pacemaker neutron resource constraints and cleanup
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
tripleo |
Fix Released
|
High
|
Marios Andreou |
Bug Description
As discussed in [1] the pacemaker resource constraints chain defined in the overcloud_
Given the current constraints chain, a "pcs resource restart neutron-server" may leave the neutron services in erroneous states after restart and any router interfaces without IP (not hosted anywhere). In the bug report at [1] the reproducer is to restart haproxy (which results in restart of all openstack services, including neutron-*).
Discussion is ongoing as to the solution, though I will link to a review momentarily which tries to implement a different constraints chain, as suggested by ajo @ [3] (will be WIP for now)
[1] https:/
[2] https:/
[3] https:/
Changed in tripleo: | |
importance: | Undecided → High |
status: | Triaged → In Progress |
assignee: | nobody → Marios Andreou (marios-b) |
Changed in tripleo: | |
status: | In Progress → Fix Released |
WIP review for discussion at https:/ /review. openstack. org/#/c/ 229466/