November in Belarus is always wet, cold and foggy. This year was not an exception and most of the time flespi team spent preparing the platform to be spread across multiple data centers, improving the performance of its internal components, and helping our users reach their goals.
Actually from the support point of view, we noticed a rapid growth in the number of people trying flespi and basing their solutions on flespi. We experienced approximately a 40-percent growth in everything countable within one month. And tremendous growth of product usage is something that inspires engineers most of all. We like our users and we like helping them and seeing immediate results. That's why flespi support is provided by developers, not by a dedicated support team.
We were able to upgrade all our servers to the latest stable version of Debian OS including gateways with zero downtime in both data centers and this is a great achievement of the expertise and technologies gathered by our team. However, the result of a month-long development affecting the flespi database storage format and internal data exchange protocol caused the unavailability of telematics database storage controllers on three occasions (414 seconds in total) during a controlled database mirrors synchronization. Although all flespi subsystems except telematics database storage operated correctly during that incident, our clever bots detected this situation and registered downtimes. We investigated the reasons immediately and changed the implementation of storage controllers to prevent similar situations in the future. We were very close to 100% uptime in November, but due to these database modifications our monthly uptime fell down to 99.984% and we were unable to keep the 4-nines this month.
- As a special reading, I recommend checking this interesting case from one of the flespi users in Singapore about the edge of IoT and vehicle telematics. For people with a technical background, I propose a high-level overview of report engines used in modern software architectures and an article on how flespi can act as middleware capable of postprocessing and data modifications between the tracking device and tracking platform.
- At the same time, another long-running project aimed at updating the heart of flespi telematics hub — its protocol parsing engine called pvmII — is now being actively deployed. We switched wialon-retranslator protocol to the new engine and already integrated the new albatross protocol using it. Basically, all new protocol integrations for flespi will be based on the new version of the PVM engine.
- We integrated automatic GPS rollover protection for a few protocols to remove the requirement for those who are using flespi to upgrade devices firmware, especially when this is almost impossible for outdated and old devices and their GPS chipsets.
- We upgraded Scania FMS protocol to a new version of API and released dedicated instructions on how to connect Scania trucks equipped with the Scania FMS system to 3rd party telematics platforms.
- In November our frontend team was working hard to stay in the same place — most of our applications are based on Vue and Quasar framework and they had to upgrade to its stable version as the previous Quasar version will become unavailable in January 2020. The problem was that there were so many changes between the versions that final diffs for the flespi.io, for example, were more than 20K lines of code. It was a very long and hard work and as you may notice on this screenshot from our status page it was a very rare case when our frontend was not updated for more than three weeks and was far behind backend in update frequency. An extremely rare case. Now toolbox is running on the new Quasar version, flespi.io panel is going to be upgraded this week, and we will continue upgrading other frontend components to keep everything in sync.
In December we plan to start the whole set of flespi services in the dedicated flespi installation in Russia and in January we plan to open it for our users. Immediately after the first regional migrations, we will be able to finally concentrate on the new features as we have a huge todo list of exciting things that will help our users solve a variety of tasks and increase the competitiveness of their software.