Update kernel in Turris OS 3.x series

I want to see any vendor that has sufficient resources to satisfy all of its customers for hundred percent. It is like simulating whole universe in all details in real time, not possible unless simulation is run on whole universe. Be reasonable. I know that you are unhappy but let me tell you, you are minority in your requests and we invest resources relative to amount of users. We will get around to fix what burns you but it won’t be immediately.

It is not solved. We can reproduce it with some setups that is why that issue is open.

This makes no sense. I really don’t want to loose time with you if you present arguments like this. I wrote “backported” so arguing that it is moot because TOS 4.0 is released shows clearly misunderstanding.

You seems to forget that you are not the only customer in the world.

I am out of this discussion. It makes no sense to burn trough my time here.

3 Likes

does not seem unreasonable to expect

being done by now but of course if it requires years to complete[3]

23 of 31 tasks completed

than it might be perceived unreasonable indeed. Happens when there is no roadmap for development cycles.


Such is not mentioned in the issue tracker, only that the attempt to revert upstream’s patches for TOS4.x did not pan out.


Of course not but apparently just one concerned that resources going to waste with TOS3.x, with the majority of the user base otherwise being seemingly content to run their nodes on TOS3.x forever.


[3] https://gitlab.labs.nic.cz/turris/turris-os-packages/issues/344

Just for curiosity sake, could you (i.e. somebody from team) publish some statistical info? Like approximate number of routers of different types, TOS level use, active forum users, etc… to give us rough idea about Turris routers users?

1 Like

I am blue Turris 1 user from the beginning and since past two years I am also not much happy with slow pace of innovation even team doing what is in their power. As I would like to innovate too and I simply grown up on my turris (I learned a lot about Linux and thanks for that!) I made uneasy decision that sooner or later it will be necessary to decommision my Turris 1 as it had been already said here in the forum in other thread.

For now I am in the middle of transition custom compiled OpenWrt MASTER kernel from end of January and have to confess it is NOT definitely way for everyoine as to have new version of kernel, kmods and packages is not always good thing as there could be some error and something eventually will stop to function here or there which leads sometimes to hopless situation when I don’t know what to do. I also got lot of things and services configured on my current turris and working on manual migration to new device which is not an easy task. My approach to upgrades now is to have 2 SSD disk one SATA and one mSATA and to be able boot from both via Grub2 from debian buster partition and have two compiled master snapshots every half or once year and move configurations files between them so I would be able to switch to newer master snapshot more easily as for now I did not find the way to update kernel, base files and kmods other ways - task done by updater developed by NIC team. Two separate disks with multiple partitions will also allow me to compile openwrt packages directly on router in debian buster distribution. I also consider to stay at 19.07.1 if possible as described here https://hamy.io/post/0015/how-to-compile-openwrt-and-still-use-the-official-repository/ but for now all my kernel compilation tests show it may not be possible to enable CGROUPS for LXC and stay on same vermagic to be able to install kmods from openwrt internet repository.

I would recommend to @anon50890781 to go and buy suitable Mini-ITX board with at least 4 core CPU with AES NI support and enough of M.2 (for 802.11ax card) and minipcie (for 802.11ac) and msata sockets and also 2 gigabit LAN ports and then go and compile his own openwrt master snapshot which will allow him to run whatever he want without any limits, Yes it will be costly as it could easily cost as much as 8000 CZK or more but once it will be up and running it would bring a peace of mind and frustration from slow inovation will go away. You will also have a great toy ! From recent @cynerd comments it seems that it could take some time to get on final release of TOS 4 and TOS 5 is even more far away in future.

Needless to say there WILL BE substantial time investment in learning linux skills but I clearly see it as good thing.

The path is more less straight forward:

  1. install recent linux distribution - i go for debian as I see it supported the most
  2. install development packages - tools, gcc, libraries etc.
  3. clone master from git
  4. deploy binaries to shiny new router device
  5. install all othe opkg’s
  6. migrate or create configuration for desired services
  7. install LXC containers and configure desired services in LXC
  8. install docker and configure desired docker images

Once I got my openwrt master device ready to deploy for production I would probably have to reserve some weekend and buy new SD CARD to make a new installation of TOS 4 on my blue Turris and then reconfigure everything as backup router and keep using it as dumb access point for forseable future. If migration script will be developed within forseable months it would be great but personally I probably not wait for that and do fresh installation of TOS 4 as import from medkit.

2 Likes

I am closing this thread. The last 4 posts are just commits and nothing else. There is no reason to post each commit, which I did in the nightly branch as it won’t be included in the stable release yet. If there’s going to be a kernel update in Turris OS 3.x, we will let you know.

1 Like