Allow disabling the network plugin configuration

[tomassedovic]

This is useful when we want to use a different networking solution
without having any networking plugin code run when the services start
up.

[openshift-bot]

Can one of the admins verify this patch?

[tomassedovic]

Not sure using the use_openshift_sdn fact is the best way to do this, but it's what I expected to happen when I set it to false.

[sdodson]

Can this not be achieved by setting os_sdn_network_plugin_name="" ? Also setting openshift_use_openshift_sdn=false seems to do the same. This assumes that openshift actually does the right thing if networkPluginName is empty, my experience has been that it does, are you seeing otherwise?

[tomassedovic]

I thought so, but it just ended up setting the config back to the default
plugin.
On Oct 9, 2015 7:51 PM, "Scott Dodson" notifications@github.com wrote:

Can this not be achieved by setting os_sdn_network_plugin_name="" ?

—
Reply to this email directly or view it on GitHub
#664 (comment)
.

[tomassedovic]

I admit I haven't tried setting the network plugin to an empty string in the conf files, but I couldn't figure out how to do that via openshift-ansible anyway.

I've tried it again to make sure: using master (commit 8e776b0), and setting os_sdn_network_plugin_name="" in the inventory.

That resulted in networkConfig -> networkPluginName being set to redhat/openshift-ovs-subnet in both conf.yaml files and all the interfaces being created.

Then I added openshift_use_openshift_sdn=false and tested that (on a new set of VMs). Interestingly, while the config is the same, this caused the origin-node services to fail to start, because they couldn't find the redhat/openshift-ovs-subnet plugin. That in turn caused the ansible run failure because the nodes never registered back to master.

Started Origin Node.
I1012 03:47:55.544844   23682 manager.go:158] Machine: {NumCores:2 CpuFrequency:2099998 MemoryCapacity:3975819264 MachineID:19d82c8902374e7caa655973bd7a6e6a SystemUUID:314AFD1D-86BA-E44A-B451-5FB22DAE0134 BootID:3e51cecc-756d-44a5-8c60-4bc6b68fd27f Filesystems:[{Device:/dev/vda1 Capacity:42932649984}] DiskMap:map[252:0:{Name:dm-0 Major:252 Minor:0 Size:107374182400 Scheduler:none} 253:0:{Name:vda Major:253 Minor:0 Size:42949672960 Scheduler:none}] NetworkDevices:[{Name:eth0 MacAddress:fa:16:3e:a7:35:60 Speed:0 Mtu:1400} {Name:eth1 MacAddress:fa:16:3e:de:07:53 Speed:0 Mtu:1500}] Topology:[{Id:0 Memory:4294430720 Cores:[{Id:0 Threads:[0] Caches:[{Size:32768 Type:Data Level:1} {Size:32768 Type:Instruction Level:1} {Size:4194304 Type:Unified Level:2}]}] Caches:[]} {Id:1 Memory:0 Cores:[{Id:0 Threads:[1] Caches:[{Size:32768 Type:Data Level:1} {Size:32768 Type:Instruction Level:1} {Size:4194304 Type:Unified Level:2}]}] Caches:[]}] CloudProvider:Unknown InstanceType:Unknown}
I1012 03:47:55.632350   23682 manager.go:165] Version: {KernelVersion:3.10.0-229.7.2.el7.x86_64 ContainerOsVersion:CentOS Linux 7 (Core) DockerVersion:1.7.1 CadvisorVersion:0.16.0}
I1012 03:47:55.632699   23682 server.go:402] Using root directory: /var/lib/origin/openshift.local.volumes
I1012 03:47:55.633269   23682 server.go:704] Watching apiserver
F1012 03:47:55.634191   23682 node.go:141] failed to create kubelet: Network plugin "redhat/openshift-ovs-subnet" not found.
origin-node.service: main process exited, code=exited, status=255/n/a
Unit origin-node.service entered failed state.

I may be missing something here, but we need a way of making sure no network plugin code gets run when launching the node service (to integrate with external SDN solutions such as Neutron). Maybe it's just a bug/oversight somewhere in the fact-gathering process? I'll admit to not being too familiar with Ansible and this codebase.

[sdodson]

Sorry, I thought something else was already taking care of this. LGTM just rebase

[mfojtik]

@tomassedovic care to rebase? :)

[tomassedovic]

Sorry. I'm sick, will do it tomorrow.
On Oct 19, 2015 11:36, "Michal Fojtik" notifications@github.com wrote:

@tomassedovic https://github.com/tomassedovic care to rebase? :)

—
Reply to this email directly or view it on GitHub
#664 (comment)
.

[tomassedovic]

@sdodson @mfojtik sorry for the delay. Rebased.

[mfojtik]

[merge]

[mfojtik]

my powers are weak in this repo :-) @sdodson

[mfojtik]

@mmahut ❤️

[tomassedovic]

Thanks!