Ashutosh Narayan

DevOps Days India 2011

Ashutosh Narayan — Sun, 11 Sep 2011 11:30:36 +0000

DevOps Days is an open event which is typically considered as an interaction between what is traditionally considered as Development activity and an Operations activity. I added up to Operations activity for this two day event held at ThoughtWorks, Bangalore on 27-28 August 2011. There were talks, interactive sessions and OpenSpace sessions altogether.
The first day began with a welcome talk by Ajey Gore, followed by talk by Nick Hines on Continuous Delivery. Nick was focusing on things like – basic fundamentals, automation, branching,right sizing, continuous improvements; Development practices like automation testing, code on mainline, branching by abstractions, non-functional automation testing, Build – development pattern : build-pipeline, canary releasing. Tooling: Chef, Puppet, CFEngine, Infrastructure as a Code.
The other interesting talk given by speakers from Infosys – “Why is it difficult to do Test Driven Development and Continuous Integration in large companies ?” was an eye opener. Mainly because of maintenance, reliability and performance, feature release, investments in process automation, culture change, dependencies. Mirror environments , need for constant change, split of activities, maturity levels. These were those immense factors which made them think twice before adopting it and deviating from patterns.
Test Load Balancer [ TLB ] was the talk which I liked the most ; it splits tests into multiple mutually exclusive and collectively exhaustive partitions and allows it to run in parallel to reduce build time. There were speakers from SlideShare who presented how they were able to build a system that deploys 5 times a day in a production environment. I attended OpenSpace sessions by techs from MakeMyTrip on Monitoring Systems using Nagios, Puppet, Ganglia. They also mentioned about identifying metric for monitoring, they have their own inventory system to deal with each clusters.
Second day witnessed a talk by Garrett Honeycutt from Puppet Labs which was so worth learning that I did mention about it’s usage in my organisation. We can have hosts on demand, reduce entropy, disaster recovery, infrastructure as a code, usage of dashboard. Automate most of the activities on nodes. Garrett’s talk’s slide is available here
I am not a Ruby guy but could definitely understand it’s power – Ajey & Nikhil spoke on Deployment Pattern in Ruby on Rails world. Ajey’s openspace session was the last thing I attended, went on for more that 45mins and got to know about reducing build times from few days to few hours using automated deployment. Things that they concentrate while setting up an environment – define a standard hardware, choose OS, configurations and provisioning them.
Had a rich experience at the event followed by meeting some old faces too. It’s four years now since I am with Systems and working in Operations, but yes is DevOps solving issues like – (1) Afraid of changing once application is delivered. (2) Risky deployments (3) It works on my machine.

The DevOps movement is still in its infancy, but it’s gathering pace – there are conferences, mailing lists, irc channels, blogs and people to get to know. I’m convinced this isn’t just a fad, and we’re on the brink of a revolution in the software industry – a paradigm shift in which developers and sysadmins start to work together, to train each other, and ultimately to blur the boundary – welcome to the world of the DevOps.

Firefox: “Save and Quit” option disappeared ?

Ashutosh Narayan — Tue, 06 Sep 2011 17:36:12 +0000

I was finding it difficult to save my currently opened tabs whenever I quit Firefox browser ; later I learnt that up from Firefox 4, one change that has been brought is to by default stop Firefox displaying warning messages. It can be changed from the preferences.
Here is how I learnt a way to get back “Save and Quit” option in Firefox 6.0 which has disappeared somehow :
1. Type about:config into the location bar and press Enter.
2. Accept any warning messages that pops up. You will be taken to list of preferences.
3. Locate the preference browser.tabs.warnOnClose, if its value is set to false, double-click on it to change it’s state to true
4.Repeat this for 3 other preferences browser.warnOnQuit, browser.warnOnRestart and browser.showQuitWarning

Happy browsing !

