Subsections
Preferences
Each Eclipse-based application provides a set of pages for changing user preferences. The basic Eclipse framework already allows for the customisation of settings like colours and fonts to be used throughout the client or key bindings for certain actions. Additional plugins may extend the existing set of preferences. The UNICORE specific plugins use this mechanism for introducing a whole range of new options that determine their mode of operation. This section discusses these UNICORE-related preference settings.
Figure 19:
The preferences page
|
|
In order to manage a large set of different preferences Eclipse organises them in categories that form a tree structure. This structure is shown to the left of the preferences panel and allows to select the category of options you wish to change. Depending on the selected category a different preference page opens on the right side of the panel. In the first example (Figure 19), the category Colours and Fonts has been selected. Clicking on Change... opens a font selector dialog. You may Apply the change or Restore Defaults. OK leaves the Preference Panel retaining the changes while Cancel leaves the panel and discards the changes.
Figure 20:
UNICORE related preferences
|
|
There is a dedicated category containing general UNICORE related preference settings. The following options can be modified:
- User Expertise Level: The client targets a wide range of users with varying experience and knowledge about the Grid. Switching to higher levels will reveal more details to the user and enable additional actions. The default is beginner.
- Default file transfer: Users can choose from different protocols to use for transferring files to and from UNICORE storages. Available protocols include the Baseline file transfer (a fast HTTP based protocol) and the OGSA Random Access ByteIO (a slower but widely supported protocol).
- Connection timeout: This option determines how quickly the client will give up when a server does not respond to its requests. The default is 20,000 milliseconds. Caution: setting this value too low will altogether prevent the client from communicating with Grid services.
- HTTP Proxy settings: If you connect to the internet via an HTTP proxy server, you can provide the connection details in this form.
Figure 21:
Grid Browser preferences
|
|
The following options can be found in the Grid Browser sub-category:
- Number of Grid Browser levels to refresh on startup: An integer number specifying how many levels of services in the Grid Browser should be refreshed when the client is started. Refreshing means that the services are contacted in order to update cached information about them. Zero or negative numbers will prevent the Grid Browser from refreshing services during startup.
- Hide failed Grid services: When enabled, this option will filter out all services that cannot be contacted by the client. These services will not be displayed in the Grid Browser anymore.
- Hide Grid services that you don't have access to: When enabled this will filter out all Grid services that could not be contacted due to access restrictions.
- Chunk size when listing many services: Target systems can execute many jobs and Grid storages can contain lots of files. If the client just tried to list all jobs, services, and files on the Grid, it could easily run out of memory. Therefore, the number of child nodes displayed in the Grid Browser when refreshing its parent node is limited. Additional nodes can be displayed by double-clicking or expanding a node called `...' which can be found at the end of the list of child nodes when the parent node seems to contain additional children. The number that is entered in this preference field determines the number of child nodes that are listed when first refreshing its parent and the number of nodes that are shown in addition, when the `...' node is opened.
Figure 22:
Preferences regarding installed application GUIs and job descriptions
|
|
The following options can be found in the Jobs sub-category:
- Default lifetime for jobs: The amount of time after which submitted jobs are destroyed and cleaned up automatically. Once a job has been destroyed, its output files cannot be restored. This option can be overridden by setting the job's lifetime in the Application view (see 3.5.1).
- Application directory: The location of a directory that holds all application specific user interfaces that have been installed so far. New user interfaces for applications can either be downloaded from the Grid (see 4.3.4) or manually copied to this directory.
- Time between polling job status: In monitoring mode, the job editor occasionally polls for changes in the submitted job's execution status until the job has finished or failed. This option will set the amount of time until the next status check occurs.
Figure 23:
Security related preferences
|
|
The following options can be found in the Security sub-category:
- Keystore file: The location of the Java keystore (.jks) file to be used for holding the user's private keys and the trusted certificates. The location can be changed by either directly entering a new path or clicking the Browse button, which opens a file browser.
- Store the user's password in the encrypted keyring file: Users have frequently asked for a way to start the client without having to supply the same password each time. When this preference is set to true, the client will save the user's password in an encrypted file managed by Eclipse, the so-called keyring file. However, using this feature is NOT RECOMMENDED by the UNICORE team, as it weakens the security model of the Grid: An intruder will be able to take over a user's Grid identity simply by starting the client.
- Show certificate expiry warnings at client startup: When this option is enabled, the client will issue a warning during startup for each certificate that has expired or will expire soon.
Figure 24:
Workflow specific preferences
|
|
The following options can be found in the Workflows sub-category:
- Default lifetime for workflows: The amount of time after which submitted workflows are being destroyed and cleaned up automatically. All jobs that have been submitted during workflow execution are cleaned up as well.
- Time between polling workflow status: In monitoring mode, the workflow editor occasionally polls for changes in the submitted workflow's execution status until the workflow has finished or failed. This option will set the amount of time until the next status check occurs.
Bastian Demuth
2010-05-25