Posts tagged with "Work"

DevOps at Kobas

I’ve been at Kobas two years now, I previously wrote about my experiences in my first month , so it seemed fitting to do an overview of my experiences since then surrounding the software we use for DevOps.

Jenkins & Continuous Delivery

Jenkins Image At Kobas we use continuous delivery rather than continuous deployment, tool of choice; old man Jenkins .

Jenkins, although riddled with UI/UX issues, is a very helpful tool for us day to day now. We have all sorts of pipelines setup for deploying to our EPoS & Cloud servers. Having the ‘one click’ ability to release to a QA or Production environment and letting Jenkins handle all the steps that go on throughout that build process is a time-saver.

Having the ability to rollback to a previous git tag when a bug has been introduced proves to be a real lifesaver. I can’t imagine going back to releasing projects via script or knowledge.

We keep our Jenkins configuration backed up in Git as the thought of losing all the work put into Jenkins and having to start again is nightmare inducing.

Codeception & Jenkins

My latest win, and one that took the longest to achieve was having Codeception and Jenkins play nicely together for automated testing.

The initial part of that went fine, just getting Codeception to run automatically via the Jenkins build process. But then I decided I wanted metrics like code coverage, the ability to run acceptance tests via Selenium & being able to reset our testing database before each test runs.

Selenium is something I’ve played with a lot on my own machine, so setting up Selenium Grid on a server didn’t cause me much trouble.

Code coverage however has been a pain. Needing c3.php to get remote code coverage working required a number of disgusting hacks / workarounds. This was due to the way our project is setup, the main directory Codeception is in, isn’t even synced to our servers.

The results where worth the pain though I now have the ability of viewing a breakdown of code coverage like this:

Code coverage report

(clearly not an image from our Jenkins, imagine a lot more red).

You also get a dashboard for coverage distribution and showing you the files with the most CRAP (change risk anti-patterns), my new favourite acronym. I can’t seem to find an example of that view online however.

Perhaps later I’ll do a blog post on how to set this up on an open source project and will link to it from here. (Making plans for 2020 already)

Puppet & Server Config

Controlling configuration manually might seem okay when you only have a handful of servers, I certainly manage my personal servers manually still.

At Kobas we have two types of servers, our ‘Cloud’ servers (hosted by AWS) and our ‘EPoS’ servers (hosted by our clients on-site). Currently, we have 4-5 AWS servers, and 170+ EPoS servers. Managing configuration change across the EPoS servers would be impossible without Puppet.

Puppet isn’t the only tool that handles configuration management, there are a number of tools to choose between; Chef and Ansible are two that come to mind. I’ve only used Puppet so can’t comment on the advantages of one over the other. I do recommend getting at least one of these tools setup to manage your server configuration though as the pay-off is huge.

Using a configuration management tool ensures that all your configuration is the same across every environment; development, qa and production. Significantly reducing the ‘But it works on my machine’ issue. It also requires you to put additional thought into making configuration changes.


Posted on September 23, 2018

My thoughts on management & time tracking

manager-time.jpg So I’ve talked about time tracking before , however that was much more a “Keep yourself on track / how to know what to bill per hour as a freelancer/contractor”. Today I’d like to share some thoughts on bigger companies and my feelings toward management and time tracking there.

I’ll preface all this with ”these are my opinions, I’m not saying I’m entirely correct here, this is just how I personally feel“.

As a programmer I love my craft, I spend hours playing around with new technologies, learning new languages and wrapping my head around computer science concepts. I’ve always known programming would be my career from a young age, finding it amazing that people would pay me to do what I love to do anyway.

What I never realised back then is how much of a programming role involves no programming at all. Meetings can be a daily occurrence, eating into your time, understandable though as things do need to be decided on/signed off. Managers need you to explain all sorts of stuff to them so that they can sound informed to whoever it is that they report to. Clients want you to explain why you can’t add five new features by next week. Customers want you to explain how to use things (and occasionally need you to fix things).. Priorities must be juggled.

All of the above I don’t really have an issue with, they are mildly annoying but to be fair, necessary evils.

What I do have an issue with however is time tracking on top of this in large organisations, usually so your manager can have some form of chart showing what their team have spent their time on the last few weeks. Of course this would all be possible without disturbing the programmers workflow at all (since we all have issue management systems) but the word “granularity” starts getting thrown around and the next thing you know you’re now in JIRA logging hours on tickets trying to justify where you spend every minute of your day.

I honestly don’t get it.

