Inspired by Richard Tappenden‘s recent post about If This Then That I’ve set up my own basic couple of “recipes”:
So since I’m going to be spamming you all with random micro brain dumps during next week, I though I should at least make the spamming as efficient as possible and automate it across three channels!
So this one is the first test..
Whilst I’ve been attempting to learn a new thing each month this year, I’ve been finding it really tricky to keep to the pretty loose schedule. As such, I though I’d try and note down every time a shiny new thing takes my interest, so that I have some idea why I’m incapable of completing a series of blog posts.
I’m really having problems with EC2 these days, and I’m constantly being dragged back into finding out what the problem(s) is/are..
Had to add entries into the rc.local to restart httpd and mysql, and also had to change the permissions on the file to allow it to be executed upon restart (sudo chmod 6755 /etc/rc.d/rc.local).
Instigator
— because the blog went down a couple of times and I didn’t notice!
WordPress 2 DropBox – killed me. As does mysqldump in general.
DB backup
mysqldump --add-drop-table -u <username> -p <database> | bzip2 -czs > <db backup filename>.bz2
website backup
tar -cjf <site backup filename>.bz2 /var/www
Instigator
— wordpress kept telling me to upgrade to the latest version, but also had a big MAKE SURE YOU BACKUP FIRST warning; and the WordPress 2 DropBox plugin killed my EC2 microinstance.
#!/bin/sh
#
# Script to check CPU usage and tweet me if it's over a threshold
#
if [ ` uptime | awk '{ print$10 }' | cut -d. -f1 ` -gt 50 ];
then
sudo python /home/ec2-user/tweepy/ec2Event.py '@rposbo LOAD: ' < /tmp/load
else
echo lo
fi
This totally doesn’t work! Uptime seems to happily report low CPU usage even when the site is dying. Maybe CPU is low and mem is high.. hmm.. may have to change that check.
Instigator
— attempting to backup wordpress keeps breaking my site, dammit!
Currently trying to get those three to work together and help me find out why the EC2 microinstance is so.. uh.. micro..
Instigator
— EC2 just keeps on dying and I need to find out why
Whilst I’ve been attempting to learn a new thing each month this year, I’ve been finding it really tricky – except for January – to keep to the pretty loose schedule.
And I don’t even have a job to take my time up at the moment!
As such, I though I’d try and note down every time a shiny new thing takes my interest, so that I have some idea why I’m incapable of completing a series of blog posts.
They’ll be short and sweet and will probably take the form:
Bit more detail
Instigator
— reason I looked at it.
Let’s see how it goes
My thoughts from the recent #devopsdays conference in London; notes, inspiration, todos, ahas, and OMGs.
(Since my handwriting was so poor on the first day and my name badge was illegible, this was the badge for day #2!)
Wow. There’s a lot to learn about this whole DevOps movement, but I actually feel that I can contribute to it having a pretty broad range of experience; I’m mainly development, but over the past decade or so have turned my hand to basic sysadmin-ing, DBA-ing, QA-ing, BA-ing, dev managing – almost all of the areas that need to be covered. I like to think of myself and a catalyst for implementation 🙂
I had initially been concerned that I lacked depth in skills like virtualisation and automation, but I’ve realise that those systems like Puppet, Chef, and Vagrant are potential solutions to a problem of configuration management (CM) and automation. Understanding the need for these tools is half the battle; being a specialist in them isn’t completely necessary, but an understanding to a level of basic implementation would be useful.
I’m already enrolled on a bunch of puppet webinars so will be getting stuck into that more soon.
Even though I’m pretty new to this, chatting with some key people did validate my thoughts that people shouldn’t be referring to “devops” as a person or a role in their presentations instead of an approach/framework/culture.
What happened at #devopsdays?
After being drawn in to the Basho booth by a raspberrypi/arduino/riak mars-rover-esque robot and chatting with John Clapham, who I didn’t realise was about to do the second presentation of the day (!), I end up deciding to sit right at the front (like a swot).
This turned out to be a great move as I ended up chatting with the organisers and even one of the founders of the DevOps movement, Gene Kim. Really nice guy, looks exactly like his twitter avatar (unlike some of us.. ahem..), and he even took a picture of the two of us together and let me know how to get onto the book review group for the long awaited DevOps Cookbook – bloody nice chap.
Chris Little, from BMC, is a really interesting guy to talk to as well – gave me a great overview of the background to DevOps, its “founders”, and helped me understand both its role in a company and in the future of the industry.
A pic Chris took from the stage
Try to play Where’s Wally and find lil old meeee…!
Each day there were four main presentations to start with, then some “ignite” talks.
I’m not going to try to go into the details of these presentations as 1) I’m lazy, 2) I tried to take notes on my phone and the battery died, and 3) other people have done it much better than I would have anyway!
