- When defining a cluster and checking the Use the Same Password for All Nodes box, an error message, saying that the user did not enter a password for any of the clusters except the first one, appeared after the user had submitted the changes. The password synchronization was triggered only once by checking the box. As a consequence, if no password was entered before checking the box, none could be copied into the other password fields. This update fixes the problem so that filling any single password field in the form causes the password to be correctly submitted for every node affected.
- Upgrading or downgrading the luci package could result in SELinux AVC (Security-Enhanced Linux Access Vector Cache) denials due to the python applications searching for local customizations in the
Homedirectory. With this update, Python's paste tool now uses the
-Esflag, and so avoids this behavior.
- If changes were made in any of the lightbox dialog boxes, a dialog box was hidden instead of being reset after it had been closed. As a consequence, when the dialog box was reopened, it was in the same state as before closing. The only way to reset the state of the dialog box was to refresh the page in the browser. Now, when the user closes the existing dialog box, the page refreshes automatically instead of trying to recreate the initial state.
- Previously, titles of certain dialog boxes in the luci UI contained inconsistent character casing. Now, character casing is consistent in all titles.
- Previously, luci did not provide any way to back up or restore the content of the database. With this update, the content of the luci database located in
/var/lib/luci/data/luci.dbcan be fully backed up and restored.
- In luci, when editing a failover domain that had both the Restricted and Prioritized boxes unchecked, adding of a node had no effect. The operation appeared to be successful, however the node was not added to the domain in the cluster.conf file. With this update, nodes can be successfully added to the failover domain. Removing nodes from the failover domain now works correctly as well.
- Previously, a bug in luci resulted in luci not being able to parse lines which contained the name of a service ending with the ".1" suffix. As a consequence, when importing a cluster, luci logged an error or displayed the Error 500 message in the browser. This update removes the bug and luci can now parse names ending with ".1" correctly.
- When the user created a file system resource and then tried to edit any field of a cluster service, an error message was printed and the changes were not applied. With this update, the source code is modified so that the changes are successfully applied.
- Previously, the Run Exclusive checkbox on the administration panel for a service group did not correspond to the configuration of this service group's entry in the
cluster.conffile. The Run Exclusive option was enabled in luci by default, without it being manually enabled, and services could therefore become exclusive without users knowing about it. Now, luci is modified to correspond with the
cluster.conffile: if the
Run Exclusiveoption is not enabled, the checkbox is not checked.
- Due to luci not showing the migrate action for virtual machines, the Error 500 error message could appear when attempting to create a cluster of KVM (Kernel-based Virtual Machine) guests. With this update, luci is modified so that the services can be started and edited successfully.
- Prior to this update, the stop/start service was performed instead of migration when migrating a virtual machine by choosing the Migrate to node... option in the drop down menu on the Service Details page. With this update, the source code is modified to successfully complete the migration of virtual machines using the web interface.
- Prior to this update, if Log debugging messages and Log messages to log file were enabled in the Logging page and the user entered a file path in the textbox below, the Error 500 message appeared. The changes to logging could not be submitted as a consequence. This problem has been fixed and changes to logging are submitted correctly.
- Prior to this update, luci served web pages and XHTML documents for most of the web browsers with the "application/xhtml+xml" content type and with "application/xml and "text/html" as failback. Web pages for the Internet Explorer browser were consistently served with the "text/html" content type. As a consequence, users were unable to open the URL of the luci server using Internet Explorer 8.0 and an error message appeared instead. With this update, web pages are permanently served with the "text/html" content type. The login page shows up correctly and users are able to log in.
- When trying to save an empty value for an option that used to be non-empty before instead of removing such option from the cluster.conf file, it was saved as the previous non-empty value. If the user attempted to clear options for a virtual machine service that he had been configured, luci did not save configuration for the service and it was therefore impossible to clear them using only the web interface. This problem has been fixed and the options are now cleared properly.
- Previously, luci did not start a service on a preferred node. When the user chose a service with a failover domain on the Service Groups page, a service started on the first node even if the user did not choose the first node as the preferred one. Now, the Submit button is used instead of a link for the headers_detail form and thus fixes the problem.
- The previous version of luci did not provide any role-based access control system. With this update, luci contains a new system for managing user roles. Multiple users can now gain various privileges for managing or accessing clusters.
- Now, a warning message is displayed when the user logs in for the first time in the luci UI. The text warns users about possible problems related to managing clusters using the UI with no further knowledge of clustering.
- This update provides a confirmation dialog box, which appears when removing cluster nodes or the whole cluster by selecting all the nodes and clicking the Delete button.
- This update adds support for the new