I feel as a developer getting told to do this makes you feel that you are not trusted to manage what little time you have to do programming yourself. That you’re possibly under performing and need to work harder. Or that your time isn’t as valuable as the managers time. All in the name of a granular report that probably gets a courtesy glance at and then binned. It definitely doesn’t create a happy team environment.

If you haven’t read Programmer Interrupted , I recommend that you do, but I’ll just include the results of their study here:

  • A programmer takes 10-15 minutes to start editing code after resuming work from an interruption.
  • When interrupted during an edit of a method, a programmer resumed work in less than a minute only 10 percent of the time.
  • A programmer is likely to get just one uninterrupted two-hour session in a day.

These are the problems that management should be trying to solve, not trying to interrupt a programmers time further by requiring that they log each and every hour that they’ve spent on each feature.

Another great article that discusses how programmers see time different to managers is the Makers Schedule.

Where I currently work (Kobas) actually deals with all of these things I’m complaining about very well. For a start there is no time tracking (making me very happy), we have a daily slack stand-up (i guess its more sit down?) where we give a few sentences of what we worked on yesterday and what we plan to do today. It’s very helpful for knowing what other people are up to without wasting time doing an actual stand-up.

Interruption wise at any point i feel I need to have an uninterrupted session I can pop in my headphones and unless something explodes I never get interrupted. Meetings for me are rare but when they occur they have an actual purpose. Developers != Support leading to less interruptions, of course I have to fix problems that come up from support but I’m not personally being interrupted by support requests.

So its certainly possible for companies to function well in these ways, why more big companies are not is beyond me.


Posted on August 29, 2016

First month at Kobas

Image of Kobas team meeting

So I’ve been working for roughly a month at Kobas now, I think things have been going very well and I wanted to highlight what I’ve been working on for the last couple of weeks.

For anyone that does not know what Kobas does, it is a hospitality management solution covering; stock control, rotas, HR, EPoS, customer loyalty and much more. It’s actually a very useful piece of software for clients, allowing them to gather lots of data from all areas of their business and providing a cloud interface that outputs that data in fancy reports. After seeing it in action I find it very surprising that not all businesses use this as it can really help you to increase your profits and avoid wasting money unnecessarily.

Anyway, on to what I’ve been doing, I’ve been mainly working on the EPoS (Electronic Point of Sale) side of things, which in layman’s terms are the Kobas tills.

To side-track just a little I think the Kobas tills are so nice to look at and use, here is an example screen from one:

Image of a Kobas EpoS

When you take that in comparison to a result from searching EPoS on Google Images (and what most places use):

Image of a generic EPoS

Bit of a difference there right. So unfortunately now I have been burdened with the curse of noticing every single EPoS system every place I go and thinking to myself, “how do they use this?“.

Anyway back to what I’m doing, basically I’ve been working on adding the functionality to accept deposits and other payment types to the EPoS. The EPoS accepted cash, card and voucher when I arrived, which was all you would really need basically.

But now with Christmas getting closer venues are going to be taking deposits for bookings and we wanted to be able to handle that within the EPoS itself.

Also with the rise of services like ‘Just Eat’ etc, venues are trying to figure out how to process payments from those services, as it’s not really a cash payment as you don’t have the money in your till and it’s not really a card payments as you haven’t put it through your card machine.

So with that in mind we also decided to create “Other payment types” which allow businesses to just tell us what other payments they want to accept and we are able to quickly add that functionality to the till and have it display in all relevant reports.

Our product manager Daisy Lang has wrote about it in much more detail over here.

Adding the ability to accept deposits and other payment types went well but I did encounter a few difficulties while doing it, for a starter constantly worrying about breaking the tax calculations (and having forgot how UK tax works).

Luckily I found a very edge case unit test wrote by Neil Mukerji (our CTO) and after converting it over to use the new version of the payment objects I was delighted to see it was still passing as expected.

Naturally I then decided to write a bunch more unit & functional tests, I’ve been on the Codeception train lately, after getting introduced to it at the end of my role at UBC and I’m determined to get it set up properly in Kobas so that moving forward refactoring and changing code is much easier (and less stressful). I’m hoping to get all our tests into Codeception shortly and add them into to the Jenkins build for automatic testing on deployment.

Also shockingly (due to not having it in my other roles) there is a whole QA team at Kobas, which has been a total life-saver for me. Actually having someone QA features you’ve added is unbelievably helpful as when you’ve been working on something for so long it’s easy to miss things.

Anyway that’s all I have to talk about for now, I know what I’m doing in the coming months but I’m not going to mention that here, but stay tuned, interesting stuff is coming!


Posted on July 03, 2016