Installation Instructions for Hot Fix A94001

Solaris for x64


Hot fix A94001 addresses the issue(s) in SAS Web Analytics 5.3_M1 on Solaris for x64 as documented in the Issue(s) Addressed section of the hot fix download page:

http://ftp.sas.com/techsup/download/hotfix/HF2/A94.html#A94001


IMPORTANT NOTES

  1. You must have SAS Web Analytics 5.3_M1 installed on your system before applying this hot fix. Refer to SN-35968 for instructions on how to determine which product releases you have installed.

  2. Files delivered in this hot fix will be backed up during the installation process. However, it is good general practice to back up your system before applying updates to software.

  3. You must have Administrator Privileges on your CLIENT or SERVER machine.

  4. All currently active SAS sessions, daemons, spawners and servers must be terminated before applying this hot fix.

  5. This hot fix should be installed using the same userid who performed the initial software installation.


INSTALLATION

1. The installer downloaded is A94001sx.bin. Verify that the installation binary has execute permission. If it does not, use the chmod command to make it executable.
$> chmod 755 A94001sx.bin
2. Set your $DISPLAY environment variable
export DISPLAY=<your_node_name>:0
3. Execute A94001sx.bin
<path_to_downloaded_file>/A94001sx.bin
    For example:
./A94001sx.bin

This will initiate the installation wizard, which will guide you through the hot fix installation process. During the installation you will be prompted for the SASHOME location to be updated. You should provide the path to the top level SAS directory where the deploymntreg directory exists.


See the Maintenance Installation Tool (MIT) documentation for more details on the installation of hot fixes.


POST-INSTALLATION INSTRUCTIONS

Step 1: Reset the values of the referrer_sk column in the WaMart.session_fact table

1.1 Back up the WaMart.session_fact table of the pre-existing web mart.

1.2 Start an interactive SAS session with access to the Web Analytics macros. See the SAS Web Analytics 5.3 Administrator’s Guide, Second Edition, Appendix 4 "Tips and Techniques", "Accessing SAS Web Analytics Macros" p. 154 for information on how to do this.

1.3 Open the following program, supplied by the hotfix, in the Program Editor window, and change the value of your-mart-name on line 67 to the name of the web mart as it was defined in SAS Data Integration Studio when the mart was created.

Windows: <sasroot>\weba\sasmisc\reset_session_referrer_sk.sas
Unix: <sasroot>/misc/weba/reset_session_referrer_sk.sas

1.4 Submit the modified program. Verify that no warnings or errors were encountered.


Step 2: Recreate the affected aggregate tables

2.1 The aggregate tables created by following jobs must be deleted and recreated. The affected tables are

Job: weba_1500_agg_search_terms
WaMart.aggr_ref_term_day
WaMart.aggr_ref_term_week
WaMart.aggr_ref_term_month
WaMart.aggr_ref_term_qtr
WaMart.aggr_ref_term_year
Job: weba_2100_agg_sebm
WaMart.aggr_sebm_day
WaMart.aggr_sebm_week
WaMart.aggr_sebm_month
WaMart.aggr_sebm_quarter
WaMart.aggr_sebm_year

If you have not yet defined goal pages or non-paid goal pages through the Administrator tab of the UI, the tables from the weba_2100_agg_sebm job will not exist, and do not need to be created.

2.2 If your data repository is SAS, follow the instructions in the SAS Web Analytics 5.3 Administrator’s Guide, Second Edition, Appendix 5 "How the Aggregate Jobs Work", "Regenerate Aggregates" p. 160 to backup and delete the existing data sets, and to rerun the jobs to create the new aggregate data sets.

2.3 If your data repository is Oracle, follow these instructions:

2.3.1 Back up the WaMart aggregate tables within a Web mart that are to be refreshed. Ensure that the ETL is not running, and that the ETL is not scheduled to start.

2.3.2 Delete the above listed aggregate tables from the WaMart library

2.3.3 For the weba_1500_agg_search_terms job:

2.3.3.a In SAS Data Integration Studio, drill down to the Web mart 5.3.1 Jobs\ETL Jobs folder and open the aggregate job to be run.

2.3.3.b Right click on the SESSION_FACT table and select Properties. Under the Physical Storage tab, next to the Library box, click the ... button. Drill down to the Web mart WaMart folder and select the mart’s WaMart library object. Close the SESSION_FACT Properties window. You should see a red O in the upper right corner of the SESSION_FACT object, indicating that it is stored in Oracle.

2.3.3.c Right click on the SQL Join object connected to the SESSION_FACT table. Select Open.

