The Partlan Blog

Information Technology & Mountain Biking

[rpartlan]

Blocking IPv6 everywhere

If you have some PCs on your network with IPv6 enabled, your PCs are probably looking for a DHCP server to get an IPv6 IP. 

And if a bad guy were on your network running a packet capture, they might see some details of the DHCP process, like the mac address of the requesting PC.

Then they can become a IPv6 DHCP server and start talking to the clients and receiving traffic and do man-in-the-middle type attacks. 

OK. so, that’s where we are now. Need to stop IPv6 on the network. And here’s how:

step1: on your switch, create an ACL, let’s call it deny-ipv6 step2: inside the ACL the action can be deny any any. (I think the first any is source, second any is destination). Point is that its blocking everything.

step3: go to the interface that you want to block IPv6 on and do the following:         ipv6 access-group deny-ipv6 in         ipv6 access-group deny-ipv6 out

this says to apply your ACL to IPv6 traffic on this interface for inbound traffic and outbound traffic.

Now we’ve blocked IPv6 for one port. Just do the same thing on every switch everywhere in your network, and we’ve blocked IPv6 everywhere. 

[rpartlan]

All Of The Services...Well, DNS and DHCP

Can DNS and DHCP services be provided from the same linux server? Yes

What if DNS is provided through dnsmasq. And DHCP is provided through isc-dhcp-server? Also yes. But then it gets a bit complicated. I think.

So really, this post is here mostly for me because I’m going to forget how to troubleshoot the problems I’m having.

Ok, let start with the first problem I’m having. isc-dhcp-server when it starts, there is no dhcpd.pid file located at  /run/dhcp-server/dhcpd.pid. status looks something like this:

So for this one:

sudo mkdir /run/dhcp-server sudo touch /run/dhcp-server/dhcpd.pid

and then sudo systemctl restart isc-dhcp-server.service sudo systemctl status isc-dhcp-server.service

That takes care of the isc-dhcp troubleshooting. Other than that, the status command will usually show if there’s any leases handed out. A few other troubleshooting commands are:

sudo tail -f /var/log/syslog cat /var/lib/dhcp/dhcpd.leases

Onto the dnsmasq troubleshooting. This one is a bit weird. I initially set up dnsmasq and it was running just fine. Then, and not sure why, after setting up isc-dhcp-server, dnsmasq started having issues starting. Here’s the screenshot:

So the main issue here is that TCP port 53 is being used by something else. Pretty sure DHCP isn’t using it because DHCP is UDP 68. But no worries. Can use netstat to take a look:

sudo netstat -tulpn | grep LIST

and output looks like:

The issue is that port 53 is being used by systemd-resolved. The solution is to go into /etc/systemd/resolved.conf and uncomment the line that says DNSStubListener=yes, then change yes to no. Then:

sudo systemctl restart systemd-resolved sudo systemctl restart dnsmasq sudo systemctl status dnsmasq

and now dnsmasq should be showing that it is active. In fact, I wonder, if I netstat now, will the service using port 53 be dnsmasq? hmm. Let’s see.

Looks like it is! That’s awesome. So now I’ve got dns and dhcp services running on the same box. Yes, dnsmasq also can provide dhcp, but I didn’t really understand the documentation for how to do multiple scopes. So I decided to try isc-dhcp. This way, if I break it, I don’t (necessarily) affect dns services. Anyway, lunch time. I shall refer to this in the future when this server reboots and everything breaks.

[rpartlan]

Scheduling PCaps

I’m lazy. On a scale of 1-10, maybe around a 5-6 in terms of laziness. I’ll do things if they need to be done, but I don’t want to waste extra energy doing things when there’s an easier, lazier way to accomplish the same task. And that’s specifically what this post is about. 

The task: Run a packet capture for roughly a day. Then stop the packet capture, save the pcap file, and start a new packet capture.

I’ve been going into the server room, sitting on the floor with the packet capture laptop, and doing this stop, save, restart process for about a week now. Its not a particularly painful process. But it can definitely benefit from improvement. (isn’t that the definition of improvement?) 

