Feature Reloading

Overview

This part describes the functionality of reloading features without server restart. Feature may be loaded to server again even if it has already been loaded. But newly loaded feature must have version different from previously loaded. Same version of feature cannot be loaded again.

If same feature is reloaded with new version and Feature Versioning is disabled then all chains within this feature are re-registered under the same names. Actors within reloaded feature are re-registered too. Previously loaded feature with another version is still in memory and all its dependent features are connected to this previously loaded feature. You may reload dependent features too in order to reconnect them to reloaded base feature. To keep correct hierarchy you may setup exact version of base feature in the dependent feature dependencies. In case if no version of base feature is present in feature dependencies then the highest version from loaded base features is used to connect dependent feature to.

Anyway in case if Feature Versioning is disabled and any chain of base feature is called from chain of dependent feature then chain from last loaded base feature is chosen and called. But if chain from dependent feature directly calls actor belonging to base feature then actor is called from base feature which dependent feature connected to. That is if dependent feature connected to previously loaded base feature and its chain calls actor from base feature then actor from previously loaded base feature will be called even if base feature was reloaded.

If Feature Versioning is enabled and on chain call version of called chain is not defined directly then special strategy is used to determine which chain version must be called basing on chain name and processing message. This strategy may be registered for any chain in corresponding feature plugin. See corresponding section of this tutorial to understand how to register this strategy. If no version strategies registered for chain then chain with maximum version is chosen on chain call by default.

There are three ways to initiate feature reloading: putting the file with feature into features subdirectory of Smartactors server directory, calling of special chain of RemoteManagement feature with path to feature file in the file system, calling of special chain of RemoteManagement feature with description of feature and description of remote repository from where feature can be downloaded.

Loading/reloading feature using features subdirectory on Smartactors server.

When Smartactors server is running you may just put the file with the feature you want to load into features subdirectory. Then server catch the file and load feature immediately. Note that file with feature must not have name which already present in features directory. If you want to reload feature which already present in features directory then just rename its file to different name (best way is to include version into feature file name).

Loading/reloading feature from file on the server

To use this way please make sure that remote-management feature is present in corefeatures directory of your Smartactors server. If so then to reload feature you may call chain load-feature-from-file with message which contains parameter featureLocation with full feature file path in value. Please see below the sample of message to call this chain through routing:

{
    "messageMapId" : "load-feature-from-file",
    "featureLocation" : "/home/user1/sm_server/features/my_feature.jar"
}

Note that load-feature-from-file chain has no external access, therefore it can be used only from inside of server. To call it from outside you may create your own externally accessible chain which calls this chain (you chain may contain some authentication).

Loading/reloading features from remote repository

To use this way please make sure that remote-management feature is present in corefeatures directory. If so then to reload features you may call chain load-features-from-repository with message which contains the description of features and repositories. Please see below the example of message to call this chain through routing:

{
    "messageMapId" : "load-features-from-repository",
    "repositories" : [
        {
            "repositoryId" : "archiva.smartactors-features",
            "type" : "default",
            "url" : "http://archiva.smart-tools.info/repository/smartactors-features/"
        }
    ],
    "features": [
        {
            "group": "info.smart_tools.smartactors",
            "name": "endpoint",
            "version": "0.6.0"
        },
        {
            "group": "info.smart_tools.smartactors",
            "name": "endpoint-plugins",
            "version": "0.6.0"
        }
    ]
}

Format of repositories and features json variables is same as in features.json file usually stored in corefeatures server directory.

Note that load-features-from-repository chain has no external access, therefore it can be used only from inside of server. To call it from outside you may create your own externally accessible chain which calls this chain (you chain may contain some authentication).