search

redhat-openstack / openshift-on-openstack

A place to write templates, docs etc. for deploying OpenShift on OpenStack.
Apache License 2.0
136 stars 87 forks source link

HA install fails - deploying on packstack --allinone #47

Closed juddmaltin-dell closed 8 years ago

juddmaltin-dell commented 8 years ago

Is it a complete anti-pattern to install OpenShift HA like this on a single hardware node running Packstack --allinone with heat, etc? If most everything is contained in different VMs, wouldn't it work? Thanks!

heat stack-create my_openshift --poll -t 180 \ -e openshift_parameters.yaml \ -e openshift-on-openstack/env_ha.yaml \ -f openshift-on-openstack/openshift.yaml

20:01:5 2016-02-24 0f732d51-9bef-4512-bb6b-17701ce885c1 [lb_host]: CREATE_FAILED Conflict: resources.lb_host.resources.host: Port ea66dec1-2c40-48ae-84d9-95c6ff681ccb is still in use. (HTTP 409) (Request-ID: req-2981aa4f-3e2e-4cee-8260-3df28be386f3) 20:01:5 2016-02-24 0661636a-3885-42fe-a15e-22fe7e185dfa [my_openshift]: CREATE_FAILED Resource CREATE failed: Conflict: resources. lb_host.resources.host: Port ea66dec1-2c40-48ae-84d9-95c6ff681ccb is still in use. (HTTP 409) (Request-ID: req-2981aa4f-3e2e-4cee-826 0-3df28be386f3)

Port Details

Name my_openshift-lb_host-w5ty7wsehank-port-bv5bpxr4yp34 ID ea66dec1-2c40-48ae-84d9-95c6ff681ccb Network ID 575dff93-bc92-433a-864a-d292b475b75d Project ID 00116efdc59f44f6954eeae263b36821 MAC Address fa:16:3e:ae:e7:bf Status Active Admin State UP Fixed IP None Attached Device Device Owner compute:None Device ID 17b7c98e-251e-44fe-9b40-19fa83d02990 Binding Host hw1.openshift.oss.lab Profile None VIF Type ovs VIF Details

    port_filter True
    ovs_hybrid_plug True

VNIC Type normal

scollier commented 8 years ago

@markllama @jprovaznik Any thoughts here?

jprovaznik commented 8 years ago

You can use all-in-one single node openstack deployments (actually I use this for development) but make sure you set heat parameter "master_server_group_policies: affinity", otherwise master instances will end in error state. But it seems this one is a different error. I wonder if it's because lb host resource fails to create and was re-created -> that would explain why the lb host port is being used, anything interesting in nova scheduler logs? Also maybe "heat event-list -n 3 my_openshift" might help.

One important note - today we merged a commit which removes dedicated loadbalancer host, instead neutron's LBaaS service is used so if you have neutron LBaaS enabled you might give it a shot.

juddmaltin-dell commented 8 years ago

It turns out my packstack storage was in an error state thanks to a known bug.

Adding docs to readme to ensure that --os-neutron-lbaas-install=OS_NEUTRON_LBAAS_INSTALL is set to "yes."