On a previous post we saw an example of a network access topology running anycast default gateways.
The idea is to save IPv4 addresses, without other methods than standard routing protocols. Just plain BGP that can be implemented on most vendors, either via hardaware appliances or virtualized network devices.
For the following examples, I’ll simulate a public /24 prefix using the 198.51.100.0/24 which is reserved by IANA as TEST-NET-2 for documentation.
Lab Network
This lab network comprises 2 AGG/BNG routers, a single core router (which will perform additional aggregation), and a single edge router. All devices are MikroTik RouterOS 6.48.6 CHR instances.
The “edge” router is peering with the “core” through their loopbacks, and just advertising a default to it (or cores, on a future stage), and accepting the entire 198.51.100.0/24.
The “core” is peering with the “edge” of course, and also with two BNGs named as agg-bng-xx. This core is advertising its default to them, and accepting all prefixes within 198.51.100.254/24, with a prefix length up to /29.
If you come from a IOS land, this syntax would be something like this.
ip prefix-list BNG
permit 5 198.51.100.254/24 ge 24 le 29
Finally, the BNGs are peering with the core, accepting a default, and allowing any advertisements from 198.51.100.254/24 from /24 to /29.
Both routers have 198.51.100.254/24 as the anycast default gateway.
If you wonder hor this works, this lab network is similar to the one on the previous post which you can check here.
BNG PPPoE and DHCP
We will start by setting up PPPoE services on our BNGs.
At this point we will work it with local secrets and keeping all the AAA process in the same router, with RADIUS as a future option.
Be aware that RouterOS by default will try its best to adjust the TCP MSS to match the PPPoE interface MTU.
Also, this being PPPoE, we have no restrictions on using the first address on the network as the PPP connection will not care about it being a network address. However, this will have the obvious restrictions and behavior if we run DHCP.
We will also skip the .254 address on the address pool as this is assigned to the ether2 interface on both routers as our anycast default gateway.
Finally, we will add some test subscribers. A dumb switch will act as the last-mile technology which could be xPON, wireless, DSL, you name it. All the subs are running RouterOS 6.48.6, and this is just to have something capable to talk PPPoE. There is also a VPCS 0.8.2 which comes by default with GNS3.
GNS3 Topology
Address me, father
Starting with sub03 VPCS, we will ask DHCP to the BNG.
Welcome to Virtual PC Simulator, version 0.8.2
Dedicated to Daling.
Build time: Aug 23 2021 11:15:00
Copyright (c) 2007-2015, Paul Meng ([email protected])
All rights reserved.
VPCS is free software, distributed under the terms of the "BSD" licence.
Source code and license can be found at vpcs.sf.net.
For more information, please visit wiki.freecode.com.cn.
Press '?' to get help.
Executing the startup file
sub03> ip dhcp
DORA IP 198.51.100.252/24 GW 198.51.100.254
sub03> ping 198.51.100.254
84 bytes from 198.51.100.254 icmp_seq=1 ttl=64 time=1.278 ms
84 bytes from 198.51.100.254 icmp_seq=2 ttl=64 time=1.234 ms
84 bytes from 198.51.100.254 icmp_seq=3 ttl=64 time=0.946 ms
^C
If you pay attention, we did get the .252 address, instead of the .253.
sub4 had probably requested this one before, as RouterOS by default comes with a DHCP client on ether1. Is this the case?
[admin@RouterOS] > /ip ad pr
Flags: X - disabled, I - invalid, D - dynamic
# ADDRESS NETWORK INTERFACE
0 D 198.51.100.253/24 198.51.100.0 ether1
Indeed, both are running DHCP. And just for reference, this is how it looks from the BNG.
[admin@agg-bng-02] /ip dhcp-server> lease pr
Flags: X - disabled, R - radius, D - dynamic, B - blocked
# ADDRESS MAC-ADDRESS HOST-NAME SERVER RATE-LIMIT STATUS LAST-SEEN
0 D 198.51.100.253 0C:04:49:87:00:00 RouterOS dhcp1 bound 2m34s
1 D 198.51.100.252 00:50:79:66:68:01 sub03 dhcp1 bound 3m52s
Same is happening with sub01 and sub02, however we’ll remove the DHCP client and add a PPPoE client.
[admin@RouterOS] /interface pppoe-client> add interface=ether1 user=sub1 password=sub1 add-default-route=yes
[admin@RouterOS] /interface pppoe-client> pr
Flags: X - disabled, I - invalid, R - running
0 X name="pppoe-out2" max-mtu=auto max-mru=auto mrru=disabled interface=ether1 user="sub1" password="sub1" profile=default keepalive-timeout=10 service-name="" ac-name="" add-default-route=yes default-route-distance=1
dial-on-demand=no use-peer-dns=no allow=pap,chap,mschap1,mschap2
[admin@RouterOS] /interface pppoe-client> ena 0
[admin@RouterOS] /interface pppoe-client> pr
Flags: X - disabled, I - invalid, R - running
0 R name="pppoe-out2" max-mtu=auto max-mru=auto mrru=disabled interface=ether1 user="sub1" password="sub1" profile=default keepalive-timeout=10 service-name="" ac-name="" add-default-route=yes default-route-distance=1
dial-on-demand=no use-peer-dns=no allow=pap,chap,mschap1,mschap2
[admin@RouterOS] /interface pppoe-client> /ip ad pr
Flags: X - disabled, I - invalid, D - dynamic
# ADDRESS NETWORK INTERFACE
0 D 198.51.100.125/32 198.51.100.254 pppoe-out2
[admin@RouterOS] /interface pppoe-client>
This config looks as follows on sub02.
