Resolution:
Description:
============
By default, in TIBCO KPSA, all pending Service Order are externalized into files allowing the application to be recovered from the filesystem. The checkpoint recovery restart is specific and has to follow the right process, otherwise you might encounter the known issue described hereafter.

Environment:
==========
TIBCO KPSA Mobile or Broadband 3.5+

Symptoms:
=========
After a checkpoint recovery, during runtime, if you encounter the following trace in the engine logs :

2012-03-02 15:00:03.717490|USR|WARN |6759|builtin.cpp(193)|Could not dequeue from &ltPROVCA:TestDriver:S[0]:Q_QueueOut> in doPump.

Cause:
======
This means that the corresponding order queue has been corrupted, very probably by the checkpoint recovery.

Resolution:
=========
Loading checkpoints can be done only once at startup in a new node and not at runtime. It should only be performed before enabling client adapters or injecting any traffic. The recommended procedure is as follows:
1/ Deploy the node.
- provadmin install kspa
- provadmin start kspa
- provadmin configura ca
- provadmin install cart
- provadmin start cart

2/ Make sure the above commands have been executed successfully and then execute the below command:  
- provadmin load checkpoints

Note that by default, this command will stop flow controller, wait for it to be stopped, load the checkpoints and then automatically restart flow controllers. But if the parameter "autoStartFC" is set to "false" in the load command, then the flow controller must be started manually using "provadmin start fc" command.

3/ Resume order injection.
- provadmin unlockall ca
- provadmin enableall ca