2 Preparing to Upgrade a Connect Instance
In a running Connect application instance, the state of the system is stored in the database and in the storage log files that are created on the different nodes of the Connect cluster. Before Connect can be upgraded and redeployed, the information in the storage files must be copied into the database by flushing the system. Thus, before starting the upgrade, you must enable System Flush, which forces Connect to stop sending messages and to process any storage log files that have not yet been saved in the database.
Enable System Flush
The Enable System Flush control is in the Deployment Editor. Perform the following steps for each active deployment in your Connect instance.
Navigate to the Deployments tab.
Select the deployment in the left hand pane. If the Deployment is active, you will see the real-time status as in illustrated in the window below.
Click the Machines tab in the right-hand pane. You will see a screen that looks like this.
Click the "Flush Status: Enable" button to begin flushing the system.
The button next to Flush Status will change to “Disable”. The flush process may take several minutes, depending on the volume of outstanding data. It it important to wait for this step to complete before upgrading Connect. Otherwise some campaign execution data may be lost.
Troubleshooting System Flush
If the flush process appears to be stalled, it may be due to stale records in the STORAGE_HISTORY table. To view unprocessed STORAGE_HISTORY records, run this query in the database:
select * from STORAGE_HISTORY where STATUS_CODE < 30
Unprocessed storage records that are non-existent and/or old may have been caused by a previous redeployment that was performed without proper flushing.The following query can be used to tell Connect to ignore any unprocessed storage records:
update STORAGE_HISTORY set STATUS_CODE = 40 where STATUS_CODE < 30
Once this query has been executed, refresh the Deployment Editor page in the browser.
Stopping Connect
After the flush operation finishes, all the individual machines running on the cluster must be stopped. To do this, mouse over the Actions menu for each machine and click the Stop action. It can take several minutes for all machines in the cluster to stop running. The status column in the center of the screen shows whether the machines are “Running” or ”Not Running.” It may be necessary to use the Refresh command in the Commands menu at the top of the right-hand pane.
IMPORTANT: QA Diagnostic Environment MUST be disabled at this point. Ensure that the QA Diagnostic is set to "OFF".
Cleaning Storage
Once all machines have stopped, the Status column should indicate a status of “Not Running” for all the machines, as shown below.
After all machines have stopped running, click Clean Storage. Upon clicking Clean Storage, the following dialog appears:
Click OK if you are ready to proceed.
Stopping Services
Each node in the Connect application cluster will have a process that is running runmonitor.sh. This process must be stopped on each machine, which can be accomplished by running the S80connect script:
$ /etc/init.d/S80connect stopDepending on how Connect is deployed, root or sudo access may be required
Backing Up Current Configuration Data
In the current installation directory, find and save the following files into a temporary directory (other than the install directory):
database.propertiesinstall.propertiescrypto.propertiesAlso ensure any files container cryptographic keys are backed up. The information in these files will be required when installing the new version of the Connect software.