Updated resource-agents packages that fix several bugs and add various enhancements are now available for Red Hat Enterprise Linux 6.
The resource-agents packages provide the Pacemaker and RGManager service managers with a set of scripts that interface with several services in order to allow operating in a High Availability (HA) environment.
- The lvm.sh agent was unable to accurately detect a tag represented by a cluster node. Consequently, the active logical volume on a cluster node failed when another node rejoined the cluster. Now, lvm.sh properly detects whether tags represent a cluster node. When nodes rejoin the cluster, the volume group no longer fails on other nodes.
- If the file system used by a MySQL resource became unavailable, the MySQL agent's validation checks prevented the resource from stopping. This bug has been fixed, and MySQL resources are now properly restarted in the described case.
- The RGManager resource agent failed to recognize that Oracle Database started successfully when notifications about non-critical errors were printed on startup. This update modifies the behavior of RGManager to ignore the non-critical errors, so that the Oracle Database service does not fail in this situation.
- Floating IPv6 addresses managed by the RGManager ip.sh agent did not send unsolicited advertisement packets when starting. Consequently, when an IP resource failed over, it took about five minutes for the tables to be updated. The packets are now sent, which optimizes the time required before an IP address is recognized as being available.
- When a node experiences a loss of quorum, the RGManager utility performs an emergency stop of all resources, not just those that are in a started state. Previously, when a separate node split from the cluster and lost quorum, the vg_stop_single() function stripped the Logical Volume Manager (LVM) tags from the Volume Group (VG) if the vg_owner was set. With this update, the LVM agent strips the tags only when the local node performing the stop operation is the owner, and the service now runs as part of the quorate partition even if the service owner's LVM tags have been removed.
- Due to a regression, some NFS options went missing in the nfsserver after updating, and it was impossible to modify the number of the NFS thread. A patch has been applied, and the number is now modifiable.
- When monitoring a cluster network interface, the IPaddr2 agent could display an "ERROR: [findif] failed" message even though the IP address and interface were working properly. This update fixes the underlying code, and the IPaddr2 agent consistently reports accurate results during the monitor operation.
- The MySQL agent failed to work if configured with a user other than 'mysql'. Consequently, MySQL failed to start due to a permission error manifested as a timeout error. A fix has been applied, and MySQL now starts and runs as the configured user.
- Under certain circumstances, the write test of the is_alive() function did not properly detect and report when a file system failed and was remounted as read-only. This update fixes the bug and in the described scenario, is_alive() now reports the status of the file system correctly.
- The Pacemaker nfsserver agent now sets the rpc.statd TCPPORT or UDPPORT via configuration options.
- The nginx resource agent now allows an nginx web server to be managed as a Pacemaker cluster resource. This provides the ability to deploy the nginx web server in a high availability environment.
- The resource-agents-sap-hana package now provides two Pacemaker resource agents, SAPHanaTopology and SAPHana. These resource agents allow configuration of a Pacemaker cluster to manage a SAP HANA Scale-Up System Replication environment on Red Hat Enterprise Linux.
Users of resource-agents are advised to upgrade to these updated packages, which fix these bugs and add these enhancements.