Now let’s get to the fun part. First, ingredients: 1 PC with RJ45 port. 1 ethernet cable. Optional: 1 USB to RJ45 adapter. Another ethernet cable. 

Prerequisites: You’ll need to already have a mirrored port on your switch. This is just about how to schedule the packet captures. Secondly, for the optional portion, you’ll need to configure a port on a switch, to use the RJ45 adapter for the PC to give it an IP you can manage it on.

Let’s look at the crontabs now:

05 12 * * * cd /root/pcap; /usr/bin/timeout 1435m /usr/sbin/tcpdump -i eno1 -w outfile 03 12 * * * cd /root/pcap; /bin/bash renameToToday.sh

Oh yeah, gotta set up this crontab as root. so sudo crontab -e. Now, as root, we change directory to the pcap directory, use the timeout command, run the following for 1435 minutes, which is 5 minutes short of 1 day. Run tcpdump on the eno1 interface and write the pcap output to the file: outfile.

This script would run from 12:05pm until noon the next day.

3 minutes of nothingness

Then go to the pcap directory again, and run the script renameToToday.sh

And that script looks like:

#!/bin/bash today=$(date +"%b%d") mv outfile $today.pcap

And since today is October 18th, it would rename the file outfile to Oct18.pcap

2 minutes later

start the packet capture again, writing to a file called outfile

[rpartlan]

Fix Duplex Mismatches Remotely

Here’s the scenario: You’re out at a remote location (think branch office, not wilderness) and you’re tasked with installing a router and switch for a MPLS circuit. You pre-configured everything and you proceed to rack, power up, and plug in the equipment. You console onto the router to make sure it sees the correct neighbors. It does. Life is good! Time to pack it up and go back to the central office. Or stop at Starbucks for some coffee first. Maybe go catch some Pokemon. 

Anyway...Eventually you get back to the office. You hop on to make sure everything from the install went correctly, and to you horror, you see:

032012: Apr 23 12:42:09.231 UTC: %CDP-4-DUPLEX_MISMATCH: duplex mismatch discovered on GigabitEthernet0/0 (not full duplex), with some.neighbor.yourdomain GigabitEthernet1/1/1 (full duplex).

And you curse to yourself. Now you have to go back out there and console onto the router and fix the duplex settings. You can’t do it remotely because changing the duplex settings kills the connection. And if you’re using the WAN link you’re editing, you’re going to kill your connection half way through the change. Then you’re really going to have to head back out there. Because instead of a duplex mismatch now you’ve taken down the branch’s WAN.

But there’s a better way. And I’m here to post the better way because I find that much too often I am the person described above.

For reference, I found the commands from this link.

The problem is that the duplex and speed cannot be hard coded onto a router interface when the interface is set to auto negotiate. Logically, then removing the auto negotiation allows for manual configuration of duplex and speed. But removing auto negotiation also kills the WAN connection. And then the configuration cannot be completed.

This solution doesn’t send the sequence of commands until they’re all ready or staged. This is also described executing commands in “batch” or “batch processing”. More modern firewalls allow for this by allowing commands to be entered before they are committed. Anyway this is the same thing but for router interfaces.

tclsh

set fixinterface { ios_config "interface gi0/0" “no negotiation auto” "duplex full" "speed 100" }

eval $fixinterface

Before you start you’ll definitely want to set a reload in 5 or 10 minutes so just in case you mess up the router will reset and you’ll be able to reach it again.

tclsh puts you into a shell style CLI instead of the Normal Cisco IOS CLI. Then you can set the commands into a variable. Then in one shot, execute the multiple commands at once. The connection does drop for a while. But then comes back.

Once you’ve successfully made the changes. do a “reload cancel” to prevent the router from reverting. And also do a “write memory” to save the configuration.

Good job. Now you can enjoy that coffee you bought and not stress out about heading back to the branch office!

[rpartlan]

Network Quarantine - By Any Means Available.

Scenario: You get a call from one of the IT staff. There’s a host/client/device that’s attempting a bunch of user logins and failing. It’s locking out their users. That’s how they found out about this host/client/device causing the issue. Your task (should you choose to resolve the problem, mission impossible style ;p ) is to immediately get the host off of the network. 

