After extremely stable October with 100% uptime, we ended November with our third worst uptime ever — 99.95%. Actually, we have four separate downtime reasons: the first was a network failure in NL-IX, two were caused by our storage system part of tests, and the fourth one was due to the installation of the second rack in the data center (expanding flespi network and introducing a new additional layer of switches connecting racks). The network failure was outside the scope of our responsibility, so we could do nothing. However, with the storage system, we are now much more stable and prepared than a month ago. In the first case we changed the database schema with stopped mdbctl service and, due to triggers, it took too much time. In the second case, we performed controlled deletion and cleanup of Gurtam Space test account with 80K devices and 9TB of storage. The most difficult was the installation of the new rack, but now that it is installed, we can add new racks to flespi LAN with no noticeable impact. Details about all four cases including technical explanation and downtime duration can be found in our NOC.
- The most visible change on our website is the introduction of our knowledge base. We started with 25 fresh articles on the most popular topics among flespi users and will add more in the upcoming months. We want to give credit our software engineer Jan Bartnitsky who created most of them. Comparing to blog with its sometimes outdated articles we will keep knowledge base in the actual state at all time.
- Our special open-source tools like TrackIt!, Toolbox, and MQTT Board finally received their own landing pages with the accumulated set of information and especially important — actual video reviews about each tool. If you are interested in more flespi videos — welcome to our youtube channel.
- Some new usability tips for the users of the flespi panel. First of all, it is now possible to automate the creation of devices connected to the channel directly from its ‘Connections’ tab. Second, we have added a special button to open specific elements of the flespi panel in the new browser tab — e.g. when working with devices you usually need to navigate between toolbox and device settings dialogue all the time. The same applies to almost every window with any type of flespi object. And finally — for those who operate multiple flespi accounts — we introduced colour schemes so that you can visually determine in which account you are at the moment.
- Subaccounts feature for large projects received own part of updates. First of all, now every flespi object has ‘cid’ field with account ID it belongs to. Second, we created special hierarchical access control, so that in terms of REST and MQTT operations parent account has items from child account appearing in its list. At the same time from telematics perspective it is possible to create a channel in parent account and, if allowed, devices from child account will have access to it. If it sounds too complicated just be sure that all operations with subaccounts are quite organic and logical from telematics use cases point of view.
- We integrated one new protocol — omnicomm. This is to connect Omnicomm devices directly to flespi instead of earlier integrated omnicomm-rpc protocol which uses API to receive messages from Omnicomm servers.
- Going further in terms of IoT and carriers connectivity we have introduced the possibility to specify ICCID instead of a phone number in device properties. ICCID by itself is printed on the SIM card and is used by carriers for unique SIM identification. Together with this, we have integrated with Jasper using its REST API. So now it is possible for those carriers who are using Jasper service to directly deliver configuration SMS messages from flespi.
- For those who are using flespi storage system directly, we have a few updates as well. First — we have changed container’s TTL value meaning: -1 means no cleanup by TTL(default) and any value greater or equal to zero will automatically clean up container’s messages supposing it’s key is UNIX timestamp in uint32_t format (number of seconds since January 1st, 1970 UTC). And the second update — to request message from a container it is possible to specify not only AND type of filter (with a comma), but also OR: e.g. “x=4|x=9|x=99”. This also applies to requesting messages from devices. Read more details on the format in the API documentation.
- We have updated the platform status page and now the last minute stats shows not only the number of telematics messages we received (e.g. sent by telematics devices) but also the number of REST API calls we received and the number of MQTT messages published/subscribed by flespi user connections. We plan to add even more information to this — like the volume of API traffic, channels traffic, streams traffic and so on. Just do not forget to bookmark this page as if you are having any kind of problems this is your starting point to check how the platform is running at the moment. And just for the history — right now all day long we are handling approximately 3 million of MQTT messages/minute. Thanks to the high volume architecture implemented in the flespi MQTT Broker we do not notice this load, while most other MQTT brokers are very proud when reaching such numbers in their benchmarks.
In December we plan to enjoy the first snow (which we had in Minsk exactly on December 1). We will add some new features to the platform, of course. Many long-running projects are about to finalize but you never know when a new gnome will appear somewhere that’s why I prefer not to market it now. Actually, we are very close to the stage of stacking multiple channels together making it possible to use normal telematics devices as a transport for the data from other devices communicating over own protocol. This will be an amazing feature opening totally new and wide areas of application, so let’s cross the fingers and wait.
We are wishing you a very Merry Christmas and a Happy New Year. You will see the next change log in 2019. Enjoy the holidays and stay warm!