SNMP can be installed in worker nodes when it shall run only in controller nodes

Bug #1955571 reported by Enzo Candotti
6
This bug affects 1 person
Affects Status Importance Assigned to Milestone
StarlingX
Fix Released
Low
Enzo Candotti

Bug Description

Brief Description

SNMP is working fine while running on controller nodes but it can also be installed on worker nodes when it shall not be possible to install in nodes different than controller.

Just to mention that once snmp is installed on worker node, snmp get/walk operations work fine but trap generation is not working.

The SNMP should not be installed on worker nodes.
-----------------
Severity

 Minor
-----------------
Steps to Reproduce

1. In STD environment install snmp app
2. Check snmp installation

-----------------
Expected Behavior

SNMP should always be deployed in controllers node
-----------------

Actual Behavior

SNMP is randomly deployed in worker nodes
-----------------

Reproducibility

Intermittent (50% of occurrence)

-----------------
System Configuration

STD lab or VM
2 controllers
2 computes
2 storages.

-----------------
Branch/Pull Time/Commit

SW_VERSION="21.12"
BUILD_TARGET="Host Installer"
BUILD_TYPE="Formal"
BUILD_ID="2021-11-26_00-00-08"
BUILD_DATE="2021-11-26 00:08:34 -0500"

-----------------
Last Pass
Stx5.0

-----------------
Timestamp/Logs

NAMESPACE NAME READY STATUS RESTARTS AGE

kube-sstem ns-snmp-7df9b5c8bd-74rll 3/3 Running 0 80s 172.16.154.2 compute-0 <none> <none>

-----------------
Alarms
N/A

-----------------
Test Activity

Regression Testing

-----------------
Workaround

No workaround available

Changed in starlingx:
assignee: nobody → Enzo Candotti (ecandotti)
Revision history for this message
OpenStack Infra (hudson-openstack) wrote : Fix proposed to snmp-armada-app (master)
Changed in starlingx:
status: New → In Progress
Ghada Khalil (gkhalil)
Changed in starlingx:
importance: Undecided → Low
Revision history for this message
OpenStack Infra (hudson-openstack) wrote : Fix merged to snmp-armada-app (master)

Reviewed: https://review.opendev.org/c/starlingx/snmp-armada-app/+/822711
Committed: https://opendev.org/starlingx/snmp-armada-app/commit/9913c1ad6c05a24f2a07bcbf55bff82fa37f1a14
Submitter: "Zuul (22348)"
Branch: master

commit 9913c1ad6c05a24f2a07bcbf55bff82fa37f1a14
Author: Enzo Candotti <email address hidden>
Date: Wed Dec 22 11:14:21 2021 -0300

    Add master NodeSelector to SNMP pods

    SNMP pods should not be installed on worker nodes, since
    trap generation will not work as expected.

    This change adds a node-role.kubernetes.io/master NodeSelector
    in order to force the pods to be Scheduled only on master nodes.

    Closes-Bug: 1955571

    Test Plan:

    PASS: Verify that tolerations and NodeSelector are installed properly.

    PASS: Add a NoSchedule taint to master nodes and verify that
    pods are not being scheduled on worker nodes.

    Signed-off-by: Enzo Candotti <email address hidden>
    Change-Id: I28c279187463cfdc79fd39f40a817be4ee4cf011

Changed in starlingx:
status: In Progress → Fix Released
Revision history for this message
Ghada Khalil (gkhalil) wrote :

screening: Adding the stx.7.0 release tag because this issue is addressed for that release

tags: added: stx.7.0 stx.fault
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.