BPM Technical Solutions

ref: https://www.ibm.com/developerworks/community/forums/html/topic?id=77777777-0000-0000-0000-000014924184
There a couple typical cases for a "memory leaks". (I put memory leak in quotes, because the system does eventually recover the memory, it just has to wait for the bean to timeout. Typically a few hours, if I recall.) The first potential leak is when the user just closes their browser window without reaching an end or a postpone. This usually isn't a big deal: the system will eventually get the memory back and majority of the time the users will do the right thing and go to the postpone/end. A few exceptions won't be a problem. The more nefarious case is when a BPM author tries to be tricky and deliberately creates a coach that doesn't end. For example, when generating a coach based report. Or when creating a coach that issues a redirect to an external web page. This is nefarious because this "leak" will happen every time that service is used and therefore can happen much more frequently in aggregate. These are often referred to as "never ending services" in some of the performance documents. This isn't to say that you should never create these kinds of services, just that you need to be aware of the memory consequences.

So, some general advice on memory:
 

• Most importantly, don't keep variables around longer than you need them. If you get a big XML document back from the integration service, and then you parse than XML into BPM variables, set the XML document to null once you are done with it. This not only minimizes the amount of memory you are keeping in the session but also minimizes the data that is being saved and loaded to the database with every execution context save.
• Don't have data in the service that you don't need. This is somewhat of a corollary to the previous one, but only send the objects to the service that are needed. No sense in sending data from the BPD to the service if it isn't needed.
• Avoid never ending services when you can.
• If you must have a never ending service (and there may be times that you do need to create), then make sure to null out all of the variables and be very conscious of your memory usage. For example, if you are creating a coach based report that will not "end", you can use a custom HTML block at the end of the page to null out all of the variables used to create the report.
• Encourage users to use postpone buttons rather than just closing the browser window. Not only does it save memory but it also ensures that the user's data is saved

ref: https://www.ibm.com/developerworks/community/forums/html/topic?id=77777777-0000-0000-0000-000014924184
There a couple typical cases for a "memory leaks". (I put memory leak in quotes, because the system does eventually recover the memory, it just has to wait for the bean to timeout. Typically a few hours, if I recall.) The first potential leak is when the user just closes their browser window without reaching an end or a postpone. This usually isn't a big deal: the system will eventually get the memory back and majority of the time the users will do the right thing and go to the postpone/end. A few exceptions won't be a problem. The more nefarious case is when a BPM author tries to be tricky and deliberately creates a coach that doesn't end. For example, when generating a coach based report. Or when creating a coach that issues a redirect to an external web page. This is nefarious because this "leak" will happen every time that service is used and therefore can happen much more frequently in aggregate. These are often referred to as "never ending services" in some of the performance documents. This isn't to say that you should never create these kinds of services, just that you need to be aware of the memory consequences.

So, some general advice on memory:
 

• Most importantly, don't keep variables around longer than you need them. If you get a big XML document back from the integration service, and then you parse than XML into BPM variables, set the XML document to null once you are done with it. This not only minimizes the amount of memory you are keeping in the session but also minimizes the data that is being saved and loaded to the database with every execution context save.
• Don't have data in the service that you don't need. This is somewhat of a corollary to the previous one, but only send the objects to the service that are needed. No sense in sending data from the BPD to the service if it isn't needed.
• Avoid never ending services when you can.
• If you must have a never ending service (and there may be times that you do need to create), then make sure to null out all of the variables and be very conscious of your memory usage. For example, if you are creating a coach based report that will not "end", you can use a custom HTML block at the end of the page to null out all of the variables used to create the report.
• Encourage users to use postpone buttons rather than just closing the browser window. Not only does it save memory but it also ensures that the user's data is saved

select proj.name,proj.short_name

from BPMDB.lsw_project proj

where is_system = 'F'

and proj.is_archived = 'F'

and proj.is_toolkit = 'F';

when users login into BPM Portal, the BPM engine queries a lot of data, and it also queries the exposed services in all process apps and toolkits which are available to that users. It will increase a lot of response payload if we expose unnecessary services which means very slow response when user login into the BPM Portal

Soft reference objects, which are cleared at the discretion of the garbage collector in response to memory demand. Soft references are most often used to implement memory-sensitive caches.

http://docs.oracle.com/javase/7/docs/api/java/lang/ref/SoftReference.html

If in DEV environment, we have problem with unnamed and unneeded snaphots, read the item1 for understanding unnamed snapshots.

For higher environments, we just have unneeded snapshots.

