[Contrail] Change using DB for analytics to a dedicated node on the deployed cluster leads to fail
Bug #1625552 reported by
okosse
This bug affects 1 person
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
Fuel Contrail plugin |
Fix Released
|
Medium
|
Illia Polliul |
Bug Description
Steps to reproduce:
1. Deploy cluster without Dedicated Analytics DB
2. Enable Dedicated Analytics DB and add new node with Contrail-
3. Redeploy the environment
Expected results: All steps are success
Actual result: Analytics API Down. Alarms may not be reported correctly.
Changed in fuel-plugin-contrail: | |
status: | New → Fix Released |
assignee: | Partner Centric Engineering (fuel-partner-engineering) → Illia Polliul (ipolliul) |
To post a comment you must log in.
I don't see a reasonable way to change it automatically back and forth.
If you switch from non-split db to split one - plugin will prepare additional Cassandra cluster and change configs. But the but DB schema is created during Contrail-Analytics provisioning.
So with this switch - you'll need to de-provision Contrail-Analytics and provision it once again with new parameters.
Therefore this action should be documented as semi-automatical, based on this document:
https:/ /github. com/Juniper/ contrail- fabric- utils/wiki/ Provisioning- separate- analytics- DB-nodes- in-Existing- cluster- (Contrail- Cloud)