Operations Dashboard in cluster
Assignee
Pavel SalamonReporter
Jaroslav UrbanLabels
Sprint
Description
Steps to reproduce
Activity
Jiri Trnka October 1, 2020 at 10:23 AMEdited
Tested cases mentioned in description on sources and release build 5.8.0.90.
Tested on cluster with SSL configuration.
Its also possible to use direct REST API on one node and the change is displayed on the other node after refresh period passes.
Tested change on one node after import of configuration from xml on the second node.
Pavel Salamon September 18, 2020 at 12:52 PM
Remaining work:
manual QA?
final completeness check?
Pavel Salamon September 18, 2020 at 12:47 PMEdited
A cluster problem was detected and reported:
Scenarios for other automated tests have been added here:
Jaroslav Urban September 3, 2020 at 8:51 AMEdited
Grooming:
if 2 users modify the same monitor - last one wins
self references should still work - probably needs some work -
Jaroslav Urban September 1, 2020 at 12:16 PMEdited
This should already work - this story is mostly describing testing use-cases.
We could extend automated tests of the API to randomly use different nodes for each call? It could bring instabilities?
The Operations Dashboard should work in Cluster:
same dashboard on all nodes (i.e. same view on each node where I can login)
information used by Monitors are calculated based on state of the whole cluster (e.g. listener failing on any node can influence a Monitor state)
changes to the dashboard are propagated to the whole cluster (e.g. I add a monitor on node1, then soon users on node2 will also see it)
dashboard API works on all nodes and provides a cluster-wide view
if 2 entities change state on 2 nodes at the same time, monitor that uses it works correctly
example - on 1 node schedule fails, on 2nd node listener fails, then the one monitor that uses them will have correct resulting state