The Template Engine works in phases. It alternately imports information about CMS file versions from the Content Management Server (import phase) and generates web documents which are then switched online (export phase). The export phase is divided into several subphases. The tasks of the individual phases are described in the following:
import: The Template Engine receives so-called update records from the Content Manager, i. e. information about updated files. If the system includes the NPS Portal Manager, the Template Engine also receives update records relating to channels and news articles.
importCB (from Version 6.7.2): The Template Engine executes a callback function that has access to the journal of import actions and may trigger additional actions.
export-init: The export requests resulting from the update records are processed and grouped. The grouping makes it possible to have several slaves export files.
export-fill: The export requests are executed which means that files are exported to the directory hierarchy switched offline. This phase (and therefore the export procedure as a whole) is aborted, if a configurable number of files could not be exported (system configuration entry
tuning.export.acceptableFailures). If the system includes the Verity Cartridge, the generated documents are indexed in parallel.
export-fillCB: The Template Engine executes a callback function with which the updated directory hierarchy can be replicated as desired.
export-switch: The updated directory hierarchy is switched live by swapping the targets of two symbolic links pointing to the online and offline hierarchies.
export-switchCB: The Template Engine executes another function with which the targets of the symbolic links can be swapped on other webserver systems as well.
export-sync: The directory hierarchy now switched offline is updated.
export-syncCB: This function allows you to update a remote directory hierarchy which is currently switched offline.
export-sync-finished: The journal of changed files which the Template Engine creates and processes during the
export-sync phases, respectively, is rotated. This ends the export phase, and the Template Engine returns to the import phase.
The Template Engine enters the export phase, when it receives a corresponding instruction from the Content Management Server or one of the
app publish or
app export Tcl commands is executed. While
app publish is used to run a complete export and publication cycle or to resume an interrupted cycle,
app export exports the files and then returns to the import phase. Since the Template Engine exports files incrementally,
app export can be used to reduce the publication process (
app publish) to swapping and synchronizing the directories, hence speeding it up. Only errors should make it necessary to use these commands since the server normally receives equivalent instructions as update records from the Content Manager by means of system jobs. Please refer to the Administration/Layout manual for details about these jobs.
If one of the export phases up to and including
export-switch fails, the Template Engine returns to the import phase. If, on the other hand, one of the next phases fails, the error state persists. In this case it is absolutely necessary to remove the error cause first. Then the export procedure can be resumed with
app publish, i. e. the directory hierarchies can be swapped and the necessary synchronization processes can be performed. If the cause of the error can not be eliminated (because, for example, a remote machine is unreachable), the functions can be disabled, if necessary, by commenting out their code (it is then required to restart the server).
The Template Engine normally delivers content located on the system on which it is installed, meaning that it exports the files into a local, offline directory hierarchy which is then switched online. Finally, the hierarchy previously online is synchronized with the hierarchy which is currently online so that the next export is based on the current data.
With distributed systems, these processes can, by means of callback functions, also be performed on one or more remote machines. The functions call the custom procedures described below. Please note that the functions need to indicate an error using the
error command. If a callback function failed and is then called again, the result must be the same as if it had been successful upon the first call. Callback functions must not modify the contents of the export directory tree.
The callback functions need to be defined as server commands in a Tcl script file which must be placed in the instance-specific
Phase importCB (from version 6.7.2), callback function importCallback
As update records are being imported, the Template Engine creates a journal of the changes made to CMS files. This callback function can process the journal and trigger additional actions depending on the file modifications.
Phase export-fillCB, callback function exportFillCallback
Prior to executing the callback function, the Template Engine finishes the export into the offline directory. After execution, an equally complete offline directory hierarchy must exist on the remote machine. It must be possible and safe to execute the callback function several times.
Phase export-switchCB, callback function exportSwitchCallback
The task of this function is to swap the online and offline directory hierarchies on a remote machine. When the callback function is called, the local online directory hierarchy contains the new data and the offline hierarchy contains the data previously online. Furthermore, the remote machine must be in the state in which it was after the
exportFillCallback was successfully executed, i. e. the offline hierarchy contains the new data, and the online hierarchy the old data. After successful execution, there must exist an online and an offline directory hierarchy which have been swapped so that their versions correspond to the respective versions on the local system. In the case of an error, the remote directory hierarchies must not have been swapped.
Phase export-syncCB, callback function exportSyncCallback
The task of this callback is to update the offline hierarchy on the remote machine so that it corresponds to the online hierarchy. Prior to calling the function, the Template Engine synchronizes the local hierarchies. The callback function is only executed if the
exportSwitchCallback was executed successfully, i. e. if the current data is online and the outdated data is offline. After the function has been successfully executed, both remote hierarchies must have the same versions.