IBM DS8xxx FlashCopy driver
The driver for IBM DS8xxx arrays FlashCopy snapshotting is merged. The OpenSVC agent now can drive snapshot lifecycle on behalf of services on HP EVA, DataCore, IBM DS8xxx and EMC Symmetrix arrays. A service snapshot refresh scheduler batch is as simple as 'service stop' -> 'service syncresync' -> 'service start'.
Bookmarking in collector views
Collector views now provide a facility to bookmark the current set of column filters. Fast recall of a filter-intensive analysis is a major productivity gain provided by this feature. Creating bookmarks also eases the drill-down -> back up (to bookmark now) -> drill-down common usage pattern.
The agent now sends to the collector rpm signature information. The packages view in the collector now allows to find easily on your whole datacenter where the rpm come from, and to track and alert on installations from untrusted locations.
New array parsers
Parsers for HP 3par and IBM DS8xxx arrays are merged in the collector, and their respective data extractors are merged in the agent. OpenSVC can now aggregate and report SAN disk information from most arrays from EMC, IBM, HP and NEC.
A forms editor in the collector
A forms editor has landed in the collector. It's primary usage is to split the compliance objects forms from the collector code base. As a result, compliance objects and their associated forms can be created by clients easily and without OpenSVC team involvement. The powerful forms description langage is also leveraged for other uses. One is a form-based service and node customization system, which from a form inputs generates a compliance rule automatically attached to a specific service or node. For example this system can be used to present a user with a weblogic deployment form, which from a bunch of question (port, username, ...) configures the compliance rules and ends up deploying the product. Forms inputs can also be fed to scripts, or be mailed to interested parties. A natural evolution for this forms subsystem is generic workflow management. Stay tuned.
OpenSVC now can handle NFS shares as service resources on HP-UX and Linux. This approach is more friendly than the previously suggested NFS app launchers. The users benefit from fine-grained error reporting and self reparation. For example, when a share is not exported to a specific target client, or exported with unexpected options, a warning is raised explaining the issue, and starting the share resource will re-export only for the affected client. The driver has been tested with hundreds of shares and proved to be fast and reliable.
Now OpenSVC services handling at least one container resource can span inside the container. If so desired, the agent must be installed in the container, and at least one service resource must be tagged as 'encap'. In this case, the master service (the part handled by the hypervisor agent) takes care of synchronizing the service configuration into the container. The service configuration file is the same. The slave agent only handles the resources flagged 'encap'. This new feature permits powerful topologies like Oracle RAC cluster in zones, or failover slave service between active containers in a master flex. The nice part is that admins logged into the container can operate on the slave service as they would on any service. The master and slaves resource status are aggregated by the master service and reported to the collector from there.
OpenSVC services were limited to handle only one container per physical host, possibly all running if in flex mode. This limitation is no more. Now containers are proper resources in the service configuration file, each with a dedicated [container#n] section. So, beware of these configuration file changes while planing agent upgrades. Please refer to usr/share/doc/template.env file for details of new configuration parameters.
OpenSVC support for Clouds
OpenSVC services can now handle cloud resources : virtual machines on OpenStack and Vcloud, and cloud disk volumes for Gandi cloud. This support enables many interesting service topologies : physical cluster in production with disaster recovery in a cloud, shared disk clustering between two cloud virtual machines, disaster recovery on a different cloud from the production, or a farm of virtual machines spanned across different clouds. These new drivers use Apache libcloud cloud login abstraction. They also support virtual machine provisioning features of the clouds API.
Veritas Cluster wrapper
The agent OpenSVC is now capable of reporting and inventoring VCS service groups. This new wrapper complements the already existing ServiceGuard and Red Hat Cluster wrapper family. The target is to be able to describe coherently whole heterogeneous datacenters through the collector. The compliance rulesets and modulesets can be attached to such wrapped services, their resources accounted.