Horizon Webpage doesn't come up after install completes

[Neal]

I'm new to Mirantis (Fuel 7.0) so stay with me here.

I'm just installed my 1st OpenStack test site using 3 systems (1 Controller, 1 Cinder and 1 Compute Node) and everything appears to be fine. When I attempt to access OpenStack by clicking "proceed to horizon" the system brings up a blank webpage that eventually times out.

Here are the network settings I'm using:

Public Range:
192.168.1.1 - 192.168.1.240

CIDR- 192.168.1.0/24

Gateway- 192.168.1.254
Float range - 192.168.1.241 - 192.168.1.253

Storage:
192.168.10.0/24 (VLAN 101)

Management:
192.168.20.0/24 (VLAN 102)

Neutron L2 Configuration (Default settings)

Neutron L3 Configuration (Default settings)

NOTE: 192.168.1.0/24 is the network external to the OpenStack deployment with 192.168.1.254 as the correct default gateway.

The system is assigning 192.168.1.2 as the horizon IP address.
When I look at tcpdump I don't see any traffic hitting the system when I try to connect to the horizon website (https://192.168.1.2).

The weird thing is I am able to connect to that same webpage locally on the controller system when I telnet to it on port 443 so the issue is outside systems don't get a response from that IP address.

br-ex is assigned IP address 192.168.1.3/24 and I'm able to ping it from outside systems. I can not ping or connect to 192.168.1.2. Wireshark shows retransmits.
When I perform 'arp -a' on the controller system I can see the system is bound to the same MAC address of 192.168.1.3 but the only presence I see of this IP address is in the /etc/hosts file.

This feels like a bug but I'm not sure. Any help?

[Neal (dn789d)]

Also I have 3 interfaces on each system:
eth0 (Admin PXE)
eth1 (private/storage/management)
eth2 (public) - Connects directly to external network 192.168.1.0/24

[Roman Podoliaka (rpodolyaka)]

Neal, could you please attach a diagnostic snapshot?

Also, we'll probably need to see the status of Pacemaker managed resources. Please attach the output of:

crm status

from the controller node.

[Neal (dn789d)]

Sorry for the late response. I; beeven stuck in training.

Please see the diagnostic per your request. As stated before I'm very new to this product but looking forward to learning

[Neal (dn789d)]

screenshot of crm status

[Neal (dn789d)]

Hello Roman,

Just checking in to see if you saw anything of note.

[Alexey Stupnikov (astupnikov)]

Hi Neal!

I have tried to reproduce reported issue and had no success: everything works is it should. Therefore I will close this issue as Invalid and will give you few tips to troubleshoot this issue.

1. You access Horizon through virtual IP address assigned to controller's haproxy netns. You can use "ip netns exec haproxy ip a" command to confirm this.

2. You can use tcpdump to capture horizon's traffic on controller host:
tcpdump -i br-ex host 192.168.1.2
ip netns exec haproxy tcpdump -i b_public host 192.168.1.2

3. You should probably doublecheck routing issues on controller host and inside all network namespaces.

4. Check pacemaker's status.

If you will not be able to find out the roots of this problem, please use openstack forums, IRC channels and maillists to gain new clues and ideas. Good luck!