[admin@RouterOS] >
[admin@RouterOS] > /ip dhcp-client
[admin@RouterOS] /ip dhcp-client> remove [find]
[admin@RouterOS] /ip dhcp-client> /
[admin@RouterOS] > /interface pppoe-client
[admin@RouterOS] /interface pppoe-client> add add-default-route=yes disabled=no interface=ether1 name=pppoe-out2 password=sub1 user=sub1
[admin@RouterOS] /interface pppoe-client> pr
Flags: X - disabled, I - invalid, R - running
0 R name="pppoe-out2" max-mtu=auto max-mru=auto mrru=disabled interface=ether1 user="sub1" password="sub1" profile=default keepalive-timeout=10 service-name="" ac-name="" add-default-route=yes default-route-distance=1
dial-on-demand=no use-peer-dns=no allow=pap,chap,mschap1,mschap2
[admin@RouterOS] /interface pppoe-client> /ip ad pr
Flags: X - disabled, I - invalid, D - dynamic
# ADDRESS NETWORK INTERFACE
0 D 198.51.100.124/32 198.51.100.254 pppoe-out2
[admin@RouterOS] /interface pppoe-client> /ip ro pr
Flags: X - disabled, A - active, D - dynamic, C - connect, S - static, r - rip, b - bgp, o - ospf, m - mme, B - blackhole, U - unreachable, P - prohibit
# DST-ADDRESS PREF-SRC GATEWAY DISTANCE
0 ADS 0.0.0.0/0 pppoe-out2 1
1 ADC 198.51.100.254/32 198.51.100.124 pppoe-out2 0
[admin@RouterOS] /interface pppoe-client>
Alright, we can ping the gateway from both. Can we get beyond it?
sub03> trace 1.1.1.1
trace to 1.1.1.1, 8 hops max, press Ctrl+C to stop
1 198.51.100.254 2.362 ms 1.396 ms 0.958 ms
2 * * *
3 * * *
^C 4
The BNGs is aware of this subscriber, however, we are not advertising anything to the core- yet.
We like connected things
If you recall, on the AGGs, there was a precise out filter on the peering to the core.
Well, the idea is to let the core know about some parts of the subnet, covered by this filter. And the easiest way is to have BGP to
Know there are some hosts running DHCP, probably via static routes pointing to the local interface.
PPPoE subs will already have a dynamic and connected route on the routing table.
Have BGP redistribute connected and statics, in case there are no PPPoE subscribers and we only have DHCP subscribers.
Aggretate all PPPoE interfaces into a supernet, because we are allowing up to /29 This supernet will be a /25, because we created our internet pool from 198.51.100.1-198.51.100.127. Same concept applies for agg02, with the consideration that the aggregate will be 198.51.100.128/25
The include-igp setting will match all IGP routes, like connected routes and iBGP routes.
You can see that the core is aware of a part of the /24 exists on this BNG.
[admin@core-01] > /ip ro pr
Flags: X - disabled, A - active, D - dynamic, C - connect, S - static, r - rip, b - bgp, o - ospf, m - mme, B - blackhole, U - unreachable, P - prohibit
# DST-ADDRESS PREF-SRC GATEWAY DISTANCE
0 ADb 0.0.0.0/0 10.0.1.1 200
1 ADC 10.0.0.0/30 10.0.0.2 ether1 0
2 ADo 10.0.1.1/32 10.0.0.1 110
3 ADC 10.1.1.1/32 10.1.1.1 lo0 0
4 ADo 10.10.1.1/32 10.255.255.2 110
5 ADo 10.10.1.2/32 10.255.255.6 110
6 ADC 10.255.255.0/30 10.255.255.1 ether2 0
7 ADC 10.255.255.4/30 10.255.255.5 ether3 0
8 ADb 198.51.100.0/24 10.10.1.1 200
9 ADb 198.51.100.0/25 10.10.1.1 200
However, with this setup, we are still advertising the entire /24 to the core. Let’s adjust the filters on both routers to advertise only anything longer than 24.
And now, from sub4 for example, let’s run a traceroute to it.
[admin@RouterOS] /tool> traceroute 1.1.1.1
# ADDRESS LOSS SENT LAST AVG BEST WORST STD-DEV STATUS
1 198.51.100.254 0% 2 1.6ms 2.6 1.6 3.6 1
2 10.255.255.5 0% 2 3.2ms 3.8 3.2 4.3 0.6
3 1.1.1.1 0% 2 4.6ms 4.7 4.6 4.8 0.1
There is a special consideration here if you still haven’t noticed it.