Your first question: What’s the IP of the host?

IT staff: A.B.C.D

And then you go about your business. First you try to figure out what the gateway of the host is. it’s probably A.B.C.1 and you’re right! Yay. Ok now, you do 

show ip arp | include A.B.C.D

and you get the IP and MAC address of the host. You also get the port on the router going towards the host. Very Yay! 

Typically you’d hop onto the switch where the MAC address is mapped to a specific port and you’ll see this with the following command:

show mac address-table

Once the port is identified. You’ll shutdown the port. And the host has been quarantined. No network for the host. No more user accounts getting locked out. Yay...or maybe? 

But what do you do if you’re on your way onto the switch to shutdown the port and you find out you don’t have access to the switch? Without access to the switch, the Mac address is pretty much useless. You’re unable to map the Mac address to a port, let alone shut down the port. 

But you do have access to the router. Now here’s what you do.

(Ahem, I really should add that you’re likely dealing with a subnet that is DHCP and that this solution is temporary.)

You don’t actually need to remove the host from the network. You just need to stop it from communicating. Routing is kind of like giving directions to places based on what you know. A router can learn or advertise routes. Typically the learning happens through adjacencies or through dynamic routing protocols. The advertising is usually configured into the router by the network administrator (you). 

So you do the following command on the router:

ip route A.B.C.D 255.255.255.255 null0

And this effectively stops all packets destined for A.B.C.D from getting to the host. You’re telling the router to tell other routers, “Hey, are you looking for A.B.C.D? Yeah? I know where A.B.C.D is. Here following me.” And then the router pushes the packets over a cliff into an abyss. And the packet never makes it to the host.

Keep in mind that in a DHCP subnet, the host you’re trying to quarantine will likely get a new IP once the lease on their current IP is up. (maybe even in half the time of the lease). So you need to keep this in mind and make sure to quarantine the new IP of the offending host, and also fix the route you added above as that IP is no longer associated with the baddie!

But also, the issue should have been resolved before the host asks for a new IP. This is just a quick fix for when you need to quarantine a host, but don’t have access to the switch layer.

[rpartlan]

Powershell: Get Alerts When Your Clock Is Off

Now, I realize this is a silly post, because most servers have NTP built in and that normally servers’ clocks won't be off the official clock by much. But at least for me, I recently experienced a use case where it would be useful to get a notification if the server’s clock is far off from official time. 

Let me explain:

At work we use Microsoft Office 365. But because we have on-premise authentication, users need to open up Office 365, then the hosted service talks to our authentication server, which authenticates the user, and then the user is allowed to use the Office 365 services. Seems simple enough. Now what would happen if the server that handles authentication is off from the official time? Because authentication is a time sensitive transaction, if the clocks between the  Office 365 server and our authentication server are off by too much, the authentication server will not communicate with the Office 365 server. Turns out the limit is 5 minutes. If our authentication server is off by more than 5 minutes it won’t talk to the Office 365 server. And guess what! Our authentication server was recently off from official time by 6 minutes. Yep, and all the users that needed to authenticate couldn’t access their email. 

Now you might be saying, “why don’t you just use NTP? It’s 2019 (almost). This shouldn’t even be a problem!” And I agree, but due to issues such as the authentication server being a virtual machine hosted on a physical server that has time problems, all I can say is that I have no control over the system’s ability to keep time. All I can do is bandage it with a script to help me monitor the server’s clock as it drifts:

And now let’s go through the code. First things first, I run the code on a server where I trust the time. I can control the server’s NTP settings and it stays up to date and doesn’t drift far from official time. The top portion of this script is for setting up the credentials. Then, I will use the credentials to send a command to the authentication server. I also get the time on the local machine. Next, using the timespan command I can compare the values between the authentication server and the local server. I chose to compare minutes but seconds or hours could also be compared. For my use case, minutes was most useful. Lastly, the script does some logic and if the clock is off by more than 2 minutes in either direction, my group gets an email saying the clock is off. Then we can intervene and fix the time on the server, thereby mitigating the problem.

