Service Impact 5.4.5 CA
This is a controlled-availability release. If you are interested in updating to this release, please contact your Zenoss representative.
This release introduces a major new feature, selective synchronization.
In previous releases, the Resource Manager Zope Object Database (ZODB) and the Service Impact graph database were tightly synchronized, and all of the relationships among Resource Manager devices and components in your environment (the "model") were copied to the Service Impact graph database. In larger environments, when Resource Manager model updates were underway, the synchronization process reduced performance in Resource Manager and caused Service Impact to make analyses with incomplete data.
In this release, you can refine the list of components that are copied from ZODB to Service Impact. This approach works well in environments where the critical services defined in Service Impact are a relatively small part of the Resource Manager model.
Additional highlights of this release:
- The SERVICES > Metatype Configuration page is the new interface for deciding which device components to exclude from Service Impact. The page includes counts of the overall number of entities in the Resource Manager model and the Service Impact model, updated regularly.
- The export/import/reconciliation process has been re-implemented to improve accuracy, reduce reconciliation conflicts, and provide progress information.
- The zenimpactworker process is no longer necessary.
For version information, please see the compatibility matrix on the Release Notes page.
In previous releases, adding a dynamic service was very quick, because the Service Impact database included all of the devices and components in the ZODB. In this release, Service Impact creates a background job to copy the devices and components it requires, which can take a few minutes to complete. Scripts that utilize the Service Impact JSON API must be updated to wait for the job to complete before proceeding.
When you change the metatype configuration, the Service Impact database is updated to add or remove nodes. Depending on which items are changed, processing the changes can take 30-90 minutes in very large environments.
(IMP-306) The list of allowed characters in the names of dynamic services and dynamic service organizers is constrained. For more information, see Dynamic service name constraints.
(IMP-685) During the
zenimpactgraph run --updateprocess, you may receive a message similar to the following:
WARNING zen.ImpactGraph: Exception while adding impact batch:'NoneType' object has no attribute '__module__', number tried: 9 (a retry number > 5)
zenimpactgraph run -x catcleanand retry the
zenimpactgraph run --updateprocess again.
|The criteria for logical nodes must consider the numbering system of source events
|The health check for webservice-status fails, which causes additional failures
|Special characters not supported in service or organizer names
|(IMP-557) Reconciliation does not consider relationships in imported graph
|Null pointer exception in Neo4j before state change events stop processing
|The zenimpactworker service times out during long graph updates
|zenimpactgraph loses connection to zodb after a long graph update
|Both device and component production state are sent to the impact server
|Exports do not include logical nodes that are not part of a dynamic service
|Inconsistent production states when graph updates are run during a maintenance window
|No retry when certain add node operations fail
|Rollback exception not handled properly
|updateGraph API cannot support large number of interlocking changes in one transaction
|Upgrading the LinuxMonitor ZenPack fails due to large number of model changes at once
|Strange behavior when following dynamic service member link
|Can't add a service or organizer because Submit button is inactive
|The Impact view does not support the dark theme in the Resource Manager browser interface
|In some cases, Impact events are not shown on a service's Events view
|The Impact View does not immediately show updates to the model
|A warning message is displayed constantly during normal use
|The upgrade.sh script is not working properly
|Impact does not provide accurate data to the Zenoss call home feature
|Loading search results is slow on larger devices with many components
|Value for check-box in Suppress service events is not persistent
|Changing a service name does not propagate everywhere
|Error when attempting to save a policy with empty fields
|Policies are not included in service export/import
|Import process does not include progress information
|Production state changes during maintenance windows are not reflected in Impact
|Events that match logical node criteria aren't processed
|Font color in Impact portlet makes text difficult to read
|Removing Impact components takes many seconds with no UI notification
|Indexing is slow
|Unable to click an object in the Impact View and go to the device in Resource Manager
|Graph update takes a long time on a large system
Could not adapt during graph update
|ZenPack update fails due to Impact state changes
|Restoring omitted components is not working
|Unable to create file for edges due to missing parent directory