Essentially, I found several of the presentations difficult to really grasp as I felt I was lacking a frame of reference. Perhaps this was just me not having enough of a Ops background, but I felt that the presentations could have benefited from a slide or two at the start laying the foundation for the remainder of the talk.
I was very glad to see at least one full presentation being completely non-technical, and instead focussing on the culture side of DevOps.
The afternoon of each day was put aside to OpenSpaces: everyone has the opportunity to propose a discussion topic, pop it on a post-it, people who want to vote can take a marker and add a dot on the post-its that they’re interested in. Those with the most dots get allocated a room and a time, and the discussions commence.
I initially thought this was a bit of a cop-out: I paid to attend a conference and half of it is the other people who paid talking to each other?! Rip off!
However, the discussions that I chose to attend helped me understand much more about monitoring, logging, database CI, puppet/chef/CF Engine/vagrant basics than I could have got from the main presentations, so I’m a convert.
DevOps is an exciting opportunity to technically innovate around CI, automation, delivery pipelines etc, and also to work with the business to introduce concepts like Impact Mapping.
This is where I’ll be focussing my professional efforts for the near future. I think there’s a lot of potential to help relieve stress on development teams and operations teams and relieve frustration from the business teams that define the work.
I’m committing to doing 12 months of “101”s; posts and projects themed at beginning something new (or reasonably new) to me. January was all about node development awesomeness. February was (sort of) all about Samsung Smart TV apps.
March is going to be about me playing with the raspberryPi!
The Raspberry Pi is a credit-card sized, single-board, low cost, barebones computer, intented to be mainly used within the classroom as a cheap way to teach programming.
There are more than one version of the Pi and I’ll be referring to Model B during these posts, as it’s what I have.
A bare printed circuit board (PCB) contains:
It’s pretty low-power as it’s just got:
Since the Pi runs from the OS loaded on to the SD card, then it’s up to your level of hackery skill to determine what software you’re going to run. Check out the wiki page to see what OSs are being, or have already been, ported to run.
The most popular ones are most likely those recommended by the fine people at raspberrypi.org, namely: a version of Debian (Raspbian), RISC OS, or a version of Arch Linux.
Raspbian is easy to pick up if you’ve ever played with Linux before.
Media centre
There’s a fantastic port of XBMC called Raspbmc which allows you to easily make a Media Centre PC (I’ll cover this in a later post).
Arcade machine
Play really old games on a really underpowered device. YES!
High Performance Computing (HPC) environment
Want to get some more power? Buy a truck of Lego and 64 raspberyPis and network them together in a HPC stylee!
Loads of others!
Srsly. (At least) One guy has sent his almost into space. Amazeballs.
It’s all up to your imagination and time; I have one but not the other so let’s see what I come up with this month..!
Knowledge of either Python of C++ seems to be the main way to get into working with the device; python is the recommended language for general development most likely due to it reading like a high level language and not needing compilation; just edit, save, execute.
Setting up the thing; getting an OS on it, getting a bluetooth keyboard working, setting up SSH – that sort of thing.
February was my 101 on developing for the Samsung Smart TV; a bit of a random subject n the first place and I also managed to get quite off track by the end after a hiatus in the middle.
I started with an intro to what Smart TVs are.
Second was an overview of what apps are, how they’re developed and then got into developing a basic app.
Next I did a post about deploying the apps to your tv for testing
I had intended to give a detailed article on developing these apps since I had spent a lot of time in January researching these posts and couldn’t find a decent article anywhere containing this info.
However, during the writing of my second or third post I found a well hidden but utterly perfect article covering everything I had planned to write about; my post would have ended up being a reproduction of that article which is a waste of everyone’s time and not very nice for the author of the original article!
The more useful resources are:
As such I had to think of something still related to Smart TV apps, but also interesting and different enough to be worth writing.
This is where the plan to do without the IDE came in and I tried to dissect the process and implement it manually.
I finally attempted to do without, Apache (done), generate the package (uh.. not quite), and scrap Eclipse (no dice).
What I gained from this was more headaches related to node’s async fun, and also opened up a few other avenues for future development; essentially I’ll be able to link Jan’s 101, Feb’s 101, and also March’s 101 all together!
Once I realised that Smart TV apps were just webpages, the creation of apps become kinda boring for a blog series. Deploying apps was still quite interesting, so I liked that one. The detail of creating an app was covered wonderfully in the other articles I found, so no point repeating that stuff.
A few things I discovered that weren’t really related; if you start your node server on port 80 and get a failure related to “ENV” and “process” that looks like it couldn’t access the port and you’re not sure what process is stealing that port, try [code]netstat -anbo | findstr :80[/code]
Hopefully March will be a more fruitful month – I’ll be getting stuck into a tasty slice of raspberryPi!