Browse Source

Set octavia services' stop grace period to 300sec

Octavia worker, house-keeping and health-monitor serivices may use some
long taskflow's flows to handle load balancers and amphorae (launch VMs,
etc...). Those flows should not be interrupted when restarting those
services (i.e when updating an overcloud, or restarting services because
of certificates rotation), it might cause resource leaks that cannot be
fixed by an admin.

As default container stop timeout is defined to 10 seconds, this timeout
value needs to be increased for octavia services (except octavia api) to
ensure a graceful shutdown.
This new value has been set to 300 seconds according to the octavia
worker default configuration introduced in
https://review.opendev.org/#/c/684201/

Closes-Bug: #1855684
Change-Id: I8911a79328769c910d03168cfa5a421d0dd0f9b6
(cherry picked from commit c595835776)
changes/38/703938/1
Gregory Thiemonge 2 months ago
parent
commit
0119358280
3 changed files with 3 additions and 0 deletions
  1. +1
    -0
      deployment/octavia/octavia-health-manager-container-puppet.yaml
  2. +1
    -0
      deployment/octavia/octavia-housekeeping-container-puppet.yaml
  3. +1
    -0
      deployment/octavia/octavia-worker-container-puppet.yaml

+ 1
- 0
deployment/octavia/octavia-health-manager-container-puppet.yaml View File

@@ -142,6 +142,7 @@ outputs:
step_4:
octavia_health_manager:
start_order: 2
stop_grace_period: 300
image: *octavia_health_manager_image
net: host
privileged: false

+ 1
- 0
deployment/octavia/octavia-housekeeping-container-puppet.yaml View File

@@ -133,6 +133,7 @@ outputs:
step_4:
octavia_housekeeping:
start_order: 2
stop_grace_period: 300
image: *octavia_housekeeping_image
net: host
privileged: false

+ 1
- 0
deployment/octavia/octavia-worker-container-puppet.yaml View File

@@ -125,6 +125,7 @@ outputs:
step_4:
octavia_worker:
start_order: 2
stop_grace_period: 300
image: *octavia_worker_image
net: host
privileged: false

Loading…
Cancel
Save