How does agg1 knows about what’s happening on agg2. For example, if from agg1 we try to reach hosts on the 198.51.100.128/25 network, the immediate next hop is agg1 itself, because we have a DAC route pointing to 198.51.100.0/24
[admin@agg-bng-01] > /ip ro pr
Flags: X - disabled, A - active, D - dynamic, C - connect, S - static, r - rip, b - bgp, o - ospf, m - mme, B - blackhole, U - unreachable, P - prohibit
# DST-ADDRESS PREF-SRC GATEWAY DISTANCE
0 ADb 0.0.0.0/0 10.0.1.1 200
1 ADo 10.0.0.0/30 10.255.255.1 110
2 ADo 10.0.1.1/32 10.255.255.1 110
3 ADo 10.1.1.1/32 10.255.255.1 110
4 ADC 10.10.1.1/32 10.10.1.1 lo0 0
5 ADo 10.10.1.2/32 10.255.255.1 110
6 ADC 10.255.255.0/30 10.255.255.2 ether1 0
7 ADo 10.255.255.4/30 10.255.255.1 110
8 ADC 198.51.100.0/24 198.51.100.254 ether2 0
9 ADbU 198.51.100.0/25 20
10 ADC 198.51.100.126/32 198.51.100.254 <pppoe-sub1-1> 0
11 ADC 198.51.100.127/32 198.51.100.254 <pppoe-sub1> 0
However, we already have all rhe routing info we need on the core.
The fix is simple and involves of course, filtering, but we’ll cover that on the next post.
Yeah, I finally remembered to make a post about this. I know it will like as a copy-paste of the previous one, because, in fact it is.
Ok, if you have been following the series, you should already know that I equally love and hate UFiber OLTs. They are affordable, deliver a lot of bang for the buck, and have an awful GUI.
Well, the GUI is lovely on v4.
Python in the middle
I wrote a quick and dirty client which acts as a sort of middleware between the HTTP inteface of the OLT and you.
It allows to provision non existing ONUs, GPON profiles, WiFi profiles, retrieve active ONU status and general configuration.
Ok, if you have been following the series, you should already know that I equally love and hate UFiber OLTs. They are affordable, deliver a lot of bang for the buck, and have an awful GUI.
Please, be aware that this can change for better or worse in the future, and at the time I’m writing this the latest firmware is v3.1.3. I trust in you UBNT, hope you can sort out this and give us a better product. I’ll keep my fingers crossed.
Python in the middle
I wrote a quick and dirty client which acts as a sort of middleware between the HTTP inteface of the OLT and you.
It allows to provision non existing ONUs, GPON profiles, WiFi profiles, retrieve active ONU status and general configuration.
This is the core of the project. It uses the OLTCLient class to provide a middleware between you and the HTTP interface of the olt.
Initialize a new OLTClient instance with:
client = olt.OLTClient(host, username, password)
The initialization will handle the login for you, altough you can call the login() method manually.
If the OLT is network reacheable, and you have provided the right credentials, and the OLT WEB GUI is alive and well, you should be ready to start.
You can also connect using cli.py:
$ /cli.py
UFiber Client for fw version 3.1.3
UFiber> help
Documented commands (type help <topic>):
========================================
connect help onu quit show
UFiber> connect 10.20.0.101
Username:admin
Password:
Logging to 10.20.0.101 ...
Connection OK
UFiber>
In a previous post we took a quick look to the Ubiquiti UFiber OLT. As always, UBNT tries to offer a non expensive solution to provide last-mile conectivity for end users. I am using non-expensive because UBNT gear is not cheap. Yeah, it can be affordable, but you only get what you pay for.
We saw that the command line is very limited, even when the software is a fork of Vyatta. There is no way to get ONUs provisioned from the command line, so forgot about Ansible (we love Ansible), netmiko, and other SSH clients tools to ease your life.
UBNT wants you to use the web GUI, period. They offer a dockerized management system called UNMS, which really comes handy after you have provisioned your customers.
Both you and me, as network operators, know that provisioning customers is one of the more boring tasks, but is is still a critical one. Fast and precise provisioning translates in more customers, more stability, faster troubleshooting, and peace of mind.
Yeah, SONAR exists, but not all operators can work with their pricing and technology supports. And don’t even think to integrate billing if you are using electronic invoicing with AFIP in Argentina.
If you are still here, don’t give up. If there is a will, there is a way.
Under the hood
The OLT has a web GUI served by HTTPS, with a self-signed certificate, on port 443. There is no easy way to use a proper certificate here, but well, it’s something.
If you are not authenticated, this is what waits for you in the URL root.
Looks like a login screen
I want to know if this is a standard HTML form. And indeed it is.
Good enough for me. Of course this should be into a try/catch structure, but ymmv.