2.3.3.d In the resulting diagram, double click on the Select object.

2.3.3.e In the resulting mapping, on the right hand side, select the SESSION_DT variable under the Target Table column listing.

2.3.3.f Right click on the selected column and select Advanced.

2.3.3.g In the Expression window, follow these instructions to convert the Oracle date value to a SAS date value:

- Select the Functions tab, open the Date and Time functions list, and select the DATEPART function.

- Select the Data Sources tab, open the SESSION_FACT table and select the SESSION_DT column.

- Click OK to close the Expression Builder window.

- Select Up to return to the full diagram window.

2.3.3.h Select Run to run the modified job.

2.3.3.i After verifying that the job ran without errors, and the aggregate tables were successfully created, restore the original settings to the job:

- Right click on the SESSION_FACT table and select Properties. Under the Physical Storage tab, next to the Library box, click the ... button. Drill down to the Web mart WaWork folder and select the mart’s WaWork library object. Close the SESSION_FACT Properties window. You should see a blue S in the upper right corner of the SESSION_FACT object, indicating that it is stored in SAS.

- If you did not save the modified job, you may simply close the job without saving it now, and the case expression modification will not be saved. If you did save the job, Right click on the SQL Join object connected to the SESSION_FACT table. Select Open. In the resulting diagram, double click on the Select object. In the resulting mapping, on the right hand side, select the SESSION_DT variable under the Target Table column listing. Scroll to the Expression column and delete the case expression. Select Up to return to the full diagram window. Save the job and close it.

2.3.4 For the weba_2100_agg_sebm job:

2.3.4.a In SAS Data Integration Studio, drill down to the Web mart 5.3.1 Jobs\ETL Jobs folder and open the aggregate job to be run.

2.3.4.b Click on the Folders tab, drill down to the Web mart WaMart library, and drag the SESSION_FACT table into the diagram.

2.3.4.c Click on the Transformations tab, expand the Data category, and drag the Extract transformation into the diagram. Connect the WaMart. SESSION_FACT object as its input.

2.3.4.d Double click on the Extract Transformation. Select the Mappings tab, and select the SESSION_DT column under the Target table.

2.3.4.e Right click on the selected column and select Advanced.

2.3.4.f In the Expression window, follow these instructions to convert the Oracle date value to a SAS date value:

- Select the Functions tab, open the Date and Time functions list, and select the DATEPART function.

- Select the Data Sources tab, open the SESSION_FACT table and select the SESSION_DT column.

2.3.4.g Click OK to close the Expression Builder window.

2.3.4.h Click OK to close the Extract Properties window.

2.3.4.i Right click on the output table of the Extract transformation and select Properties.

2.3.4.j On the Physical Storage tab, type SESSION_FACT in the Physical Name box. Check the Create as view checkbox. For Location, select Redirect to a registered library. Click the ... button next to the Library box, and select the WaWork library object for your Web mart.

2.3.4.k Click OK to close the Extract Properties window.

2.3.4.l Right click on the line connecting the original SESSION_FACT object to the Search Engine Bid Management Join transformation. Select Delete.

2.3.4.m Drag the output icon of the new Extract transformation and connect it to the Search Engine Bid Management Join transformation.

2.3.4.n At the bottom of the screen, select the Control Flow tab in the Details section of the window. Select the new Extract transformation and use the up arrow button to move it to the top of the list of transformations.

2.3.4.o Now right click on the DETAIL_FACT table and select Properties. Under the Physical Storage tab, next to the Library box, click the ... button. Drill down to the Web mart WaMart folder and select the mart’s WaMart library object. Close the DETAIL_FACT Properties window. You should see a red O in the upper right corner of the DETAIL_FACT object, indicating that it is stored in Oracle.

2.3.4.p Select Run to run the modified job.

2.3.4.q After verifying that the job ran without errors, and the aggregate tables were successfully created, restore the original settings to the job:

- Right click on the DETAIL_FACT table and select Properties. Under the Physical Storage tab, next to the Library box, click the ... button. Drill down to the Web mart WaWork folder and select the mart’s WaWork library object. Close the DETAIL_FACT Properties window. You should see a blue S in the upper right corner of the DETAIL_FACT object, indicating that it is stored in SAS

- If you did not save the modified job, you may simply close the job without saving it now, and the modifications will not be saved. If you did save the job, delete the new WaMart.SESSION_FACT table and the new Extract transformation. Reconnect the original WaWork.SESSION_FACT table to the Search Engine Bid Management Join transformation. Save the job and close it.

2.3.5 Reschedule the ETL.



This completes the installation of hot fix A94001 on Solaris for x64.