7.212. resource-agents

Bug Fixes

BZ#714156

Previously, the status action in the netfs interface failed to write any output to the /var/log/cluster/rgmanager.log file. Consequently, it was not possible to verify if the status check of an NFS mount was successful. The bug has been fixed, and results of the status check are now properly stored in the log file.

BZ#728365

For HA-LVM to work properly, the /boot/initrd.img file, which is used during the boot process, must be synchronized with the /etc/lvm/lvm.conf file. Previously, the HA-LVM startup failed when lvm.conf was changed without updating initrd.img. With this update, this behavior has been modified. A warning message is now displayed, but the startup is no longer terminated in the described case.

BZ#729812

Prior to this update, occasional service failures occurred when starting the clvmd variant of the HA-LVM service on multiple nodes in a cluster at the same time. The start of an HA-LVM resource coincided with another node initializing that same HA-LVM resource. With this update, a patch has been introduced to synchronize the initialization of both resources. As a result, services no longer fail due to the simultaneous initialization.

BZ#817550

When the oracledb.sh script was called with the status argument, it restarted the database after checking its status without any notification to the rgmanager application. This bug has been fixed, and the unwanted restart no longer occurs.

BZ#822244

Previously, the /usr/sbin/tomcat-6.sh script parsed configuration files and set shell variables before starting the Apache Tomcat 6 servlet container. Consequently, the default configuration was ignored. This bug has been fixed and the aforementioned problem no longer occurs.

BZ#839181

Previously, an output of HA-LVM commands that contained more than one word, was not correctly parsed. Consequently, starting an HA-LVM service with the rg_test command occasionally failed with the following message:

too many arguments

With this update, the underlying source code has been modified to add quotation marks around variables that expand to more than one word. As a result, the aforementioned startup errors no longer occur.

BZ#847335

If the contents of the /proc/mounts file changed during a status check operation of the file system resource agent, the status check could incorrectly detect a missing mount and mark the service as failed. This bug has been fixed and rgmanager's file system resource agent no longer reports false failures in the described scenario.

BZ#848642

Previously, rgmanager did not recognize CIFS (Common Internet File System) mounts in case their corresponding entries in the device field of the /proc/mounts file contained trailing slashes. With this update, a patch has been introduced to remove trailing slashes from device names when reading the contents of /proc/mounts. As a result, CIFS mounts are now recognized properly.

BZ#853249

Prior to this update, when running a file system depending on an LVM resource in a service, and that LVM resource failed to start, the subsequent attempt to unregister the file system resource failed. This bug has been fixed, and a file system resource can now be successfully unregistered after a failed mount operation.

BZ#860328

Previously, when using the HA-LVM resource agent in the Pacemaker cluster environment, several errors and failed actions occurred. With this update, several scripts have been added to prevent these errors. These scripts repair the treatment of whitespace within HA-LVM and the processing of non-zero codes in rgmanager. In addition, the member_util utility has been updated to use Corosync and Pacemaker when rgmanager is not present on the system.

BZ#860981

Previously, when a node lost access to the storage device, HA-LVM was unable to deactivate the volume group for the services running in that node. The underlying source code has been modified to allow services to migrate to other machines that still have access to storage devices, thus preventing this bug.

BZ#869695

Previously, SAP instances started by the SAPInstance cluster resource agent inherited limits on system resources for the root user. Higher limits were needed on the maximum number of open files (ulimit -n), the maximum stack size (ulimit -s), and the maximum size of data segments (ulimit -d). With this update, the SAPInstance agent has been modified to accept limits specified in the /usr/sap/services file. As a result, system resources limits can now be specified manually.

Enhancements

BZ#773478

With this update, the /usr/share/cluster/script.sh resource, used mainly by the rgmanager application, has been enhanced to provide more informative reports on causes of internal errors.

BZ#822053

With this update, the nfsrestart option has been added to both the fs and clusterfs resource agents. This option provides a way to forcefully restart NFS servers and allow a clean unmount of an exported file system.

BZ#834293

The pacemaker SAPInstance and SAPDatabase resource agents have been updated with the latest upstream patches.

BZ#843049

A new prefer_interface parameter has been added to the rgmanager ip.sh resource agent. This parameter is used for adding an IP address to a particular network interface when a cluster node has multiple active interfaces with IP addresses on the same subnetwork.

Enhancement

BZ#1001519

This update adds support for the Pacemaker resource agents under the Heartbeat OCF provider.

Bug Fix

BZ#978775

Usage of lvm.sh with tags resulted in the stripping of cluster tags when the node rejoined the cluster. This was because the lvm.sh agent was unable to accurately detect the tag represented by a cluster node. Thus, the active logical volume on a cluster node failed when another node re-joined the cluster. This update properly detects whether tags represent a cluster node, the node-name, or, if fqdn is returned by the corosync-quorumtool -l output. When nodes re-join the cluster, tags are not stripped of LVM volume groups, and the volume group no longer fails on other nodes.

Enhancement

BZ#972931

Previous versions of the Oracle Resource Agent were only tested against Oracle 10. With this update, support for the Oracle Database 11g has been added to the oracledb, orainstance, and oralistener resource agents.