Shut up and take my ONUs
So, we are logged in. What about ONU configuration? The GUI allows to update firware, but this is done automatically starting from software version 3.1.3. We can also set many parameters of the ONU configuration, and provision them via OMCI, but the ONU should already exist in the ONU list.
Configuring existing ONUs
There is no way to add non-existing ONUs to the configuration, which makes pre-provisioning impossible.
What happens when we click on save?
A POST request is made, passing an interesting payload to the /api/edge/batch.json endpoint. Let’s see how does it looks like.
This POST puts a payload with all the ONU configuration, and a bit more. It uses a JSON structure which looks like this.
Of course this is a fake ONU with a UBNTffffffff serial number. Yeah, we can fool the GUI and send whatever values we want.
This comes real handy because you can pre-provision ONUs before they show up in the PON port.
As an ISP, this means a time saver, because you can deliver ONUs ready to plug and play, and the OLT will hand out all the configuration without further intervention.
The web GUI is handy (altough not so stable), but it really doesn’t makes sense to have to manually provision each ONU when they are connected in the PON port. Technicians have to spend to a lot time on customers houses waiting for the NOC to configure each new customer. This is no-bueno in pandemic times. You want to install as many new customers as possible, as fast as possible, and staying in foreign homes as less as possible.
It seems it’s time for me to code something.
If you are reading this, Robert Pera, please make me a CLI.
As some of you might know, currently I’m working as a network engineer on a medium size ISP. The company had a long history working as a WISP, and in later times they moved into FTTH, trying several vendors among the lead players of the industry.
As some of you might also know, Argentina has a history of economic meltdowns, currency devaluations and import restrictions. Considering this, the best solution to implement a network here is usually the one you can afford, which can provide the performance you need, and over all things, the one you will be able to keep buying in the future.
So, considering all these factors, when planning for a GPON network for a medium size operator…while trying to keep costs low for both the company and customers:
It really doesn’t matter if Calix supports XGS-PON technologies…
Or if Huawei gear is compatible with almost everything…
Or if Furukawa Electric has some great management software…
The real questions to ask were:
Can the company afford the OLTs, and the ONUs for the planned customer base?
Will they be in the market in the years to come?
Enter Ubiquiti UFiber
UFiber offers internet and telecom service providers a cost‑effective fiber optic delivery system for Triple Play Services (data, voice, IPTV/VoD) with speeds of up to 2.488 Gbps downstream and 1.244 Gbps upstream.
OLTs come with dual hot-swap power supplies, 4 and 8 PON ports versions. Every PON port supports 128 CPEs, 20 Km maximum range. The uplinks are two SFP+, which can with in LACP.
The ONUs options, at the time when I’m writing, are:
UFiber Nano – one PON (of course), one Gigabit Ethernet, a fancy LCD display. Passive PoE powered.
UFiber Loco – a PON, a Giga Eth, passive PoE powered or external micro USB power.
UFiber Wifi, like above, but with 4 Giga Ethernet ports, and a 802.11n interface.
UFiber Instant, a nice SFP ONU.
Ok, sounds nice. How do we manage them? There is a web GUI…
Hey, the GUI looks like EdgeOS !
Once logged in, the GUI has a nice dashboard which looks like this. And it crashes from time to time.
They tried
But this not EdgeOS, the OLT is a different product! Let’s ssh into it to get the real feel.
ssh admin@olt
The authenticity of host 'olt (olt)' can't be established.
ECDSA key fingerprint is SHA256:thnWRB2bImsdNuu1ar74GryFwv5r7PoHJsHhJOkHnCQ.
Are you sure you want to continue connecting (yes/no)? yes
Warning: Permanently added 'olt' (ECDSA) to the list of known hosts.
Welcome to EdgeOS
By logging in, accessing, or using the Ubiquiti product, you
acknowledge that you have read and understood the Ubiquiti
License Agreement (available in the Web UI at, by default,
http://192.168.1.1) and agree to be bound by its terms.
admin@olt's password:
Linux olt 4.4.159+ #1 SMP Fri Feb 22 15:28:22 UTC 2019 mips
Welcome to EdgeOS
Last login: Tue May 26 15:50:16 2020 from 190.211.80.70
admin@olt:~$
Ok, this is definitely EdgeOS. So we got a fully featured command line interface, with commands similar to Juniper JunOS.
admin@olt:~$ show configuration | display set
-vbash: display: command not found
Ok, maybe the command line is not so-fully-featured. No worries, I’ll write Ansible playbooks to manage the OLTs anyway. Most of the configuration is Juniper-like, so all I need at this moment is find out how to configure GPON profiles, and provision ONUs.
admin@olt:~$ show configuration | match onu
admin@olt:~$ show configuration | match profile
admin@olt:~$ show configuration | match gpon
gpon {
There you are! Let’s go into configuration mode.
admin@olt:~$ configure
[edit]
admin@olt# show system gpon
isolation enable
mtu 1518
[edit]
admin@olt#
Well, this is awkward. There is nothing about GPON in the command line. Neither in the working configuration, and of course being Ubiquiti, there are no command line manuals.
