[Neutron] neutron.db.agentschedulers_db ERROR in logs after ban 2 DHCP and 2 L3 agents
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
Mirantis OpenStack |
Won't Fix
|
Low
|
MOS Neutron | ||
10.0.x |
Won't Fix
|
Low
|
MOS Neutron | ||
9.x |
Won't Fix
|
Low
|
MOS Neutron |
Bug Description
MOS 9.2 snapshot #655
During running the test for verification https:/
http://
2 configurations have been checked:
1) 3 controllers, 2 computes, 1 cinder node, Neutron VLAN
2) 3 controllers+mongo, 2 computes+cinder, Neutron VLAN DVR, Ceilometer, Sahara
Steps to reproduce:
1. Update quotas for creation a lot of networks
2. Create 50 networks, subnets, launch and terminate instanсe
3. Ban two dhcp agents related to the first network
4. Ban two l3 agents
5. Check /var/log/
Expected results:
Only Quota errors exist in neutron server logs (as neutron can't create more than about 34 networks).
Actual error:
neutron.
The following script can be used for reproducing this situation:
http://
It was notices that the bug can be reproduced only after creating networks more than quota value.
After creation only OverLimit logs exist, but after banning agents db log also appears on ONE controller.
Changed in mos: | |
milestone: | none → 9.2 |
assignee: | nobody → MOS Neutron (mos-neutron) |
description: | updated |
Changed in mos: | |
status: | New → Confirmed |
Changed in mos: | |
status: | Confirmed → Won't Fix |
I believe this is a race between neutron servers to delete agent-to-net binding. Should not affect functionality, only traces in logs.