[rpartlan]

FFMPEG convert for iMovie

So this is really more for my notes than anything else. I like to make timelapse movies the hard way: 

set an interval on a DSLR.

import the photos to a computer

use FFMPEG to compile the photos at a specific frame rate and output to video

use some video editor to add an audio track

and done.

(And this could all be done from an iPhone, but why do it the easy way when we learn so much doing it manually!)

Now since I have an iTunes library with some songs I like. I’d like to use iMovie to add the audio track. But sadly, after I create the video file from FFMPEG using: 

ffmpeg -r 10 -pattern_type glob -i "IMG_*.JPG" -q:a 0 -s hd720 -vcodec libx264 -crf 15 time-lapse.mp4

The output file isn’t compatible with iMovie. Something about a wrong or deprecated codec. I don’t exactly remember. But after googling around I found this useful snippet from commandlinefu : 

ffmpeg -i time-lapse.mp4 -c:v prores -profile:v 2 -c:a copy foriMovie.mov

Once I used that command to create my iMovie ready timelapse, I was able to open it up in iMovie, and add my audio track. Simple (though not as simple as the iPhone timelapse) and now I can reference this whenever I need it! Hopefully it helps you out too!

[rpartlan]

Timelapse With a DSLR and Linux

Lately I’ve started getting into photography. I’m also still mountain biking quite a bit. And I’m also still playing around with Linux. Why not combine all three hobbies by using Linux to compile photos shot from a DSLR of my attempt at replacing a dropper post on my mountain bike!

The Bike

Let’s start with the mountain bike. I had recently gone to Moab, Utah to do some biking. The dropper post on my bike was a Rockshox Reverb, which uses hydraulic actuation to trigger the seat post movement. My friends seatpost failed due to the elevation and the trigger losing hydraulic fluid. My post started getting slower as well. I thought it would be best to replace the seat post altogether. I chose a cable actuated seat post: The X-Fusion Manic. Unfortunately, the seatpost sticks out too far on my bike, so I ended up not using it. Instead I’ll be replacing the Rockshox Reverb with a KS Lev Integra seat post.

The Camera

Onto the Camera. I used a Canon 80D with a Canon 10-18mm EF-S lens set at 14mm. The 80D also has a function that allows me to set the camera to take a photo every n seconds for between 1-99 photos at a time, or unlimited. I set the camera on a tripod, and let it take a photo every 5 seconds for unlimited number of times.

The Linux

Once I got the photos onto my computer, I simply used the command line tool, FFMPEG. This tool is quite powerful. I the following:

ffmpeg -r 15 -pattern_type glob -i "IMG*.JPG" -q:a 0 -s hd720 -vcodec libx264 -crf 15 attempt2.mp4

The -r flag sets the frame rate that the output would have. I set my frame rate to 15 frames per second. I tested it with 10 frames per second earlier, but it looked too choppy. The -pattern_type glob flag allows for me to do pattern matching by file name. Since my photos were shot on a Canon camera, the default filename is something like IMG_2556.JPG. I wanted to tell the ffmpeg tool to look for files that were named starting with IMG and ending in JPG. It then figured out to pull the files in order. I don’t really know what the other parameters do. I think the codec libx264 created a mp4 video with the h.264 format. This created the .mp4 video timelapse I was hoping to make. Then I used another tool called Openshot. Openshot is a video editor, kind of like iMovie or Windows Movie Maker. It isn’t all that advanced, but it sort of, kind of gets the job done. I used Openshot to add an audio track to my timelapse video. 

And that’s all there is to it! I hope you enjoy!

[rpartlan]

Watching New Bike Inventory

Scenario: I want to buy another bike. But I don't really need one, and I don't want to spend the money for the bike just yet (because then I won’t be able to buy some other toy I don’t need). But there's a specific model in my size on sale, and the quantity is low. I'd like to be able to keep an eye on the inventory of that model without having to click through the website every day.