Yahoo! Open Hack India – 2011

Ashutosh Narayan — Sat, 06 Aug 2011 15:25:56 +0000

It was the fourth hack event conducted by Yahoo! in India and my first hack event in Bangalore. I do not use most of the Yahoo! technologies while at work, but yes did some stuffs using Y! Pipes, Yahoo! Travels. What called me to this event was the enthusiastic crowd and the kind of hack ( software / hardware ) which was going to happen during this two day event. JavaScript guru Douglas Crockford, has come down for the event.
The event kicked off with some technical talks by Y! folks. YQL, Geo-hacking and OAuth sessions were worth attending. Most of the applications developed during the hack used one of these. YQL turns data on the web into databases. Geo-hacking involved things like – finding a current location, turning a location into a place, finding geographical hierarchy, geo-tagging information, displaying geo-information. Learnt about BOSS [ Build your Own Search Service ], since the back-end is proprietary and owned by Microsoft, one has to pay 80cents per 1000 query. DuckDuckGo uses BOSS for it’s search engine. Did interact with Y! folks discussing stuffs related to systems rather than web scale applications. Got to know about happenings at Y! Labs.
Hack challenges : Location based, Mobile based, Entertainment based. Hackers were asked to come up with applications using Y! open technology mostly available at Y! Developer Network and their latest open platforms. It was time to come up with some ideas and put them in some structure. Teams were formed and few did an individual hack as well. An applications named ‘FakeMyTrip’ wherein a user can put his photo to any fake location (s)he wish to visit, was applauded a lot. There were hardware hacks too – showing mind’s concentration while moving a remote control toy car, LED indicators which shows amount of CPU, memory consumed when a site is loaded, LED indicator glows red when a child tries to access unauthorised websites. 50 finalist were selected and they were given 90 secs to demo it. Grand prize was a trip to New York to participate in Y! Hack All Stars Event, where winner will compete with winners of other open hacks from around the world.
The grand prize winners were actually two – [1] Spotlight – Illuminate your part of the World ( Find out best place to hangout ) [2] D.A.M.N. – Single Serving Friend ( connects people on demand to share meter fare in auto-rickshaw / taxis )
It was a privilege meeting techies from Yahoo!. The event was a place of good learning and experience.

How far will you get with “rm -rf /” on Linux ?

Ashutosh Narayan — Thu, 21 Jul 2011 12:57:06 +0000

Have you ever left your Terminal logged in, only to find a so called friend/ colleague came up and typed ” rm -rf / ” ?

I actually did so on a virtual machine (VM). It’s a CentOS VM which was used for some applications test. I created a backup of this VM and thought of shooting a video as I proceed to run ” rm -rfv / ” on it.

You can watch my video recordings here.

I also did so to learn what all directories do not get deleted with this command. And also, after the command has executed will “rm” have removed itself ? It seems that OS protects certain directories like /sys, /proc, /dev. At the end of the video you can see that “ls” command is not found and so is “/bin/rm”. Needless to say /bin, together with /bin/echo has been deleted. So, issuing ” echo * “ just displayed * at the promt. I rebooted the VM and ended up with a black screen. OS is gone !! After the command finishes, disk is wiped off, including the OS. The Kernel and current processes will continue to run from memory, but many processes will die because they fail to access some file. OS will obviously fail to boot next time.

Still unanswered – Files that is currently open ( i.e rm ) even if deleted, won’t disappear from disk until closed. I am not sure, if it is safe to assume, the Kernel could very well load the removed file into memory and remove it from disk immediately ; and keep this memory copy running until the command is open ( i.e rm is running )

Mistakenly, do not run this command on host system’s terminal. I did so on a VM, else you will land up putting your head on hand, hand between knees ! as one who’s world has just come to an end.

New Net Protocol challenges

Ashutosh Narayan — Wed, 09 Mar 2011 04:34:09 +0000

IPv6 as we all know allows greater number of addresses. As it does so, organisations will face challenges adapting new system. Lack of awareness coupled with new vulnerabilities are likely to create security problems.

What happens when we type a website address ? The computer breaks it up into a mix of unique 32 zeroes and ones. This binary code identifies and locates the computer where the website resides..Techies call it Internet Protocol (IP). As I mentioned in my previous post on IPv6 addressing it is clear that IPv6 will have a mix of 128 zeroes and ones. It will offer a practically unending number of website addresses. Right now the main issue is “lack of awareness” among network professional, security architect. Proper training is required in-order to handle this new protocol. Since the migration is going to be gradual, systems running IPv4 and IPv6 will co-exist. We may need newer device-complaint with the new protocol in-oder to make older devices to recognise IPv6. This however can be dangerous..IPv6 has some vulnerabilities when it comes to sending emails..Emails are broken up into parts when on way to recipients. Every part consists of header – a group of information on destination and how it should join with other parts when it has reached destination..

Moreover, if your home modem is five years’ old, it will not be compatible with IPv6 protocols. I am happy that my home modem is new and can support IPv6 addressing.

