Network Management
Troubleshooting Parent-Child dependencies in the SolarWinds Platform
Parent-Child dependency
First published date
Last published date
Overview
Product section
Resolution
Working with alert dependencies can be sometimes be difficult.
The one thing we need to make sure is that you do not have a Node down alert that includes those nodes that are also covered in dependency alert, because it doesn't matter if you have a dependency alert if you also have a node down alert, for all nodes the alert will trigger.
Here are some articles of alerts that talks about when the parent is down, the child will be in an unreachable state and alerts on the child will be suppressed:
- Create a dependency between network objects
- Create dependencies between network objects automatically
- View active alerts on child objects when the parent object is down
Troubleshooting - Things to double-check:
- Make sure the child does not have multiple parents.
- If it is a dependency of group then set rollup mode to "Show best status"
- Verify that "Enable Auto Dependencies", in Polling Settings, is CHECKED
Dependency Logic:
Here is how the dependency logic works:
- A child object can have several parents,
- The child's status only turns to Unreachable if all parents in all defined dependencies go down.
Example:
- A Node becomes unresponsive
- The node goes to warning
- After 120 sec the system tries to move the node from Warning to Down
- before that the system makes calculation:
- Is the NodeID a member of any Containers
- IF YES
- Check to see if that ContainerID is a Child for any dependencies
- IF NOT
- Is the NodeID a Child for any dependencies?
- IF NOT
- set it to DOWN
- IF YES
- CHECK Parent status on ALL Dependencies that the Node is a child for
- IF Not all Parents are down
- Set Node status Down
- IF ALL Parents are Down
- make node status Unreachable
- IF Not all Parents are down
- CHECK Parent status on ALL Dependencies that the Node is a child for
- IF NOT
- Is the NodeID a Child for any dependencies?
- before that the system makes calculation:
Take a look at this article and select "When the parent goes down" option:
The status of dependent child objects are UNREACHABLE instead of DOWN or UNKNOWN. Any alerts where the trigger condition includes Node status is equal to DOWN are not triggered.