Love the smell of undocumented commands in the morning
So, I saw two interesting lines before: Linux olt 4.4.15, and -vbash: display: command not found, which tells me this is Linux, not BSD as in Junos, and we have bash.
admin@olt:~$
Possible completions:
add Add an object to a service
clear Clear system information
configure Enter configure mode
connect Establish a connection
copy Copy data
delete Delete a file
disconnect Take down a connection
generate Generate an object
initial-setup Enter initial configuration dialog
no Disable or reset operational variable
ping Send Internet Control Message Protocol (ICMP) echo request
ping6 Send IPv6 Internet Control Message Protocol (ICMP) echo request
reboot Reboot the system
release Release specified variable
rename Re-name something.
renew Renew specified variable
reset Reset a service
restart Restart a service
set Set system or shell options
show Show system information
shutdown Shutdown the system
telnet Telnet to <hostname|IPv4 address>
terminal Control terminal behaviors
traceroute Track network path to <hostname|IPv4 address>
traceroute6 Track network path to <hostname|IPv6 address>
No signs of bourne again shells in the horizon. Does my magic have any power here?
admin@olt:~$ sh
sh-4.4$ whoami
admin
sh-4.4$ sudo su
root@olt:/home/admin#
Finally a decent shell. Which world is this?
root@olt:~# uname -a
Linux olt 4.4.159+ #1 SMP Fri Feb 22 15:28:22 UTC 2019 mips GNU/Linux
root@olt:~# ls -l /etc/ | grep apt
drwxr-xr-x 6 root root 117 Feb 22 2019 apt
We have apt, so this is a Debian world. I checked on /etc/apt/ and there are no repositories, but I am sure I could run cowsay on this. But the fun can wait.
Where is my GPON configuration? It should say “onu” somewhere.
root@olt:/# grep -r "onu" / | more
grep: /proc/sys/net/ipv4/route/flush: Permission denied
/config/onu_config.json: "onu-policies": {
/config/onu_config.json: "onu-list": {
/config/onu_config.json: "onu-profiles": {
/home/admin/.history:show configuration | match onu
/home/admin/.history:show configuration | match onu
Binary file /lib/mipsel-linux-gnu/libbsd.so.0.8.3 matches
Binary file /lib/mipsel-linux-gnu/libnss_hesiod-2.24.so matches
Binary file /lib/udev/hwdb.bin matches
/lib/udev/hwdb.d/20-OUI.hwdb: ID_OUI_FROM_DATABASE=Monument Labs, Inc.
/lib/udev/hwdb.d/20-OUI.hwdb: ID_OUI_FROM_DATABASE=Optical Zonu Corporation
/lib/udev/hwdb.d/20-OUI.hwdb: ID_OUI_FROM_DATABASE=Presonus Corporation
/lib/udev/hwdb.d/20-usb-vendor-model.hwdb: ID_VENDOR_FROM_DATABASE=PreSonus Audio Electronics, Inc.
Binary file /opt/bcm68620/bcm68620_appl.bin matches
Binary file /opt/bcm68620/bcm_dev_ctrl_linux.ko matches
Binary file /opt/bcm68620/bcm_user_appl matches
/opt/vyatta/share/vyatta-cfg/templates/system/gpon/logging/module/node.def:syntax:expression: $VAR(@) in "main", "oltsys", "onu", "session", "events", "mon_th", "sdk"
/opt/vyatta/share/vyatta-cfg/templates/system/gpon/logging/module/node.def:allowed: echo main oltsys onu session events mon_th sdk
I bolded the interesting information.
There is a /config directory, which has a JSON file called onu_config.json
What kind of users? Well, a special user called Ansible, which will use SSH keys to login into remote devices, allowing for full automation on playbooks.
Creating a new key
If you have been following the series, maybe you remember that we already created keys on the Juniper Junos SSH Keys post.
To create a new key, let’s issue the ssh-keygen command as follows. The -f flag tells the output path, and the -C flags specifies a comment.
This inventory only has one host, localhost, and uses three main variables:
ansible_ssh_user, which tell Ansible to use the user ansible
ansible_ssh_private_key_file, which indicates the key for this user
ansible_python_interpreter, just to avoid non needed logs
The playbook will looks like this. Notice we don’t need to gather_facts here, and we will instruct ansible to use become to gain privileges on the destination host.
---
- hosts: all
become: yes
tasks:
- name: Make sure we have a "wheel" group
group:
name: wheel
state: present
- name: Allow 'wheel' group to have passwordless sudo
lineinfile:
dest: /etc/sudoers
state: present
regexp: '^%wheel'
line: '%wheel ALL=(ALL) NOPASSWD: ALL'
validate: 'visudo -cf %s'
- name: Create "ansible" user
user:
name: ansible
comment: Ansible Automation User
groups: wheel
- name: Add ssh key
authorized_key:
user: ansible
state: present
key: "{{ lookup('file', './ansible.key.pub') }}"
First, we want to make sure there is a group called wheel which will group users with administrative privileges.