Prerequsites:

Linux computer

make sure pip is installed (sudo apt install pip)

make sure beautifulsoup is installed (sudo pip install bs4)

make sure smtplib and re are installed (I’m assuming they are by default).

gmail account

The solution:

scrape the web page for the data

look for the quantity in stock

email the quantity value to myself

schedule the script to run daily

And here’s how it went:

At first I tried using my normal tools, which is a combination of the python request module, and beautifulsoup to parse the html. However, I was having difficulty because the data I was looking for wasn't in the webpage when I did the HTTP get request. It's weird, but there's some information you only get from visiting the web page, and when you do an HTTP get on the webpage, some of that data is missing. I think it's because the data may be in a PHP script, or some other backend script that the client has no access to.

Enter Selenium.

You can install it with

sudo pip install selenium

And use it by:

from selenium import webdriver driver = webdriver.Chrome(executable_path='/path/to/chromedriver/') driver.get('https://website.to.scrape.from/')

and then somewhere before the end of the end of the script:

driver.quit()

One quick thing to note: When using Selenium with a web browser as I am in this case, and specifically with Chrome, I need to download the chrome driver for Selenium to function properly. This can be found at:

http://chromedriver.chromium.org/downloads

Once downloaded, it needs to be unzipped, and then the path of the executable needs to be in the line that says “webdriver.Chrome(executable_path=”

Selenium is different from requests, in that it doesn't just perform an HTTP get. It actually opens a browser and then gets all of the data from the loaded webpage. From there, I could then use beautifulsoup to parse for my html tag that had the inventory count.

To use beautifulsoup:

from bs4 import BeautifulSoup soup = BeautifulSoup(driver.page_source, 'html.parser') thread = soup.html.body.findAll('div',{'class':'items_left'})

for i in thread:     print(i)

And then to email the content to myself:

First make sure that the variable i is converted to a string:

count = str(i)

Next make sure you have imported the smtplib module:

import smtplib

And then do the following:

fromaddr = '[email protected]' toaddr = '[email protected]'

msg = 'this is how many bikes left:  '+count subject = "bike inventory"

message = 'Subject: %s\n\n%s' % (subject,msg)

username = '[email protected]' password = 'yourpassword'

server = smtplib.SMTP('smtp.gmail.com:587') server.starttls() server.login(username,password) server.sendmail(fromaddr, toaddr, message) server.quit()

Good, now almost every piece is done. At this point you can test your script to make sure it actually gets the content from the webpage and emails it you. I tested it for myself and it worked.

Lastly, I need to schedule this script so that it runs daily, and we do that with Cron. First off, we need to make the that at the top of our python script we have:

#!/usr/bin/python

Then, normally the Cron syntax looks like:

00 14 * * * /usr/bin/python /path/to/my/script.py

But this time, because Selenium actually opens a web browser, We need to do it a little differently:

00 14 * * * export DISPLAY=:0; /usr/bin/python /home/rpartlan/Documents/python/selenium-t130.py

We need to say “export DISPLAY=:0 so that the Cron job runs in an environment with a GUI, instead of just command line.

And that’s it! So now I’ll be ready when there’s only 1 of the bikes left in stock. By then hopefully I either forgot about the bike (and thus won’t want to buy it), or I’ve already spent the money on something else (and can’t afford to buy it). 

This is definitely a strange use case. It’s sort of like the opposite of the ticket bots buying tickets as soon as they go on sale. But this could help for when you want to buy something down the road, and just want to know if it’s still in stock, so you can react before they run out.

And finally, here’s the github gist, in case you’d like to see the whole script all together.

https://gist.github.com/partlan/debb86d4e1df1fea8baef1c5a1b9f54e

[rpartlan]

Configure Network Devices With Friends

It’s always more fun with friends! Or maybe it’s just that you can’t quite figure out how to configure the network device. So your coworkers are all crowding around your little laptop screen. Now there’s too many heads in the way. Someone says, “look at interface G 0/0″ and another person says, “wait, I didn’t see the show run”. It gets chaotic. Wouldn’t it be better if there were a way to collaborate, Google Doc style?

Maybe there is.

The Concept: 

set up a Linux computer with the serial cable to the network device.

have everyone connect to the Linux computer via SSH

start a multiuser screen session

connect to the serial port

Let’s put some prerequisites in place and get started. We must assume you’ve got a Linux computer, a console cable, a network device, and SSH enabled on your Linux computer so people can remotely connect to it.

