Ceilometer Mongo DB backend cannot failover
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
Fuel for OpenStack |
Fix Released
|
High
|
Igor Degtiarov | ||
5.0.x |
Fix Committed
|
High
|
Unassigned | ||
5.1.x |
Fix Committed
|
High
|
Unassigned | ||
6.0.x |
Fix Committed
|
High
|
Unassigned |
Bug Description
{"build_id": "2014-09-
Steps to reproduce:
1. Ubuntu with nova
3 controllers with mongo + ceiloemeter
1 compute
2. Deploy
3. Run ostf
4. ssh on primary controller and drop traffic to the mongo port
iptables -I OUTPUT 1 -p tcp --dport 27017 -j DROP
iptables -I INPUT 1 -p tcp --dport 27017 -j DROP
5. Run ostf
6. ssh on other controller and request alarm-list
Actual result:
OSTF tests are failed. alarm list failed. Seems we do not monitor mongo at all
Expected:
mongo runs under crm - so in case of failover - user data will be available
INFO: the same result if I kill mongo pid , but it works fine in case if node was shut down
Changed in fuel: | |
status: | New → Confirmed |
Changed in fuel: | |
assignee: | Fuel Library Team (fuel-library) → Sergii Golovatiuk (sgolovatiuk) |
summary: |
- Ceilometer randomply fails in ha if mongo was blocked on primary + Ceilometer randomly fails in HA if mongo was blocked on primary controller |
Changed in fuel: | |
assignee: | Sergii Golovatiuk (sgolovatiuk) → nobody |
assignee: | nobody → Fuel Library Team (fuel-library) |
Changed in fuel: | |
assignee: | Fuel Library Team (fuel-library) → Igor Degtiarov (idegtiarov) |
This needs to be converted to a blueprint for next release.