- Previously, the resource and service names in the
/etc/cluster/cluster.confconfiguration file that contained non-standard characters, like hash (
#), question mark (
?), or slash (
/), were not properly handled by the luci application. Consequently, when processing such configuration file, luci failed with the following message:
Error 500 We're sorry but we weren't able to process this request.This bug has been fixed, and luci now handles resources and services whose names contain the aforementioned characters without complications.
- When the
fenceinstance was configured with the
delayattribute in the
/etc/cluster/cluster.conffile, the luci application ignored the subsequently enabled
unfenceinstance that was configured without the
delayattribute. The unfence status was incorrectly displayed as disabled in the
luciinterface, but unfencing was performed without complications. With this update, the underlying source code has been modified to address this issue. As a result, unfence is now properly reported in
- Previously, it was possible to create a fencing device with an invalid name (starting with a number) using the luci application. The device was successfully created, but the
/etc/cluster/cluster.conffile did not pass the schema validation check. The bug has been fixed, and a warning message is now displayed to prevent users from setting invalid device names in the
- Previously, certain errors related to the communication between the luci and
ricciapplications could have been dropped without notification to the user. Also, the following message could occur in the
No object (name: translator) has been registered for this threadWith this update, this behavior has been modified and the described errors are now properly written to the log file.
- Prior to this update, a double click on thebutton in the dialog window led to listing the cluster twice. With this update, the underlying source code has been modified to address this issue, and the cluster is now listed only once regardless of how many times the button was pressed.
- Previously, when attempting to create a service that referenced the same global resource twice, the luci application terminated unexpectedly with the following message:
A resource named "<name>" already existsThis bug has been fixed, and luci now accepts multiple references inside a service group.
- Previously, the luci application allowed the
__max_failuresvariables to be set without setting their corresponding timeout variables (
__failure_expire_time), and in the opposite way. This behavior has been changed, and an error is now issued in case the corresponding variables are not set.
- When the self_fence property was enabled using the
luciinterface, the corresponding entry in the
/etc/cluster/cluster.conffile was written incorrectly. A value was assigned in the form of
self_fence="yes". Consequently, fencing actions failed. The bug has been fixed, and self_fence is now assigned with the correct value. As a result, fencing now works properly when enabled with luci.
- Prior to this update, resource and service attributes that accept boolean input did not use consistent values to denote enabled or disabled status. The accepted values were:
false. With this update, only the values
0are accepted in attributes that use boolean input.
- Previously, after renaming a fencing device with an enabled
unfenceinstance was not updated with the new name and referred to a non-existent device. This bug has been fixed, and an
unfencereference is now correctly updated when a fencing device was renamed.
- Prior to this update, the luci resource template searched for the
oracletypeattribute instead of
typewhen processing the
/etc/cluster/cluster.conffile. Consequently, the oracledb attribute was always displayed as
Defaultin the luci interface, regardless of its actual assigned value. This bug has been fixed, and
oracletypetype is now correctly displayed by luci.
- With this update, support for the IBM iPDU fence device has been added to the luci application.
- With this update, a new user table has been added to the luci interface. It is now possible to remove users from luci.and pages of the
- With this update, support for configuring the
privlvl(privilege level) attribute used by the fence_ipmilan fencing agent has been added to the luci application. As a result,
privlvlcan now be successfully configured by luci.
- With this update, support for the
nfsrestartoption for the file system and cluster file system resource agents has been added to the luci application. This option provides a way to forcefully restart NFS servers and allow a clean unmount of an exported file system.
- This update adds the fence_eaton agent to support Eaton ePDU (Enclosure Power Distribution Unit) devices in Red Hat Enterprise Linux 6, into the luci package.
- With this update, an interface for configuring and displaying the fence_hpblade fence devices has been added to the luci application.
- Previously, luci did not include concurrent additions to parameters for some fence devices (including "cmd_prompt", "login_timeout", "power_timeout", "retry_on", "shell_timeout") or respective instances ("delay") as happened in the fence-agents package. Consequently, the valid parameters could be dropped from the respective part of the configuration upon submitting the dedicated forms in luci. This update restores the capability of luci to work with a full intended set of fence agents parameters and, in turn, prevents luci from unexpectedly discarding the already configured ones.
- Previously, luci did not include concurrent additions to fence devices coverage as happened in the fence-agents package. Consequently, Dell iDRAC (idrac), HP iLO2 (ilo2), HP iLO3 (ilo3), and IBM Integrated Management Module (imm) devices or agents were not honored in luci, leading to an inability to properly work with or to setup a cluster comprising them. This update restores the capability of luci to work with a full intended set of fence devices.