[Murano] After “ScaleNodesDown” action Application isn't available

Bug #1590460 reported by Mykola Stolyarenko
10
This bug affects 1 person
Affects Status Importance Assigned to Milestone
k8s-docker-suite-app-murano
Confirmed
High
Dmytro Dovbii

Bug Description

Detailed bug description:
Murano applications isn't available after "Scales Nodes Down" action. Master Node is available and the count of nodes are correct.

Expected results:
Application should be available by URL

Actual result:
Application isn't available

Steps to reproduce:

1
Log into OpenStack Horizon dashboard as admin user
2
Navigate to “Murano” > “Application Catalog” > “Environments”
3
Create environment:
4
Click on “Create environment” button
5
Set name of environment = “test”
6
Click on “Create”
7
When environment will be created, click on the name of the environment to navigate to the environment configuration page
8
Configure environment: click on the button “Add component”
9
Add Kubernetes Cluster application to the environment:
10
Find “Kubernetes Cluster” in the list of applications
11
Click on “Add to Env” button
12
Fill “Step 1” of the creation form:
13
Specify “Cluster Name” = “TestKubeCluster”
14
Set initial number of minions to 2, maximum number of minions to 2
15
Select checkbox “Assign floating IP to Kubernetes nodes”
16
Set Kubernetes node hostname pattern to ‘kube-#’
17
Select checkbox “Enable cAdvisor monitoring”
18
Set initial number of gateway nodes to 1, maximum number of gateway nodes to 2
19
Select checkbox “Assign floating IP to gateway nodes”
20
Specify “Gateway hostname pattern” = “gateway-#”
21
Field “Custom Docker registry URL” leave empty
22
Click on “Next” button
23
Fill “Step 2” of the creation form:
24
Select “Instance Flavor” = ‘m1.medium’(2 VCPU, 4G RAM)
25
Provide keypair, set “Availability zone” = “nova”
26
Click on “Next”
27
Select “Continue application adding”
28
Click on “Create”
29
Add Kubernetes Pod to the environment
30
Find “Kubernetes Pod” in the list of applications
31
Click on “Add to Env” button
32
Fill “Step 1” of the creation form:
33
Specify “Pod Name” = “testpod”
34
Spesify “Labels” = “testkey=testvalue”
35
Select created earlier Kubernetes Cluster in the dropdown menu “Kubernetes cluster”
36
Set “Replicas” parameter to 0
37
Click on “Next”
38
Select “Continue application adding”
39
Click on “Create”
40
Add some application to the environment
41
Click on “Deploy This Environment” button
42
Make sure that deployment finished successfully, all nodes are active, app is installed inside the container
43
Navigate inside the created environment to the list of its components
44
Select “Deployment History” tab.
45
In “Actions” column against your deployment click to “Show details”.
46
Navigate to “Logs” tab and find endpoints to Kubernetes API, cAdvisor and your applications.
47
Check that applications deployed correctly (put URLs from environment reports to your browser or terminal with telnet). (cAdvisor, Kubernetes API, Application)
48
Navigate inside the created environment to the list of its components
49
In “Actions” column against “TestKubeCluster” component click to ⤋ button and select “ScaleNodesDown” action.

50 Try to Open the Influx DB URL

Reproducibility:
Always

Description of the environment:
ENVIRONMENT: MOS 9.0 ISO 443

Changed in mos:
importance: Critical → High
Revision history for this message
Stan Lagun (slagun) wrote :

How long have you been waited after the action? Can you confirm that it is ot available after 5 minutes?

Revision history for this message
Mykola Stolyarenko (mstolyarenko) wrote :

Confirmed. Were checked with different applications and timeout was more then 15 minutes.

Dina Belova (dbelova)
Changed in mos:
status: New → Confirmed
Dina Belova (dbelova)
Changed in mos:
assignee: Murano Dev (murano-devs) → Stan Lagun (slagun)
tags: added: mos-murano
removed: murano
Revision history for this message
Stan Lagun (slagun) wrote :

This is a murano-apps bug, not murano (at least until investigation show it is not so). AFAIK murano-apps are released independently from MOS and thus do not affect 9.0 release

Revision history for this message
Dina Belova (dbelova) wrote :

adding non-release tag due to the previous comment

tags: added: non-release
Dmytro Dovbii (ddovbii)
Changed in mos:
assignee: Stan Lagun (slagun) → Dmytro Dovbii (ddovbii)
Revision history for this message
Dmytro Dovbii (ddovbii) wrote :

it is not a bug of murano or murano application. This is just a behavior of kubernetes itself. If user wants to keep pod alive after manipulating with k8s nodes, he should specify replicas parameter to create replication controller which aims to control access to the application.

Since v1.1 kubernetes provides control of pods created without replication controller [1]. So, this issue should be fixed with updating k8s murano image to higher version from current v1.0.6

1. https://github.com/kubernetes/kubernetes/issues/16846

Changed in mos:
status: Confirmed → Won't Fix
Revision history for this message
Oleh Baran (obaran) wrote :

After updating kubernetes to v1.3.0 in ubuntu14.04-x64-kubernetes image I received environment "deploy failure" when run scaleNodeDown action.

In murano-engine logs I observe next exception:
[murano.engine.system.agent.AgentException]: {'errorCode': 100, 'message': u'Script removeMinion returned error code', 'extra': None, 'details': {u'stdout': None, u'stderr': u'Error from server: client: etcd cluster is unavailable or misconfigured', u'exitCode': 1}, 'time': u'2016-07-18 13:50:24.200586'}

Changed in mos:
status: Won't Fix → Confirmed
Dmytro Dovbii (ddovbii)
affects: mos → k8s-docker-suite-app-murano
Changed in k8s-docker-suite-app-murano:
milestone: 9.0 → none
Changed in k8s-docker-suite-app-murano:
milestone: none → ongoing
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.