1. Clear unnamed snapshots:

http://pic.dhe.ibm.com/infocenter/dmndhelp/v8r5m0/index.jsp?topic=%2Fcom.ibm.wbpm.admin.doc%2Fmanaginglib%2Ftopic%2Fmanaging_snapshots_j.html

2. Clear unneeded snapshots:

http://pic.dhe.ibm.com/infocenter/dmndhelp/v8r5m0/index.jsp?topic=%2Fcom.ibm.wbpm.admin.doc%2Fmanaginglib%2Ftopic%2Fmanaging_snapshots_i.html

http://www-01.ibm.com/support/docview.wss?uid=swg21601357

http://pic.dhe.ibm.com/infocenter/dmndhelp/v8r5m0/index.jsp?topic=%2Fcom.ibm.wbpm.admin.doc%2Ftopics%2Fmanaging_orphans_with_policyfile.html

date_picker.css

.Date_Picker_String .dijit div.dijitReset{
    background-image: url('dateTimePicker_calendar.png');
}

Add a zip file contain this date_picker.css and dateTimePicker_calendar.png and named as date_picker.zip

Upload this zip file as a managed file(Web file)

In CV, refer to this date_picker.css which resides in date_picker.zip file.

indexingUIContainer: top level CV coach View Control ID. This line below to re-render the whole sub cv

com_ibm_bpm_global.topLevelCoachViews.indexingUIContainer.context.refreshView()

Objective: Sending an XML message to Websphere MQ to trigger UCA

Example Solution:

1. Provided a BPD as below

2. UCA defined:

Event Message: TestUCA. This name is critical to use in event message later

3. DummyService:

Input variable name = request. This name is used later in message

4. prepare a message in xml

<?xml version="1.0" encoding="UTF-8" standalone="no"?>
<eventmsg>
    <event processApp="DANSBOX" ucaname="TestUCA">TestUCA</event>
    <parameters>
        <parameter>
            <key>request</key>
            <value>7300050241</value>
        </parameter>
    </parameters>
</eventmsg>

In this message. TestUCA = Event Message. DANSBOX = processApp acronym. request = name of input variable in dummyService of UCA

5. Find the Websphere MQ configure in WAS

This page on Websphere will show Queue name, host, queue manager and port.

Use these information and send the xml message above to trigger the UCA

REF: http://www.ibm.com/developerworks/bpm/library/techarticles/1207_chandra/1207_chandra.html

dojo.query('iframe')[1].contentDocument.getElementById('id')

Provided:

Configure in Process Admin:

Group: "SuperLeader 2"

User: "user1", and  "user2" belong to "SuperLeader 2"

Objective: rout to user or group

Solution:

Use custom routing , even though it's deprecated in BPM8.5. In expression of Custom(Deprecated) routing, specify: user or role

Rout to  user: specify user name example:"user1"

Rout to Group: specify: ROLE:<<groupName>> example:  "ROLE:SuperLeader 2"

OBJECTIVE: call a store procedure which return a list of data

PROVIDED: we have procedure in database

PROCEDURE retrieve_allusers_same_roles(p_user_name IN VARCHAR2,
                        user_details_list   OUT refcursor)  IS
    BEGIN
        OPEN user_details_list FOR
        SELECT DISTINCT user1.user_id as usrUserID,
                    user1.user_name as usrUsername,
                    user1.full_name as usrFullname
        FROM BPMDB.LSW_USR_XREF user1 inner join BPMDB.lsw_usr_grp_mem_xref usergrp2 on user1.user_id=usergrp2.user_id
        WHERE usergrp2.group_id in (
            SELECT usergrp1.group_id
            FROM BPMDB.LSW_USR_GRP_MEM_XREF userGrp1 inner join BPMDB.LSW_GRP_GRP_MEM_XREF grpgrp1 on usergrp1.group_id = grpgrp1.group_id
                ,BPMDB.lsw_usr_grp_xref grp1
                ,BPMDB.LSW_USR_XREF user2
            WHERE upper(user2.user_name)=upper(p_user_name)
                AND user2.user_id = usergrp1.user_id
                AND grp1.group_id = grpgrp1.container_group_id
                AND grp1.group_name = 'Business Team');
            
    END retrieve_allusers_same_roles;

SOLUTION:

1. Prepare param list

tw.local.parameters = new tw.object.listOf.SQLParameter();

var parameter = new tw.object.SQLParameter();
parameter.value = tw.local.username;
parameter.type = 'VARCHAR';
parameter.mode = 'IN';
tw.local.parameters.insertIntoList(tw.local.parameters.listLength, parameter);

var parameter = new tw.object.SQLParameter();
parameter.value = tw.local.results;
parameter.type = 'ORACLE_CURSOR';
parameter.mode = 'OUT';
tw.local.parameters.insertIntoList(tw.local.parameters.listLength, parameter);

NOTE: the tw.local.result datatype: list of "UserDetail"

"UserDetail" complex object  defined in PD:

usrUsername: String
usrFullname: String
usrUserID : String

With this technique of controlling the Output of proc, BPM automatically maps the output of this pro "retrieve_allusers_same" into list of UserDetails for us

2.  Set the sql for the "SQL Call Stored Procedure"

CALL retrieve_allusers_same_roles(?,?)

3. Configure the "SQL Call Stored Procedure" service

tw.local.results is list of "UserDetail". BPM automatically map the result list into a list of business object for us.