Ask Your Question
0

vm CAN reach internet and ping gateway..... cannot ping or ssh [mitaka] [closed]

asked 2016-09-06 17:00:01 -0500

updated 2016-09-06 19:26:44 -0500

From the NoVNC CONSOLE: I can configure a static ip address and ping the gateway (physical router). I can configure /etc/resolv.conf and get to the internet...... sweet. I can ssh TO another host (physical) on the same layer2 I can ssh TO another host (physical) on a different network (layer 3-7)

ISSUES: Instance CANNOT DHCP. I CANNOT ssh or ping the instance, even from the same layer 2.

security group is default. tcpdump is not available on cirros.

Physical dhcp server reports cirros getting a lease:

15   cirros fa:16:3e:d9:49:3f   10.1.15.3   VCI: udhcp 1.20.1 Hostname: cirros  6 Days 1 Hours 8 Minutes 2 Seconds  Leased out  cirros
15   localhost  fa:16:3e:80:aa:fc   10.1.15.4   VCI: anaconda-Linux 2.6.32-642.el6.x86_64 x86_64 Hostname: localhost    5 Days 23 Hours 39 Minutes 34 Seconds   Leased out  localhost
15   cirros fa:16:3e:2b:43:ea   10.1.15.5   VCI: udhcp 1.20.1 Hostname: cirros  6 Days 1 Hours 8 Minutes 3 Seconds  Leased out  cirros
15   cirros fa:16:3e:b2:1d:2e   10.1.15.6   VCI: udhcp 1.20.1 Hostname: cirros  6 Days 1 Hours 8 Minutes 4 Seconds  Leased out  cirros
15   cirros fa:16:3e:ec:ac:a8   10.1.15.7   VCI: udhcp 1.20.1 Hostname: cirros  6 Days 1 Hours 8 Minutes 5 Seconds  Leased out  cirros
15   cirros fa:16:3e:b0:18:9f   10.1.15.8   VCI: udhcp 1.20.1 Hostname: cirros  6 Days 1 Hours 8 Minutes 5 Seconds  Leased out  cirros
15   cirros fa:16:3e:cd:09:07   10.1.15.9   VCI: udhcp 1.20.1 Hostname: cirros  6 Days 1 Hours 8 Minutes 5 Seconds  Leased out  cirros
edit retag flag offensive reopen merge delete

Closed for the following reason too localized by bcollins
close date 2016-09-15 08:40:37.331713

2 answers

Sort by ยป oldest newest most voted
0

answered 2016-09-14 13:01:20 -0500

dbaxps gravatar image

updated 2016-09-14 13:16:15 -0500

Question:-

If you are using more than one node for OpenStack (i.e., not an all-in-one installation), then you must use VLANs.

No idea where you got it . Since RDO Juno VXLAN tunneling is default for packstack multinode deployment. You may obviously update answer-file for VLAN tenant's segregation. Worked fine on Kilo && Mitaka. On Liberty I had to file a bug regarding packstack failure properly configure Compute Nodes for VLAN (vm/data) connections to Network (Controller/Network) Node. See ( same answer-file would work for RDO Mitaka)
http://dbaxps.blogspot.ru/2015/10/rdo...

Question :-

ISSUES: Instance CANNOT DHCP. I CANNOT ssh or ping the instance, even from the same layer 2.

Enable DNSMASQ Logging to see what's going on.

edit flag offensive delete link more

Comments

This confused me. I restructured for vlans and am now debugging the openvswitch agents on the compute nodes.. I will tell you one odd thing that i noticed.

Every other packstack disconnects and drops gateway for my primary (mgnt) interface..... interesting times....

bcollins gravatar imagebcollins ( 2016-09-15 07:35:57 -0500 )edit
0

answered 2016-09-14 12:00:23 -0500

updated 2016-09-14 12:05:54 -0500

Reading through RDO documentation because i am playing with packstack and the centos openstack mitaka repo and noticed this line:

If you are using more than one node for OpenStack (i.e., not an all-in-one installation), then you must use VLANs.

Is this correct?

Guess I will test this out and update:

Rough Plan: 
Shutdown openstack services.
Reconfiguring by removing neutron networking objects (test instances, routers, subnet,network). 
Deleting the bridges for each vlan and ifcfg-bond0.xxxx vlan interfaces.
Create one bridge for vlans (br-vlan) and leave untagged bridge interface.
Bring up openstack services.
add network with provider segmentation_id and type vlan.
edit flag offensive delete link more

Get to know Ask OpenStack

Resources for moderators

Question Tools

1 follower

Stats

Asked: 2016-09-06 17:00:01 -0500

Seen: 123 times

Last updated: Sep 14 '16