First, install cu and screen

sudo apt install cu && sudo apt install screen

Then test cu to make sure it works. Assuming that you have the console cable plugged into the computer, do:

ls /dev

And you should see output like this:

The little thing over on the right outlined in red says ttyUSB1. That’s what were going to use in this instance for our serial connection (mmm cereal!!)

To connect to the serial connection, open your terminal and type in:

cu -l /dev/ttyUSB1 -s 9600

If you get an error about not having permission, change the permission on /dev/ttyUSB1:

sudo chmod 777 /dev/ttyUSB1

Good. Now we’ve got cu working. Let’s make sure screen is working. The goal to set up a screen session with a name, for multiuser access. The person doing the set up on the server will do:

screen -m -S friends

The other person, or people, will SSH onto the server, check that there is indeed a screen session active:

screen -ls

which will have output looking like:

rpartlan@ubuntoo:~$ screen -ls There is a screen on: 30325.friends (08/30/2018 10:50:50 AM) (Detached) 1 Socket in /var/run/screen/S-rpartlan.

and connect to the screen session by:

screen -x friends

once you’re in the screen session, someone can connect to the serial port using the cu command above. And now you can all sit at your own desk and talk over the phone, or skype, or slack, or spark, or discord, or irc. And no one has to crowd over anyone else while configuring network devices! 

[rpartlan]

The Annual Oyster BBQ

Step 1: get to the secret spot.

Step 2: BBQ.

Step 3: Hang out with everyone!

Step 4: Get Silly!

Step 5: Send a drone out!

Step 6: Do some fishing

Step 6: Have a great time!

[rpartlan]

Portrait Of The Wride

We meet up on Wednesday Evenings

And we go ride trails!

And we look good doin’ in!

And we have fun!

And then we have beer!

[rpartlan]

Sending email with Powershell...on Linux!

Have you tried to send an email with an attachment from a python script? I haven’t. But I have started trying to figure out how to do it. And me being lazy, and resorting to the top StackOverflow links from my searches, I’ve decided that it’s too difficult to send email with attachments in python. I give up. Time to find a different way!

Enter Powershell. I’m not saying Powershell is better than Python. I’m just saying in this one instance, I couldn’t figure it out with Python, but with Powershell. it’s as simple as:

send-MailMessage -From "[email protected]" -To "[email protected]" -Subject "Attachment has been attached to the email" -Body "blahblahblahblah stuff that goes in the body" -SmtpServer "ip.of.relay.server" -Attachments 'C:\path\to\attachment.doc'

Alrighty. So now I have my send email script written in Powershell. It works great on my windows laptop. But I want this to run on my Linux server. Here’s the link to install powershell core on Ubuntu 16.04:

https://docs.microsoft.com/en-us/powershell/scripting/setup/installing-powershell-core-on-linux?view=powershell-6#ubuntu-1604

With powershell core installed, I can type in “pwsh” and i get the powershell shell where I can run powershell scripts. Awesome!

Last step! I want to schedule my powershell emails. Let’s crontab powershell to email out on linux.

crontab -e 

add in the following:

<your crontab stars go here> cd /home/user; /usr/bin/pwsh powershellEmail.ps1

And I’m Done. :)

[rpartlan]

Do You Really Trust Automated PowerBall Number Selection?

If you live one of the 44 United States that take part in the PowerBall Lottery game, then you might have thought about participating last week, when the Jackpot was $560 Million. Too late now, someone from New Hampshire already won.

While I don’t think playing the Lottery is a good investment, I thought it was interesting that the Power Ball machine gives the customer/client/victim/loser the choice to either pick their own numbers, or have the machine “Randomly” choose numbers for the person. I have no data on how prevalent the use is of the automated number selection, but from my own observations, people rarely choose their own numbers.

Hmm. Now doesn’t it seem like a potential conflict of interest: The machine that chooses your numbers is made by the same company that stands to lose if you get the correct numbers? I’d say yes. I’m sure they’d say the machine really is “random”, or even pseudo-random. But it’s just a little hard to believe that the machine would really generate a winning combination. 