Then, the /etc/sudoers file will be edited by allowing the wheel group to gain privileges, with a failsafe using a visudo validation.
Once the group has been created, the new user will be created, and a SSH key will be added to it.
It seems allright, but, how should we run the playbook, if the default user is ansible and this user does not exists yet? Let’s give it a try.
$ ansible-playbook create-user.yml -i inventory.yml
PLAY [all] ************************************************************************
TASK [Gathering Facts] ************************************************************
fatal: [localhost]: UNREACHABLE! => {"changed": false, "msg": "Failed to connect to the host via ssh: ansible@localhost: Permission denied (publickey,password).", "unreachable": true}
PLAY RECAP ************************************************************************
localhost : ok=0 changed=0 unreachable=1 failed=0 skipped=0 rescued=0 ignored=0
It fails, as expected, because the ansible user does not exists in the host.
Well, there is a way to provide a one-time password by connecting a as different user. You will need to install sshpass with your favourite package manager, like apt install sshpass.
One installed, run the playbook once again with the following arguments:
-e “ansible_ssh_user=xxxxx”, where xxxxx is a valid user on the remote host
-kK, which tell Ansible to ask for a login and a sudo password
$ ansible-playbook create-user.yml -i inventory.yml -e "ansible_ssh_user=arturo" -kK
SSH password:
BECOME password[defaults to SSH password]:
PLAY [all] ************************************************************************
TASK [Gathering Facts] ************************************************************
ok: [localhost]
TASK [Make sure we have a "wheel" group] ******************************************
changed: [localhost]
TASK [Allow 'wheel' group to have passwordless sudo] ******************************
changed: [localhost]
TASK [Create "ansible" user] ******************************************************
changed: [localhost]
TASK [Add ssh key] ****************************************************************
changed: [localhost]
PLAY RECAP ************************************************************************
localhost : ok=5 changed=4 unreachable=0 failed=0 skipped=0 rescued=0 ignored=0
Awesome, we have sucessfully created a new user!
Let’s try to connect using the ansible user with its key, as defined in the playbook.
Ansible is an IT automation tools which can configure systems, deploy files and software and orchestrate almost every possible task you can imagine.
Unlike other configuration management tools like Chef or Puppet, Ansible is agent-less, and does all its magic over SSH. Forget about keeping up client daemons updated and additional certificates. Just use your existing SSH keys, or usernames and passwords.
Due to it agent-less nature, we just need a host to initiate SSH sessions towards the managed hosts. This node is called a control node in the Ansible jargon.
VirtualBox is a general-purpose full virtualizer for x86 hardware, targeted at server, desktop and embedded use. Get it from here.
Get Vagrant
We’ll use Vagrant to quickly spin up a control node, based on a ubuntu/xenial64 box. If you are not familiar with Vagrant, you just need to know it is a tool capable to provision VMs on different hypervisors, using a Vagrantfile a VM template.
In order to make everything easier, we’ll init a new git repository to keep track of all the configuration changes. Also, we can push this repository to a remote and clone from wherever we need it, making a portable Ansible control node.
Although everything can be done from a shell, I prefer to use the integrated terminal on VSCode, and make use of the universe of plugins. There is support for Ansible and Ansible Playbooks, Vagrant, a lot more!
Also, even if this post uses a Windows system, everything is reproducible on Linux platforms because all the tools are multiplatform.
Init the Vagrant environment
The Vagrant CLI is self explanatory and easy to understand.
C:\Users\Arturo\Desktop\ansible-control-node>vagrant
Usage: vagrant [options] <command> [<args>]
-v, --version Print the version and exit.
-h, --help Print this help.
Common commands:
box manages boxes: installation, removal, etc.
cloud manages everything related to Vagrant Cloud
destroy stops and deletes all traces of the vagrant machine
global-status outputs status Vagrant environments for this user
halt stops the vagrant machine
help shows the help for a subcommand
init initializes a new Vagrant environment by creating a Vagrantfile
login
package packages a running vagrant environment into a box
plugin manages plugins: install, uninstall, update, etc.
port displays information about guest port mappings
powershell connects to machine via powershell remoting
provision provisions the vagrant machine
push deploys code in this environment to a configured destination
rdp connects to machine via RDP
reload restarts vagrant machine, loads new Vagrantfile configuration
resume resume a suspended vagrant machine
snapshot manages snapshots: saving, restoring, etc.
ssh connects to machine via SSH
ssh-config outputs OpenSSH valid configuration to connect to the machine
status outputs status of the vagrant machine
suspend suspends the machine
up starts and provisions the vagrant environment
upload upload to machine via communicator
validate validates the Vagrantfile
version prints current and latest Vagrant version
winrm executes commands on a machine via WinRM
winrm-config outputs WinRM configuration to connect to the machine
First, run vagrant intit to initialize a new environment.
