparedStatementCallback; uncategorized
SQLException for SQL [SELECT FLOW_NM, AVG(RESPNSE_TIME_MILLISECONDS) FROM
DECISIONSERVICES.DCSV_MONITOR_FLOW_RESPNSE_TIME GROUP BY FLOW_NM]; SQL state
[HY000]; error code [0]; sort error - No free space found in temporary
directories - operating system directive write failed - No such file or
directory; nested exception is java.sql.SQLException: sort error - No free space
found in temporary directories - operating system directive write failed - No
such file or directory
        at
com.sas.dcsv.monitor.webservice.ws.proxy.MonitorSoapProxy.getSystemStatistics(Mo
nitorSoapProxy.java:271)
        at
com.sas.analytics.crm.rdm.RDMMonitorWS.getSystemStatistics(RDMMonitorWS.java:219
)
        at
com.sas.analytics.crm.rdm.RDMMonitorWS.getNodeCounts(RDMMonitorWS.java:236)
        at
com.sas.analytics.crm.rdm.RDMMonitorWS.getCountMap(RDMMonitorWS.java:75)
        at
com.sas.analytics.crm.flow.inbound.IBFlow.refreshLiveCountsWithClear(IBFlow.java
:2712)	
...

Workaround

To work around this issue, periodically delete older monitor records. See the Full Code tab for an example of SQL code that can be used to do this in an Oracle database.

Click the Hot Fix tab in this note to access the hot fix for this issue.

After the Hot Fix

Counts are computed less often

Before the hot fix, SAS Customer Intelligence Studio gathers counts for each active campaign when it is opened. If there is a problem with the monitoring service that provides the counts, the campaigns cannot open. After the hot fix is applied, SAS Customer Intelligence computes and displays counts only when you click Refresh Counts.

New configuration options are available

• Query interval - The interval in seconds when for polling the engines for statistics. The default value for this parameter is 5 seconds.
  JVM argument: -dcsv.monitor.query.statistics.interval.seconds
• Engine time out - The maximum time to wait before the call to the engine to collect statistics is timed out.
  JVM argument: -dcsv.monitor.engine.query.timeout.seconds
• Engine query thread pool parameters - The parameters for setting up the thread pool, for the threads that collect statistics from the engines in the cluster.
  JVM argument: -dcsv.monitor.execution.thread.pool.*
• Clear old records - If true, this will periodically clear old records. The default value is True.
  JVM argument: -dsasds.monitor.clearOldRecords
• Length of time to keep records - If sasds.monitor.clearOldRecords is True, then this argument triggers deletion of records older than the specified number of days from the current date. The default value is 30.
  JVM argument: -dsasds.monitor.keepRecordsForDays

Operating System and Release Information

select count(*) from ora_dsdev1.dcsv_monitor_activity_metadata where extract(day from (current_timestamp - sample_dttm)) > 30;
select count(*) from ora_dsdev1.dcsv_monitor_event_counts where extract(day from (current_timestamp - end_dttm)) > 30;
select count(*) from ora_dsdev1.dcsv_monitor_flow_respnse_time where extract(day from (current_timestamp - end_dttm)) > 30;
select count(*) from ora_dsdev1.dcsv_monitor_node_counts where extract(day from (current_timestamp - end_dttm)) > 30;

delete from ora_dsdev1.dcsv_monitor_activity_metadata where extract(day from (current_timestamp - sample_dttm)) > 30;
delete from ora_dsdev1.dcsv_monitor_event_counts where extract(day from (current_timestamp - end_dttm)) > 30;
delete from ora_dsdev1.dcsv_monitor_flow_respnse_time where extract(day from (current_timestamp - end_dttm)) > 30;
delete from ora_dsdev1.dcsv_monitor_node_counts where extract(day from (current_timestamp - end_dttm)) > 30;

select count(*) from ora_dsdev1.dcsv_monitor_activity_metadata where extract(day from (current_timestamp - sample_dttm)) > 30;
select count(*) from ora_dsdev1.dcsv_monitor_event_counts where extract(day from (current_timestamp - end_dttm)) > 30;
select count(*) from ora_dsdev1.dcsv_monitor_flow_respnse_time where extract(day from (current_timestamp - end_dttm)) > 30;
select count(*) from ora_dsdev1.dcsv_monitor_node_counts where extract(day from (current_timestamp - end_dttm)) > 30;

A fix for this issue for SAS Marketing Automation 6.1_M1 is available at:

A fix for this issue for SAS Decision Services 5.6 is available at: