Bug 1431588

Summary:	[RFE] 'oc adm diagnostics' feature to ignore default nodeSelector
Product:	OpenShift Container Platform	Reporter:	Nicolas Nosenzo <nnosenzo>
Component:	oc	Assignee:	Fabiano Franz <ffranz>
Status:	CLOSED DUPLICATE	QA Contact:	zhaozhanqi <zzhao>
Severity:	low	Docs Contact:
Priority:	medium
Version:	3.4.0	CC:	aos-bugs, bmeng, dma, hongli, jokerman, lmeyer, mmccomas, myllynen, smunilla, xiazhao
Target Milestone:	---
Target Release:	---
Hardware:	Unspecified
OS:	Unspecified
Whiteboard:
Fixed In Version:		Doc Type:	If docs needed, set a value
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2017-06-29 07:57:54 UTC	Type:	Bug
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:
Embargoed:

Description Nicolas Nosenzo 2017-03-13 11:21:56 UTC

> 1. Proposed title of this feature request
'oc adm diagnostic' feature to skip default nodeSelector setting

> 3. What is the nature and description of the request?  

'oc adm diagnostics' doesn't take into account that there might be default node selectors, which when doing a network check causes pods to not be deployed and fail during the MatchNodeSelector predicate.
  
> 4. Why does the customer need this? (List the business requirements here) 

Example Scenario: 

Within a cluster with a default nodeSelector set within the master-config, any attempt of running the 'oc adm diagnostic' tool will fail on creating the pod-diagnostic* with MatchNodeSelector condition.

  
> 5. How would the customer like to achieve this? (List the functional requirements here)  
  
running the 'oc adm diagnostic' tool through the CLI, with a parameter that might allow it to deploy on all nodes and ignore default node selectors.



> 10. List any affected packages or components.

Diagnostic tool.

Comment 2 openshift-github-bot 2017-06-17 08:57:16 UTC

Commit pushed to master at https://github.com/openshift/origin

https://github.com/openshift/origin/commit/4ad812926d6b55464d3f41c3a7f2ecbd7f257a1a
NetworkCheck diagnostic: create projects with empty nodeselector

fix bug 1431588 https://bugzilla.redhat.com/show_bug.cgi?id=1431588
fix bug 1459241 https://bugzilla.redhat.com/show_bug.cgi?id=1459241

When creating the projects in which the test pods will deploy, create
them with an empty node selector so that they can run on any nodes, not
just nodes with the cluster default node selector.

Comment 4 Hongan Li 2017-06-21 06:06:27 UTC

tested in OCP 3.6.121 with defaulNodeSelector configured, the pod "pod-diagnostic-test" can be deployed in default project, but if run "oc adm diagnostics" in user project, the pod will stay in pending status. 

# oc describe pod pod-diagnostic-test-771f0
<...snip...>
Node-Selectors:	dig=enabled
Tolerations:	<none>
Events:
  FirstSeen	LastSeen	Count	From			SubObjectPath	Type		Reason			Message
  ---------	--------	-----	----			-------------	--------	------			-------
  1m		2s		8	default-scheduler			Warning		FailedScheduling	No nodes are available that match all of the following predicates:: MatchNodeSelector (1).

Comment 5 Luke Meyer 2017-06-21 13:55:23 UTC

The customer case was specifically about NetworkCheck, while pod-diagnostic test is from a different diagnostic. Since there's already bug 1459241 to address this for NetworkCheck we can rescope the work remaining on this one to just making the diagnostic pod run on any host. This involves creating a temporary holding project like the NetworkCheck does so it's a bit larger change.

Comment 6 zhaozhanqi 2017-06-29 07:57:54 UTC


*** This bug has been marked as a duplicate of bug 1459241 ***