Experiences with USB-Modem

Ashutosh Narayan — Mon, 07 Mar 2011 09:04:25 +0000

wvdial – no process found was the first output thrown on double clicking my new USB Modem icon. I gave enough trial when my system was armed with Fedora-Core 13 but all went in vain. I then read few instructions which came along with USB-Modem kit and concluded to say ‘no’ to Fedora-Core 13 :(

I installed Ubuntu 9.04 and whoa the script ran successfully. It threw some console outputs : dial creator
–> WvDial: Internet dialer version 1.60
–> Cannot get information for serial port.
–> Initializing modem.
–> Sending: ATX0
OK
Later it initialized the Modem on providing the password and was waiting for the carrier. CONNECT was the next echo on terminal wherein it fetched two DNS servers to resolve IPs.
CONNECT
–> Carrier detected.
–> Starting daemon at mm-dd-yyyy hh:mm:ss
nameserver x.y.z.a
nameserver x.y.z.b

While playing with 1X speed ; I learn t a new stuff – The OPERA Turbo (which is a server side optimization and compression technology) that speeds up data transfer and reduces the amount of data that needs to be downloaded in order to view the page by upto 80%. The compression can be achieved by image down-sampling hence the graphic distortion. More can be read on the portal – http://www.opera.com/portal/turbo/

Browsers and the Web

Ashutosh Narayan — Sat, 26 Feb 2011 22:47:28 +0000

I recently bought a usb modem as the PSTN line at my home was dead since high time and I was unable to browse at home.

After getting the device talk to DNS I installed Google Chrome on my desktop. As I launched Chrome I saw a document under

http://www.20thingsilearned.com/home

As illustrated by Christoph ; I found it very useful and thought readers of my blog would like it so thought of sharing this page.

Happy reading !

Web on console

Ashutosh Narayan — Tue, 14 Dec 2010 09:46:01 +0000

Most of the people think “graphical interfaces” when they think of surfing the Web. And there are indeed great programs like Firefox, Chrome under X11. But console in not the waste land. They also have lots of utilities for downloading or uploading content.

Lets say you want to surf the web. The first utility to use is one of the oldest command Lynx. I also started using Lynx when a machine was not able to handle X11. It is just a basic command , you give a file name or URL you want to visit. So, say you want to hit Google :

lynx http://www.google.com

Lynx then asks you whether you want to accept cookies Google is trying to set. Once you either accept or reject the cookie Lynx loads the page. You will notice there are no images. But all the links and text box to enter the search query is there. You can navigate from link to link using arrow keys. Since contents are text based , the items are at different locations. You can handle more than one URL when you use Lynx. Use can use option= accept_all_cookies to avoid warning messages of asking about cookies when a URL is launched.

Lynx does have few issues, it has a hard time with HTML interpreting and it does not handle frames. You can mention proxy with the option -http-proxy host:port

For text replacement of Lynx you have Elinks ; which has colors, tables, background downloading, tabbed browsing. Useful options are anonymous 1 and -lookup. When you use this Elinks prints out the resolved IP addresses for a given domain name.

Now you can look the Web content from the command line, how can you interact with the Web ? Say, you want an offline copy and read it at your leisure on a terrace. You can use curl to do that. It can do things like HTTP, POST, FTP, SSL connections. Just specify a name value pair.

curl http:// your_site. {a, b,c}.com

which hits all three sites.

curl http://your_site.com/doc[1-5].odt

which downloads all five files.

What if you want to copy an entire site for offline viewing ? wget is the one which you need to use.

wget -p -k -r www.your_site.com

-r re-curses through the site path, -k rewrites downloaded file, -p for an extra content eg: images. And if you want to upload a content to a Web, use wput.

So, now you would be able to interact with the Web without having to use Graphical Interface. Yet another reason to keep you on command line :)

10 things about IPv6 addressing

Ashutosh Narayan — Sun, 31 Oct 2010 16:44:42 +0000

IPv6 has been inching toward becoming a mainstream technology. But where to begin when it comes to IPv6 adoption. Here I share 10 pointers which I learnt and hope it makes you understand how IPv6 addressing works.

