Old chassis must be manually removed from OVN southbound DB
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
networking-ovn |
Confirmed
|
Wishlist
|
Unassigned |
Bug Description
I follow the instructions in this link http://
At first, I install single node (9.70.33.149, hostname ovn1) successful, and I can see this node by command "ovn-sbctl show"
Then I add another compute node (9.70.22.148, hostname ovn2), i also can see this compute node by command "ovn-sbctl show", but i find in this compute node, the nova-compute can not run, so i remove this compute node and add another compute node, the ip is also 9.70.33.148.
Then in my setup, there are 2 9.70.33.148 in ovn db, but only 1 ovn2 in nova db.
why the old compute node can not be deleted by timeout mechanism because it disappears.
steve@ovn1:
Chassis "a19a83dd-
Encap geneve
ip: "9.70.33.149"
Chassis "22af05a5-
Encap geneve
ip: "9.70.33.148"
Chassis "5c1a106e-
Encap geneve
ip: "9.70.33.148"
steve@ovn1:
+----+-
| ID | Hypervisor hostname | State | Status |
+----+-
| 3 | ovn1 | up | enabled |
| 4 | ovn2 | up | enabled |
+----+-
steve@ovn1:
082bec78-
Bridge br-int
fail_mode: secure
Port "tap6f9f462a-56"
Port "ovn-22af05-0"
Port "ovn-5c1a10-0"
Port br-int
Changed in networking-ovn: | |
importance: | Undecided → Low |
Changed in networking-ovn: | |
status: | New → Incomplete |
Changed in networking-ovn: | |
importance: | Low → Wishlist |
tags: | added: ovn-upstream |
summary: |
- The new chassis can not replace old chassis + Old chassis must be manually removed from OVN southbound DB |
Changed in networking-ovn: | |
status: | Incomplete → Confirmed |
It's up to the administrator to delete an old chassis from the db right now. You can do it with ovn-sbctl.
# ovn-sbctl chassis-del CHASSIS
where CHASSIS is either the chassis name or UUID.