C:\Users\Arturo\Desktop\ansible-control-node>vagrant init
A `Vagrantfile` has been placed in this directory. You are now
ready to `vagrant up` your first virtual environment! Please read
the comments in the Vagrantfile as well as documentation on
`vagrantup.com` for more information on using Vagrant.
Wait! You have a new file on your git repository. Make it the first commit.
On the newly create Vagrantfile, we can set configuration and provisioning options for the VM.
First, replace config.vm.box = "base" with config.vm.box = "ubuntu/xenial64" to use this box.
If you want to see the VM, config vb.gui = true. By default this setting is commented out, so the VM will start in headless mode.
Feel free to customize the VM memory by commenting out the vb.memory = "1024" setting and adjusting it to your needs.
Once you are done, save your changes, commit, execute vagrant up, and watch Vagrant create a brand new VM for you.
C:\Users\Arturo\Desktop\ansible-control-node>vagrant up
Bringing machine 'default' up with 'virtualbox' provider...
==> default: Importing base box 'ubuntu/xenial64'...
==> default: Matching MAC address for NAT networking...
==> default: Checking if box 'ubuntu/xenial64' version '20191114.0.0' is up to date...
==> default: Setting the name of the VM: ansible-control-node_default_1574353915423_60685
==> default: Clearing any previously set network interfaces...
==> default: Preparing network interfaces based on configuration...
default: Adapter 1: nat
==> default: Forwarding ports...
default: 22 (guest) => 2222 (host) (adapter 1)
==> default: Running 'pre-boot' VM customizations...
==> default: Booting VM...
==> default: Waiting for machine to boot. This may take a few minutes...
default: SSH address: 127.0.0.1:2222
default: SSH username: vagrant
default: SSH auth method: private key
default: Warning: Connection reset. Retrying...
default: Warning: Connection aborted. Retrying...
default: Warning: Remote connection disconnect. Retrying...
default:
default: Vagrant insecure key detected. Vagrant will automatically replace
default: this with a newly generated keypair for better security.
default:
default: Inserting generated public key within guest...
default: Removing insecure key from the guest if it's present...
default: Key inserted! Disconnecting and reconnecting using new SSH key...
==> default: Machine booted and ready!
==> default: Checking for guest additions in VM...
default: The guest additions on this VM do not match the installed version of
default: VirtualBox! In most cases this is fine, but in rare cases it can
default: prevent things such as shared folders from working properly. If you see
default: shared folder errors, please make sure the guest additions within the
default: virtual machine match the version of VirtualBox you have installed on
default: your host and reload your VM.
default:
default: Guest Additions Version: 5.1.38
default: VirtualBox Version: 6.0
==> default: Mounting shared folders...
default: /vagrant => C:/Users/Arturo/Desktop/ansible-control-node
Well, how do we login into the new VM? Try vagrant ssh
C:\Users\Arturo\Desktop\ansible-control-node>vagrant ssh
Welcome to Ubuntu 16.04.6 LTS (GNU/Linux 4.4.0-169-generic x86_64)
* Documentation: https://help.ubuntu.com
* Management: https://landscape.canonical.com
* Support: https://ubuntu.com/advantage
0 packages can be updated.
0 updates are security updates.
New release '18.04.3 LTS' available.
Run 'do-release-upgrade' to upgrade to it.
vagrant@ubuntu-xenial:~$ whoami
vagrant
vagrant@ubuntu-xenial:~$
Type exit and you will return to your system prompt.
You’ll see that you have new items on your directory. A .log file, with the output from Vagrant, and a .vagrant/ directory.
To exclude them from the repository, create a .gitignore, with the following contents.
.vagrant/
*.log
Save and commit.
Provisioning the VM
Return to the Vagrantfile, and add a section like this:
This block instructs Vagrant to execute commands on the shell. First, the package list is updated, then the software-properties-common package is installed. To install Ansible, the ppa:ansible/ansible repository is added, and then Ansible is installed.
Notice the usage of -y to avoid interactive prompts.
Save the file, commit, and run vagrant provision. This will re-provision the VM with the new instructions from the Vagrantfile.
Once Vagrant finishes, connect once again with vagrant ssh, and run ansible --version.
That’s great! We have just installed Ansible on our new VM.
Seek and Destroy
You might wonder, where is this VM? Head to the VirtualBox Manager, and you’ll find it inside.
Here it is!
Without asking anyone.. kill the machine and delete it.
Hard power off
Delete all files
Head back to the console, and you will see that the SSH session has died. Of course, this makes perfect sense as the VM no longer exists.
vagrant@ubuntu-xenial:~$ Connection to 127.0.0.1 closed by remote host.
Connection to 127.0.0.1 closed.
If you try to run vagrant provision, it will fail, as there is no VM to provision.
C:\Users\Arturo\Desktop\ansible-control-node>vagrant provision
==> default: VM not created. Moving on...
But what happens if you run vagrant up once again? Surprise! The VM will be recreated with all the previous configuration settings.
Using Ansible
Well, the VM is ready, Ansible is installed.. now what? Let’s write an inventory file and see what can we do.
