as we want to give you some summary of what is going on, we are releasing another developer update so you know what we were up to recently. This time it covers what we were up to in May. Format of this one will be slightly different from the previous one and we are definitely interested in your opinion which of the formats you liked more and whether this effort make sense at all.
Now, let’s get back to what has happened last month in various projects.
As you know, at the beginning of May, we released another big release - 3.10 It was in RC phase for quite some time as it brought few new functions and a lot of changes underneath. When we released it, we started getting feedback and looking into issues we missed. Based on that we started preparing 3.10.1 with various fixes.
Parallel to that we were continuing our efforts on 4.0 mainly towards being able to upgrade from current release to 4.0 on old Turris 1.X routers. It’s not that easy as we are facing few issues regarding OpenWRT way of packaging core system component (they are hiding their presence from package manager as opkg
could easily break the system while trying to update them) and we are going to migrate to different libc - from uClibc to Musl. And there is quite some changes in both how our upstream distribution behaves nowadays and how do we want to do our downstream distribution (we are doing quite some cleanup as part of this effort).
Why are we focusing on old Turris migration? Well, old Turris is much simpler and there is not so many kernel patches that we need to take care of. On the other hand, with Omnia we would need our kernel guys who are currently working around the clock on MOX. And the trickiest part will be the seamless userspace migration anyway which we can test and on either of those routers. Once we get everything sorted out, it will apply to Omnia as well and they’ll just add few kernel bits and it will work.
You couldn’t miss that we are launching a new Turris - Turris MOX. Campaign was running through whole May and although campaign at this stage is mainly about marketing and PR (and HW behind the scenes), we are already working on software for MOX and doing some tests to validate what is possible. Currently we have nightly builds of Turris OS 4.0 for MOX based on current OpenWRT master that contains all the basic stuff including the kernel.
You might recall this project from the last news or from various talks. It’s purpose is to replace aging uCollect with new better (in terms of performance, stability and maintainability) approach. It will still collect data from
routers and provide dynamic firewall, but hopefully with bigger insight and even faster updates.
As this project is already in progress for some time, to test it out we started deploying some parts of the chain and putting various components together and doing some test runs. We are already getting data from HaaS, evaluating them and constructing new ban list. All deployment is Ansibled, so we can easily deploy it later in production once we are satisfied with results of our tests.
Apart from testing we continued work on other modules so it would be eventually feature complete and ready to replace uCollect. Big pile of work went into authentication of the routers and we prepared the modules that will act as minipots in old uCollect. Altogether project is progressing nicely and is nearing a state when we could start some public testing.
Cleanup of Foris internals still continues. Latest efforts goes into killing first run configuration wizard. Don’t worry, we still want something to guide new users through the initial setup but nowadays it is a separate application which makes it harder to maintain. What is being worked on is the guide that will guide you through the initial setup but directly in Foris configuration interface. So if you would need to reconfigure something later, you will face the same dialog as during the initial setup.
As we released version 3.10 and people started playing with Pakon even more, they started finding some small issues that we haven’t noticed during testing. Mostly concerning minor browsers compatibility and some corner cases.
Similarly to Pakon, as it was first time released, we gathered some feedback and started polishing it a little bit for example with notifications if something goes wrong during the backup.