<html>
<head>
<style><!--
.hmmessage P
{
margin:0px;
padding:0px
}
body.hmmessage
{
font-size: 12pt;
font-family:Calibri
}
--></style></head>
<body class='hmmessage'><div dir='ltr'><br><br><div>> Date: Sun, 12 Jul 2015 15:45:26 -0400<br>> From: amuller@redhat.com<br>> To: bderzhavets@hotmail.com<br>> CC: apevec@gmail.com; rdo-list@redhat.com<br>> Subject: Re: [Rdo-list] What happens to RDO Kilo on Fedora 22 ?<br>> <br>> Can you paste openvswitch agent and L3 agent logs on the network node<br>> when you spawn a VM and the router loses its qr/qg interfaces?<br><br>I've rebuild via packstack 3 F22 VMs hosting Controller,Network,Compute Node<br>Libvirt subnet for emulating external sub-net was as follows<br><br># cat public.xml<br>
<network><br>
<name>public</name><br>
<uuid>d0e9965b-f92c-40c1-b749-b609aed42cf2</uuid><br>
<forward mode='nat'><br>
<nat><br>
<port start='1024' end='65535'/><br>
</nat><br>
</forward><br>
<bridge name='virbr2' stp='on' delay='0' /><br>
<mac address='52:54:00:60:f8:6d'/><br>
<ip address='172.24.4.225' netmask='255.255.255.240'><br>
<dhcp><br>
<range start='172.24.4.226' end='172.24.4.238' /><br>
</dhcp><br>
</ip><br>
</network><br><br>Attempt to create public sub-net via dashboard failed with message, that <br>172.24.4.225 is incorrect gateway dor 172.24.4.224/28<br>Then I created this sub-net via CLI as admin. This time ( more powerful box ) <br>qrouter-namespace kept staying in normal shape (ifconfig) , but I was unable to ping<br>gateway 172.24.4.225 from Network Node with qg-xxxxx (172.24.4.226) and br-ex 172.24.4.232<br><code><br># </code>source keystonerc_admin<br># ip netns exec qrouter-namespace ping -c 3 172.24.4.225 <br> hanged<br><br>Same results I got for <table height="1" width="657" bgcolor="#CCCCCC" cellpadding="2" cellspacing="1"><tbody><tr bgcolor="#EEEEEE"><td>Network <code>174.24.4.192 / 26 gateway </code><code><code>174.24.4.193</code> </code></td><td><br></td></tr></tbody></table><br> Then I rebuilt external (neutron) and corresponding libvirt sub-net in way it always been used for AIO <br>and two node installs (network XXX.XXX.XXX.0/24 , gateway XXX.XXX.XXX.1) .<br><br># cat public.xml<br><network><br> <name>public</name><br> <uuid>d0e9965b-f92c-40c1-b749-b609aed42cf2</uuid><br> <forward mode='nat'><br> <nat><br> <port start='1024' end='65535'/><br> </nat><br> </forward><br> <bridge name='virbr2' stp='on' delay='0' /><br> <mac address='52:54:00:60:f8:6d'/><br> <ip address='192.168.174.1' netmask='255.255.255.0'><br> <dhcp><br> <range start='192.168.174.2' end='192.168.174.254' /><br> </dhcp><br> </ip><br></network><br><br>In other words used 192.168.174.0/24 as external sub-net. This time I succeeded with sub-net<br>creation via Dashboard. I became able to ping 192.168.174.1 via both qrouter or qdhcp namespaces<br>Outbound/inbound connectivity was tested OK for Cirros 0.3.4 and F22 cloud VMs.<br><br>In meantime I believe that network calculator invoked for creating sub-nets is broken.<br>Another issue is creating ext net via CLI which allows to create unreachable gateway.<br><br>> <br>> ----- Original Message -----<br>> > To enable packstack on F22 there is a hack :-<br>> > <br>> > # dnf install -y https://rdoproject.org/repos/rdo-release.rpm<br>> > # dnf install -y openstack-packstack<br>> > # dnf install -y fedora-repos-rawhide<br>> > # dnf --enablerepo=rawhide update openstack-packstack<br>> > <br>> > along with disabling Neutron verification section in provision_demo.pp it<br>> > allows AIO and<br>> > Two node "Controller&&Network and Compute" RDO Kilo packstack installs on F22<br>> > VMs and bare metal boxes.<br>> > However, splitting Controller and Network Nodes, e.g. three node packstack<br>> > install appears not to be able to work<br>> > in meantime, regardless successful packstack completion.<br>> > What, actually, happens ? I am able to create external net and sub-net only<br>> > via CLI (as admin) and change status to shared.<br>> > <br>> > Attempt to launch VM results qrouter-namespace to loose qg-xxxx, qr-yyyy<br>> > interfaces.<br>> > Outbound/inbound packets forwarding gets broken as soon as VM gets started.<br>> > Only when nova and neutron<br>> > services are running on the same node it doesn't happen. It's just my<br>> > personal experience.<br>> > RDO Kilo Three node deployment on CentOS 7.1 works fine.<br>> > <br>> > Thanks.<br>> > Boris.<br>> > <br>> > <br>> > <br>> > _______________________________________________<br>> > Rdo-list mailing list<br>> > Rdo-list@redhat.com<br>> > https://www.redhat.com/mailman/listinfo/rdo-list<br>> > <br>> > To unsubscribe: rdo-list-unsubscribe@redhat.com<br></div> </div></body>
</html>