Release Notes #

Previously, ldirectord did not support HTTP checks for IPv6.

This issue was fixed by adding the packages perl-Net-INET6Glue and perl-IO-Socket-IP as dependencies to ldirectord.

Some commands were undocumented or did not cover all use cases.

The interactive help has been overhauled, and the online help for the crm shell has been updated and improved.

An unlink would generate unnecessary journalling requests. This would hinder any write operations that would follow.

The unlink path has been changed to avoid the unnecessary journalling.

Support fencing via SCSI reservations.

The fence-agents package is now included which includes an agent capable of fencing via SCSI reservations.

Haproxy is useful in some scenarios where LVS cannot be used.

Haproxy 1.5 is now provided.

The current version of Relax-and-Recover (abbreviated as 'rear', formerly 'ReaR') comes with many new features and fundamental changes. In particular the file system creation is totally different (via the so called "layout workflow" instead of the old "dr workflow" that has been removed in version 1.14).

Warning: Users who already use it and want to upgrade (in particular when upgrading from SUSE Linux Enterprise 11), must carefully re-validate whether their particular disaster recovery procedure still works.

A new and different RPM package name is used that contains the version (e.g. "rear116"). It conflicts with all other rear packages to avoid that an installed version could get accidentally replaced with another version. This means "just a normal package installation" that would do a version change of an already installed Relax-and Recover software is intentionally prohibited.

Furthermore this version upgrade supersedes the formerly provided package rear-SUSE.

MariaDB is an alternative to MySQL, and the resource agent for HA should support both.

The resource agent now supports MariaDB.

A user with limited ACL access may not have the correct permissions to apply the configuration created by a wizard. Hawk should detect this and warn before allowing the wizard to be applied.

Hawk warns if the user lacks necessary write permissions to apply a configuration wizard.

Ensuring that all nodes in the cluster have the correct packages and file system configuration can be difficult. Sometimes, a configuration wizard could be improved if it could make system-level changes or inspect the state of all cluster nodes before applying a new CIB configuration.

The crm shell now includes cluster scripts, which are able to validate and modify the configuration of cluster nodes.

The crm shell is an excellent tool for configuring pacemaker, but the cluster stack consists of additional software. It should be possible to leverage the abilities of crmsh for configuring these as well.

crmsh can now configure the cluster infrastructure directly, via the new cluster and corosync subcommands.

A cluster report generated by crm_report contains much of the information needed to debug issues with a cluster. It should be possible to open these reports in the history explorer and crmsh.

Both hawk and crmsh can now open reports generated by crm_report. However, reports generated by hb_report contain additional information and are preferred when possible.

Some commands were undocumented or did not cover all use cases.

The interactive help has been overhauled, and the online help for the crm shell has been updated and improved.

The configuration of OCFS2 has been simplified, and the Hawk configuration wizard as well as the configuration template in crm shell should reflect this.

The Hawk wizard for cLVM2/OCFS2 has been updated.

To make the configuration of an NFS server easier, Hawk should include a configuration wizard for this purpose.

Hawk includes a configuration wizard that configures an NFS server.

Enabling the use of resource sets and templates in location constraints makes configuration easier and can simplify the CIB.

Both crmsh and hawk now supports creating location constraints that reference resource sets. The syntax for crmsh is described in the Admin Guide.

It may be desired to start and stop multiple resources all at once, without having explicit dependencies between those resources.

This feature adds resource tags to Pacemaker. Tags are collections of resources that do not imply any colocation or ordering constraints, but can be referenced in constraints or when starting and stopping resources.

When scripting cluster configuration changes, the scripts are more robust if changes are applied with a single commit. Creating a shadow CIB to collect the changes makes this easier, but has previously required naming the shadow CIB.

crmsh now allows the creation of shadow CIBs without explicitly specifying a name. The name will be determined automatically and will not clash with any other previously created shadow CIBs.

Allowing the use of plain IP addresses can make it easier to configure csync2.

csync2 allows the use of IP addresses for connectivity via a special syntax. This is documented in Procedure 3.8: Configuring Csync2 with YaST in the Admin Guide.

Sometimes, it is desired that two resources should run on the same node. However, there should be no further dependency implied between the two resources, so that if one fails, the other one can keep running.

This can be accomplished using a third, dummy resource which both resources depend on in turn. To make it easier to create a configuration like this, the command "assist weak-bond" has been added to crmsh.

Simplify wizards with multiple identical config instances.

In a wizard workflow, multiple steps to reference the same wizard template with different sets of parameters are supported.

This is done by adding a type attribute to the template instance in the workflow, which can be separate from the name identifying the step in the workflow. Each step in the workflow can also override any attribute of the wizard template, including the description.

This change is backwards compatible.

Sometimes, a system adminstrator may want to use the Hawk history explorer to examine a cluster report on a machine that is not part of the cluster.

This feature allows hawk to be installed and run without requiring pacemaker to be running, and the history explorer can function as expected.

Configuring a cluster node correctly is complicated, and problems with hardware, network connectivity and system software can cause issues when configuring cluster resources.

The crm shell now includes a cluster health script, which can run both on an already-configured cluster node as well as a potential cluster node, to perform various system-level health checks. It then presents a list of potential issues, such as not having NTP configured correctly or warning about low disk space.