Profile cover photo
Profile photo
Richard Vowles
613 followers -
Software Developer and Mentor
Software Developer and Mentor

613 followers
About
Richard's posts

I had a couple of hours tonight to try and get the esp8266 nodemcu stack working in a VM - my advice after that wasted time - don't bother, use the docker image.

I'm working on making changes to the enduser_setup module, so it can take extra query parameters and write them to a file. That means at setup time, you can gather device specific information (such as which room the device is in, or which window) without jumpers or flashing the module specifically.

Once I have done that, its on to OTA support for my modules.

I decided that it was time to investigate the Compute Engine and so decided to take my website - timetable ninja - and investigate what it would cost to run on Google's infrastructure.

My AWS box is a shared 1.7Gb memory box - it costs me $US50 per month. The same box on GCE costs $US17 per month.

I figured I could do better, I only need nginx + a java web server - so I dropped to a 600mb shared, for a $5 per month, and a separate Cloud SQL instance (again 600mb) for $7 per month.

I could probably have gone cheaper going to Digital River or similar, but I really wanted to try out GCE.

It took me about an hour to get the runit, ssl keys, nginx config (i was using apache2 on aws), database imported, etc.


Post has attachment
Its interesting what +Kevin Marks says in this article. Having full traceability across the stack is something I insisted we needed on the last project, it really makes a difference in terms of discovering where problems are.

That carried into our current mixed Java/Node stack on Kubernetes that uses REST, and we recently added it to all of the SOAP headers, cross app event bus and Message Q meta-data. When e2e tests fail, whatever the cause, we get a full trace of where they went, what they did and where it failed - which is pretty useful.

One last thing I have laboured this last week on is full between service dialog logging - why? Because it allows you to create an automatic sequence diagram of exactly what happens across your microservice calls for easy documentation. And of course it inherits all the existing traceability - response times, etc.

https://www.datawire.io/microservices-observability-groundhog-day-microservices-com-practitioner-summit/


Post has attachment

Post has attachment
Little packets are rolling in
Photo

Post has attachment
Sure you don't want to come down for this +Roger Parkinson​ ?

Post has shared content
In case you haven't seen it yet
Become a Beta Tester for Google+

If you’re a user who’s as passionate for Google+ as we are, apply to become a beta tester for our product. We’re looking for people who want to test out new features, as well as discuss with our product team and fellow beta testers how to make Google+ the place to share your interests. Want to know what it takes?

Here’s what we’re looking for:
● Active poster on Google+
● Motivated to learn more about Google+ and contribute to product discussion
● Eagerness to provide high level feedback on our product

If you’re interested in applying, fill out the application form. Thanks!

https://docs.google.com/forms/d/e/1FAIpQLSf1RQl1bWCtk8iKgnjhkAjenK2nNcHFegC7RfdK5yYpIm-frg/viewform
Photo

Post has attachment
Door handle broke off so my wife applied a quick patch!
Photo

Post has attachment
When I am testing the profile next, I better do it outside.

(Just adding blobs of paste to see at what point it melts, in this case it needed to stay for around 1 minute at the 250 degree temperature).

Photo

Post has attachment
+SAP - moving light slow lightning. Very slow lightning. Very. slow.


Photo
Wait while more posts are being loaded