I’ll create a simple file to connect to a testing docker instance on 192.168.85.253. No need to worry about ssh keys now, as we are testing, username and password will do the trick.
The [all:vars] section configures the connection settings for all hosts in the file. The [docker] section contains the lists of hosts, in this case just 192.168.85.253.
Let’s run: ansible -m ping -i docker all, where -m is the module to run, -i is the inventory file, and all is the section of the inventory file which contains the hosts. Notice this is a very special ping.
vagrant@ubuntu-xenial:~$ ansible -m ping -i docker all
[DEPRECATION WARNING]: Distribution Ubuntu 16.04 on host 192.168.15.253 should use /usr/bin/python3, but is using /usr/bin/python for backward compatibility with prior Ansible releases. A future Ansible release
will default to using the discovered platform python for this host. See https://docs.ansible.com/ansible/2.9/reference_appendices/interpreter_discovery.html for more information. This feature will be removed
in version 2.12. Deprecation warnings can be disabled by setting deprecation_warnings=False in ansible.cfg.
192.168.15.253 | SUCCESS => {
"ansible_facts": {
"discovered_interpreter_python": "/usr/bin/python"
},
"changed": false,
"ping": "pong"
}
Whoa! That definetly doesn’t look like a ICMP ping. First, this is an old box about to be decomissioned, so it has been very unmantained. Ansible warns us about an old version of Python, and a old version of Ansible which was installed previously on the remote hosts.
Now, try ansible -m setup -i docker all. Be aware this will throw a ton of data to you. The setup module gathers data about the host.
Experts agree that we retain only 10% of what we read but 90% of what we do. Perhaps this explains why the global pass rate for most IT exams is a ghastly 40%.
101 Labs’ mission is to turn you into an IT expert by doing instead of reading. Using free software and free trials, the labs take you by the hand and walk you through every aspect of the protocols and technologies you will encounter in your IT career.
Configuration tips and tricks are shared with you as well as how to avoid the common mistakes many novice engineers make, which can quickly become career-ending.
You can get a free preview or buy the ebook below.
101 Labs – Linux LPIC1 takes you through the beginner exam syllabus, the LPI Linux Essentials in case you are a novice and want to learn the basics of Linux or simply just brush up your skills. You then tackle all the main LPIC1 topics broken into the two exams 101 and 102.
I upgraded a CHR cluster with the main objectives of reduce costs, improve network redundancy and provide an easy administration for CHR instances. As explained in previous posts, CHR can be run on many popular hypervisors, and most users are having great success using Hyper-V Failover clusters or vSphere HA to provide highly available routers without depending on VRRP or other gateway redundancy protocols.
These virtual routers currently provide two main services besides routing for ISP customers. They act as PPPoE concentrator for FTTH users, and provide traffic shaping and policing depending on the customer service plan.
Server Hardware
For this node, I will use a 32 core Dell R730, with 32 GB of RAM, and a 500 GB RAID 10 storage. On future post, new hosts will be added to the cluster.
Unracking the server
Network Conectivity
This server comes with a 4 port Gigabit Ethernet NIC, which could be used without any issues with the ixgbe driver.
First idea was to use two ports in a LACP bundle, and the other two in separate port groups.
I had previous Netflow analysis where I saw a predictable traffic behavior, where most of the bandwidth usage was going from and to a CDN peer of the ISP network. Customers had a mix of public and private addresses of the Class B segment, and they were being moved to CG-NAT ranges. In other words, traffic from a specific set of addresses were going from and to a specific set of addresses.
Why not configure two port-channels, instead of using separate port groups? I tested and due to the nature of the IP addressing on the customer side of the routers, none of the available hashing modes for LACP allowed to achieve a decent distribution on both links of the port-channel.
So, for the purposes of this cluster, I added an Intel X520 dual SFP+ card, providing 20 Gbps conectivity to the CHR instances. Peak bandwidth usage was around 4200 Mbps, so this card is more than enough to allow for future grow.
Installing the Intel X520 NIC
The Intel X520 only supports Intel branded SFP modules, and this behavior can be tuned configuring the kernel module. However, for this particular scenario, where both ports will be connected to a top of rack Dell Force10 S4048-ON switch, I choosed to use DAC cables to keep things simple.
DAC cables on the switch
Connecting the server
The server is using ESXi 6.5 for the hypervisor. After booting, I noticed the NICs were being recognized as vmnic5 and vmnic6, but they were using the ixgbe driver and only establishing links at 1 Gbps.
I downloaded the ixgben driver which is provided by VMware itself here and uploaded it to ESXi via SFTP.
Then I followed the KB article to disable the native ixgbe driver and use the new one. First, I placed the host on maintenance mode, and then I executedthe following to disable the driver.
[root@esxi] esxcli system module set --enabled=false --module=ixgbe
After a reboot, the new ixgben driver was loaded, and the NICs were establishing links at 10 Gbps.
I added the new NICs to the previously created virtual switches, checked the correct assignments of port groups, and then migrated the VMs to this host.
