Service Impact 5.3.1
This release features the enhancements that were introduced in version 5.3.0:
- Faster graph updates and state propagation in environments with many nodes.
- A new right-click menu option enables you to center the Impact View on a node that you select. The graph shifts to display the selected node in the center, but does not change the zoom level. The new Selected Node field enables you to select from a list of nodes that match the partial text string that you specify. Enhanced filtering enables you to find and view nodes by availability and performance states and partial name string.
The following table identifies the files to download for each supported installation or upgrade scenario.
| Resource Manager | Required files |
|---|---|
6.x 5.3.x 5.2.x |
ZenPacks.zenoss.ImpactServer-5.3.1.0.0-py2.7.egg |
ZenPacks.zenoss.Impact-5.3.1.0.0-py2.7.egg |
|
install-zenoss-impact_5.3_5.3.1.0.0.run |
Download files from delivery.zenoss.io. Use your Zenoss Support credentials to log in.
Compatibility
For version information, please see the compatibility matrix on the Release Notes page.
Fixed issues
| ID | Description |
|---|---|
| IMP-39 | GraphML file that was exported from an older version of Service Impact cannot be imported into latest version. |
| IMP-284 | Implicit bug in setting event state during propagation process. |
| IMP-285 | Improve performance and responsiveness under high workload environments. |
| IMP-290 | Do batching of production state updates in Service Impact server. |
| IMP-307 | In a large database, cannot remove nonexistent nodes because requests time out. |
| IMP-309 | Impact state propagation takes too long on devices with thousands of components. |
| IMP-316 | Multithreaded implementation of related event processing. |
| IMP-320 | Unable to delete top level logical node organizer when it contains children. |
| IMP-321 | A state node accumulates best state events. |
| IMP-322 | Import of a GraphML file associates incorrect devices to components. |
| IMP-323 | Need to repair Neo4j objects with large number of non-relevant events. |
| IMP-349 | Refactor handling Neo4j resources in multi-threaded environment. |
| IMP-351 | When updating a graph, errors occurr for some ZenPack objects. |
| IMP-357 | State propagation process takes an unacceptable amount of time on devices with thousands of components. |
| IMP-369 | graphreset (update) does not prune orphaned nodes. |
Considerations and workarounds
Reidentifying devices
Resource Manager provides the ability to reidentify devices. The reidentification process deletes and re-adds the device with a new ID and GUID. When the GUID is changed, the device is removed from that particular Service Impact service. If you reidentify a device, you must manually re-add the device to the service model.
Impact graph update best practices
Graph updates and maintenance window operations use the same target graph database. Concurrent access of shared data increases time required to complete the operations, risk of transaction rollbacks and deadlocks, and risk of inconsistent state updates to the graph database. For best performance when updating an impact graph,
- Do not manually rebuild an impact graph during a time that overlaps a maintenance window or another graph update operation.
- Do not import dynamic service models while a graph update operation is in progress.
Upgrade considerations
When you initially upgrade from Service Impact 5.2.3 to 5.3.1, the new version of the product removes accumulated event history that is no longer needed. A temporary performance slowdown might occur while nodes are read.
Limitations
After Service Impact is installed, Service Impact and Resource Manager are interdependent. However, Resource Manager issues affect Service Impact more frequently than Service Impact issues affect Resource Manager. For this reason, the list of known issues for a given Service Impact release might include items that manifest in Service Impact but are not caused by Service Impact software. Such items are noted in the list of known issues.