TIBCO ActiveMatrix(R) BPM 4.1.0 Hotfix 007 is available

TIBCO ActiveMatrix(R) BPM 4.1.0 Hotfix 007 is available

book

Article ID: KB0102269

calendar_today

Updated On:

Products Versions
TIBCO BPM Enterprise (formerly TIBCO ActiveMatrix BPM) 4.1.0

Description

Product Name: TIBCO ActiveMatrix(R) BPM
Release Version: 4.1.0_HF-007
Release Date: January 2018

================================================================================

Closed Issues in 4.1.0_HF-007 (This Release)

Note: Where the closed issue is related to a TIBCO Support case, the relevant
case number (CN-xxxxxxxx) or Support Request ID (SR-xxxxxx) is given in brackets
after the Issue number.

WRM-10176
---------
An active process instance may hang after a failure and restart of a BPM node,
if that process instance had been running a child service process instance when
the BPM node failed.

Because service processes run in-memory, they must be recovered after a node
failure, allowing the parent task to complete and its instance to continue.
Note that any data generated during the in-memory instance's processing cannot
be recovered. This task is performed by a process engine recovery thread. (This
thread runs periodically, as specified by the
com.tibco.bx.engine.recovery.failure.threshold JVM property - by default, after
30 seconds of inactivity of the JVM process that runs the node).

However, this thread only recovers service process instances associated with
valid BPM nodes (as identified in the PVM_ENGINE table in the BPM database). Any
service process instance associated with a no longer existing BPM node (that is,
without a corresponding entry in the PVM_ENGINE table) would not be recovered.
This meant that the parent process instance could not continue, and so would
hang indefinitely.

Now, an additional process engine recovery thread is run to recover any such
'orphaned' service process instances, allowing the parent process instances to
continue. The thread runs periodically, as specified by the
pvm.bpel.bg.inMemoryRecovery.interval JVM property, which takes a periodic value
(in minutes) with a minimum value of "P5".

WARN messages are written to the BPM log file identifying service process
instances that have been recovered by either of these threads. Messages
containing the string '_recoverEngine' identify instances associated with the
BPM node being recovered. For example:
 
   [WARN ] com.tibco.pvm.system.PvmSystem - _recoverEngine(1): recovering
   in-memory instance [101]; spawner task [CallSubProcess3]; parentProcessId
   [33]

Messages containing the string 'IdbInMemoryInstanceRecovery' identify orphaned
instances being recovered. For example:

   [WARN ] com.tibco.pvm.system.PvmSystem - IdbInMemoryInstanceRecovery:
   recovering in-memory instance [40]; spawner task [CallSubProcess2];
   parentProcessId [18]

WRM-10175
---------
An active process instance could hang waiting for a child service process
instance to complete. The child service process instance would never complete
because it had been suspended or cancelled by an event intended for a different
child service process instance, sent by that child's parent process instance.

This problem could occur because service process instance IDs were not globally
unique. Now, globally unique service process instance IDs are allocated from a
database sequence value (and stored in the PVM_INST_PROCESS_IM_SEQ database
table).

WRM-10139 (CN-01543457)
-----------------------
If a parent process instance sent a SUSPEND, CANCEL or RESUME event to a child
service process that had failed, the event could not be processed. A WARN
message was written to the BPM log file. For example:

   [WARN ] com.tibco.pvm.system.PvmSystem - Unable to schedule life cycle event
   on in-memory process: PVM-MODEL-102501: Life Cycle command 'SUSPEND' is not
   llowed from 'FAILED' status

The process engine continued to try and process these events for the failed
service process instances. If a backlog of such events built up, processing of
service process instances would be slowed and CPU usage on the BPM node would
increase significantly.

Now, completed, halted or cancelled service process instances are removed from
the in-memory work list. Any associated events are also cleared.

================================================================================

Refer to the associated readme document for any additional information.

Issue/Introduction

TIBCO ActiveMatrix(R) BPM 4.1.0 Hotfix 007 is available

Environment

All Platforms

Resolution

The hotfix can be downloaded from the TIBCO Support Customer Portal Web UI,
using your username and password for the TIBCO Support Web.

Once logged on the hotfix can be found under the Downloads Menu:
AvailableDownloads/AMX_BPM/4.1.0/TIB_amx-bpm_4.1.0_hotfix007

Additional Information

TIB_amx-bpm_4.1.0_HF-007_readme.txt

Attachments

TIBCO ActiveMatrix(R) BPM 4.1.0 Hotfix 007 is available get_app