Note: This bug is displayed in read-only format because the product is no longer active in Red Hat Bugzilla.

Bug 1231292

Summary:	Openshift-sdn not watch to the correct etcd path due to path change
Product:	OpenShift Container Platform	Reporter:	Akram Ben Aissi <abenaiss>
Component:	Networking	Assignee:	Rajat Chopra <rchopra>
Status:	CLOSED CURRENTRELEASE	QA Contact:	Meng Bo <bmeng>
Severity:	unspecified	Docs Contact:
Priority:	unspecified
Version:	3.0.0	CC:	ccoleman, dario.minonne, pep, tscherf, xtian, yadu
Target Milestone:	---
Target Release:	---
Hardware:	Unspecified
OS:	Unspecified
Whiteboard:
Fixed In Version:		Doc Type:	Bug Fix
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2015-09-09 08:50:35 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 Akram Ben Aissi 2015-06-12 15:03:03 UTC

Description of problem:

Openshift-sdn is not operating configuration anymore, this seems to be due to the fact that paths for minions in etcd have changed between beta 2 and beta 4

In beta2 minions where announced under the path: “/registry/minions”.
In beta4, minions are announced under “kubernetes.io/minions”.

This seems to be stated in this issue as well:
https://github.com/openshift/openshift-sdn/issues/71

Version-Release number of selected component (if applicable):
3.0.0-beta4

How reproducible:


Steps to Reproduce:
1. Start Openshift with SDN enabled and at least 2 minions
2. Deploy a replication controller with at least 2 pods
3. Ensure that pods are deployed on 2 differents nodes
4. Connect into one container using docker exec -ti /bin/bash and launch a nc -L 9000
5. Connect into the other container using docker exec -ti /bin/bash and launch a nc IP_ADDRESS_OF_FIRST_CONTAIER 9000

Actual results:
Networking connectivity between pods does not work.
Also use on the minion host the ovs-vsctl command to check that ovs is correctly configured.

Expected results:
Pods should be abled to communicated with each other. OpenVSwitch configuration should have been setup and working.

Additional info:

Is there any available workaround?

Comment 2 Clayton Coleman 2015-06-12 15:25:48 UTC

This is because you're running the old package for SDN but the new package (renamed to openshift-sdn-ovs) is not installed.  In beta4 the package was updated to have a different name, and the upgrade process won't automatically install the new package for you.  You'll need to install the new package and uninstall the old one.  A clean beta4 install will use the new package.

Comment 3 Akram Ben Aissi 2015-06-12 15:57:25 UTC

Thank you for your responsiveness, we will be testing this very shortly and get back in touch very soon.

Indeed we packaged an image here which references the incorrect package.

That would certainly be a false flag.

Comment 4 salvatore dario minonne 2015-06-12 16:23:12 UTC

Writing on behalf on Eric Mountain:

We got this 
[root@eric-rhel71-permissive-beta4-fix ~]# subscription-manager repos \
> --enable="rhel-7-server-rpms" \
> --enable="rhel-7-server-extras-rpms" \
> --enable="rhel-7-server-optional-rpms" \
> --enable="rhel-server-7-ose-beta-rpms"

Error: rhel-server-7-ose-beta-rpms is not a valid repository ID. Use --list option to see valid repositories.
Repository 'rhel-7-server-rpms' is enabled for this system.
Repository 'rhel-7-server-optional-rpms' is enabled for this system.
Repository 'rhel-7-server-extras-rpms' is enabled for this system.

Can someone please point out the name of the replacement repo?  I couldn't see an obvious contender using "subscription-manager repos"...

Thanks,
Eric

Comment 5 Yan Du 2015-06-15 09:45:43 UTC

Test on latest Openshift Enterprise env, two container on different node can communicate normally. Move bug to verified.

oc v0.6.0.1-78-g1c816ed
kubernetes v0.17.1-804-g496be63

Container1:
# oc describe pod hooks-1-cjqyu
Name:				hooks-1-cjqyu
Image(s):			openshift/mysql-55-centos7:latest
Host:				minion1.cluster.local/10.14.6.107
Labels:				deployment=hooks-1,deploymentconfig=hooks,name=mysql
Status:				Running
IP:				10.1.0.109


Container2:
[root@Daphne sample-app]# oc describe pod hooks-1-0vn1c
Name:				hooks-1-0vn1c
Image(s):			openshift/mysql-55-centos7:latest
Host:				minion2.cluster.local/10.14.6.135
Labels:				deployment=hooks-1,deploymentconfig=hooks,name=mysql
Status:				Running
IP:				10.1.1.114


Container1:
bash-4.2$ nc -l 9000
hello
hi

Container2:
bash-4.2$  nc 10.1.0.109 9000
hello
hi

Comment 6 Josep 'Pep' Turro Mauri 2015-09-09 08:50:35 UTC

It seems this was fixed before OSE 3 GA but left open. Closing it.