So without any further delay, a pseudo-random number generator for PowerBall players, written in python, about 10-20 lines long:

https://gist.github.com/partlan/8ad523a0bcae5a9e9e9407de75b3bc15

There’s not too much to it. Generate 5 numbers, sort. Generate the PowerBall number, and print. Now, if you don’t mind, I’m going to spend my spare change on snacks and coffee.

[rpartlan]

Power To The Terminal Servers

I attended Cisco Connect yesterday and in the vendor alleys I found some interesting terminal server + smart rack mounted power distribution unit (PDU)  solutions.

The first vendor was Opengear

They make and sell terminal servers with built-in cellular so you can connect to your branch office and make changes to the network gear. Very cool, but it gets semi-cooler-ish!

The next vendor was Vertiv. They acquired a company called Avocent (back in 2009) that made terminal servers. Cool, but they also sell smart rack mount  PDUs. The idea is, you’ve got hardware in a remote location, and you can use the terminal server to connect to the out-of-band management interface of your hardware. And after things go wrong and the out-of-band management CLI stops responding to your commands, you get fed up and decide to reboot the hardware. Luckily, you bought the smart rack mount PDU. You can connect to the PDU, and effectively pull the power plug on your hardware. And then plug it back in.

Above is the Vertiv terminal server

Below is the Vertiv rack mounted PDU

But the last vendor takes the cake. Or bagels, or whatever snacks were provided at the conference. The last vendor is WTI. They make the a smart rack mounted PDU as well, but instead of having a separate unit for a terminal server, they have integrated the terminal server into the back of the PDU. On the front side, you get the power ports. On the back side, you get the terminal server ports. And to top it all off, you get the cellular feature similar to what was built-in to the Opengear hardware. With this all-in-one package, I think I’d like to have me some WTI boxes. 

[rpartlan]

Quickly Change DNS in Windows

I have this problem with my laptop where when I’m at home, the default DNS Servers provided by my ISP often don’t work. As a result, I’ll go into my network settings on my laptop and change my DNS server settings to the Google DNS servers. Problem solved! 

But then I go to work, and I need to have my DNS servers be the internal DNS servers instead of Google DNS servers. Once again I go into the network settings and change my DNS server settings back to DHCP. But now that I’m making this change daily, there must be a faster, easier, more efficient way to get this done.

***Powershell enters stage right, waves to the audience and takes a bow***

And here’s where Powershell can come in and help. I made a small powershell script that takes input from the user (static, or dhcp) and depending on the input, either sets the DNS servers to the Google DNS servers, or sets the DNS servers back to DHCP.

https://gist.github.com/partlan/bfbc79dd6023d69aa53044038fbab1b3

Now there is a small issue that needs to be addressed. Running the Set-DnsClientServerAddress requires running Powershell as administrator. So here’s the quick way in Windows 10 to launch Powershell as administrator:

hit the windows key

type “powershell”, (make sure you’re on “Windows Powershell” and not “Windows Powershell ISE”)

then hit ctrl+shift+enter

That will launch Powershell as administrator. If you’re having the same problem, hopefully you’ll find this post useful.

[rpartlan]

More US Area Codes Than 3 Digit Primes

Are there enough prime numbers between 100 and 1000 to cover every US area code? The answer is: no.

And here’s how I figured it out:

1. Get the list of primes between 100 and 999 from http://www.answers.com/Q/What_are_the_prime_numbers_between_100_and_999

2. Save them to a file called “primes.txt” and count them up with:

cat primes.txt | sed 's/,/\n/g' | wc -l

3. Go to Wikipedia to find out all the US area codes: https://en.wikipedia.org/wiki/Category:Area_codes_in_the_United_States_by_state

Notice that at the end of each state there’s a number in parentheses. That is the number of area codes in that state.

4. cURL the Wikipedia page and save it to a file called “codes”

curl -s https://en.wikipedia.org/wiki/Category:Area_codes_in_the_United_States_by_state > codes

5. parse it with python and add up the area codes:

https://gist.github.com/partlan/7a7547472ef867f0845e183fc0a98241

The results were that there are 142 prime numbers between 100 and 999. And that there are 291 area codes in the USA. So no, there are not enough 3 digit prime numbers to cover all US area codes.