4.14. conga

Bug Fixes

BZ#514679

Previously, a non-English locale caused luci, to display improper copyright information in the pages' footer. The respective template has been fixed so as not to translate the application-specific part and the whole footer now displays information correctly even in non-English locales.

BZ#853018

Prior to this update, luci contained non-visual links dedicated to better browsing experience in the agents supporting it, such as navigation to the site map, the access to which resulted in an error. The page titles also contained some rare inconsistencies. The luci templates clean-up has been provided to fix respective corner cases.

BZ#872645

For extremely large cluster.conf files, the luci serializer could not handle the complexity of a cluster configuration object as the serializer run into the recursion depth limit. Consequently, error messages were returned in luci. Caching of the object representing cluster configuration has been abandoned so that this error no longer occurs.

BZ#883804

Previously, the ricci(8) and luci_admin(8) manual pages of the respective conga packages were installed with incorrect permissions. File permissions have been corrected in the installation procedure and the manual pages are now installed correctly.

BZ#887170

Each time luci was used to start or restart a cluster, or to have previously inactivated node rejoined the cluster, it made cluster services such as cman, rgmanager or clvmd enabled on boot on the respective cluster nodes. This can interfere with the user's preferences, for instance, when running a 2-nodes cluster without a quorum disk and having the services disabled on purpose on one of the nodes to prevent fence races. To avoid this, Conga has been changed so that it no longer modifies the existing settings in the mentioned cases, while still enabling the services when the cluster is created or a new node is added.

BZ#965785

Previously, a segmentation fault occurred in the ricci daemon when processing cluster.conf files with very large values. This was caused by allocating large amounts of memory that were not available on the stack. A patch has been introduced to allocate memory on the heap and provide an error message if not enough memory is available. As a result, the segmentation fault no longer occurs.