IPv6 addressing are 128-bit hexadecimal numbers. As we have all seen that IPv4 addresses are made up of four numerical octets, which combine to form a 32-bit address. Each octet consists of a decimal ranging from 0-255, and are seperated by periods (.) But in IPv6 addresses are expressed in a series of eight 4-character hexadecimal numbers, which represents 16-bits each ( for a total of 128-bits)

Link local unicast addresses are easy to identify. IPv6 reserves certain type of headers for different types of addresses. Eg: Local unicast address always begin with FE80 ; multicast always begins with FF0x ( x is place holder, 1-8 )

Leading zeroes are suppressed.Because of their long bit lengths IPv6 addresses tend to contain lot of zeroes. So, when a section of an address starts with one or more zeroes, those zeroes are nothing more than placeholders and can be suppressed. Eg: FE80:CD00:0000:0000:0CED:1342:2A11:756D so if this is a real address, then any leading zeroes can be suppressed and result would be – FE80:CD00:0:0:0CED:1342:2A11:756D. This way the address is shortened.

Inline zeroes can sometimes be suppressed. Real IPv6 address tend to contain nothing but long sections of zeroes, which can also be suppressed. Eg: FE80:CD00:0000:0000:0000:0000:2A11:756D. Rather than suppressing leading zeroes, one can get rid of all by using two colons. Hence the address now becomes – FE80:CD00::2A11:756D. Two colons in the address tells the OS that everything in between them is a zero. Two things that has to be kept in mind is that, one can suppress a section only if it contains nothing but zeroes. Another, is one can use double colon notation only ONCE in any given address.

Loopback address do not even look like addresses. The designated loopback address in IPv6 looks like this – 0000:0000:0000:0000:0000:0000:0000:0001. So, once all the zeroes have been suppressed, the address becomes – ::1

No traditional subnet mask. In IPv4 every IP address comes with correponding subnet mask. IPv6 also uses subnets but the subnet ID is inbuilt into the address. In IPv6, first 48-bits are network prefix , next 16-bits are subnet-ID, last 64-bits are interface ID. Thus, using a 16-bit subnet and 64-bit device ID provides for 65,535 subnets, with quintillions ( number that can be respresented as a one followed by 18 zeroes ) of possible device ID per subnet.

DNS is still valid. IPv6 uses AAAA resource records, referred as Quad A records.

IPv6 can tunnel its way across IPv4. IPv6 is generally not compatible with IPv4 networks. As a result, a number of transition technologies use tunnelling to fascilitate cross network compatibility. Teredo and 6to4 are two of them.

You might be using IPv6. Beginning with Windows Vista, MS began installing IPv6 by default. Because Windows implementation of IPv6 is self configuring, your computers could be broadcasting IPv6 traffic w/o your even knowing it.

Windows does not fully support IPv6. For Eg: it is possible to include an IP address within a Universal Naming Convention ( \\127.0.0.1\C$ ), however one can’t do this with IPv6 address, because as Windows sees a colon it assumes one is referring to drive letter. And the work around for this is that – MS has established a special domain for IPv6 addressing. All that needs to be done is – replace colons ( : ) with dashes ( – ) and append .ipv6.literal.net

PyCon-India 2010

Ashutosh Narayan — Sat, 16 Oct 2010 09:10:46 +0000

Yet another open source event, PyCon India 2010 is the primary Python conference in India. It is been hosted for the second time in India. It was a two day event held on 25 – 26 Sept’10.

Got up early in the morning, caught a bmtc bus and reached the venue, MSRIT before time. I was happy to see some old faces and even happier when they also recognized me.

David Goodger was the keynote speaker. After the inaugural ceremony, I moved to lecture hall to attend a session on Python 2to3, followed by functional programming in Python. During break I had discussions with FOSSEE, a team from IIT-Bombay who are working on project to enable students and faculty to use open source software tools for science and education, thereby improving quality of instructions and learning. SAGE session by them was worth attending. I missed my laptop for their hands on, but gave a spin with their linux distribution at home.

Post lunch, I attended GUI programming using PyQt, followed by Twisted programming ( the speaker was my interviewer as well ). And how mobile phones can be forgotten ? Compiling Python runtime libraries on mobile phone, this reminded me of OpenMoko talk.

It was day two, I spent most of the time in discussions and learning what type distro folks over there were using. Met few ThoughtWorkers who actually cleared most of my doubts regarding mobile-based application and Django. Missed sessions on Google App Engine, as there were some other interesting talks at the same time.

The conference ended with Indian Python Software Society annual meeting and I was happy to be a part of it.