Known bugs in Turris OS 3.9 and recommended solutions

Dear Turris users,

we have registered a couple of bugs, which have come up with the release of Turris OS 3.9. They should be marginal problems and only affect a small number of users. We are working on solving the bugs as quickly as possible and want to apologize for any inconvenience they might cause you:

  • if you have any trouble with Updater, please try to first reboot the router.
  • Updater fails on packages, which users have created themselves. A fix will be released in Turris OS 3.9.1.
  • An error in Foris, which causes the SSH Honeypot option to be Unchecked. Please check that you have this window checked (if you want it to be), the solution is to check the option again.

Thanks for your understanding and happy routing,

Team Turris

3 Likes
  • Foris may be giving 404 until browser’s cache for pre-3.9 Foris expires.
  • 5G wi-fi missing (in Foris at least) until you make yet another restart
1 Like

In kernel log every 10 seconds there are such entries

[98425.631182] unregister_netdevice: waiting for lo to become free. Usage count = 1
[98435.700540] unregister_netdevice: waiting for lo to become free. Usage count = 1
[98445.810059] unregister_netdevice: waiting for lo to become free. Usage count = 1
[98456.049069] unregister_netdevice: waiting for lo to become free. Usage count = 1
[98466.168414] unregister_netdevice: waiting for lo to become free. Usage count = 1
[98476.207697] unregister_netdevice: waiting for lo to become free. Usage count = 1
[98486.336972] unregister_netdevice: waiting for lo to become free. Usage count = 1

Window in browser Opera and Chrome - window Luci-Processes-free space left and right side of lines it outside the browser window

<img

This is a bug in Luci. We don’t consider that to be important enough to fix it on our side. If you wish, report that to the maintainers of Luci.

But thanks for all your effort… :slight_smile:

Same as @prezes_kk since OS 3.9 update and the Load Average increase up to 6.0 in Luci status.

2017-12-14T15:34:06+01:00 emerg kernel[]: [30807.660633] unregister_netdevice: waiting for lo to become free. Usage count = 1
2017-12-14T15:34:16+01:00 emerg kernel[]: [30817.840659] unregister_netdevice: waiting for lo to become free. Usage count = 1
2017-12-14T15:34:26+01:00 emerg kernel[]: [30828.020635] unregister_netdevice: waiting for lo to become free. Usage count = 1
2017-12-14T15:34:37+01:00 emerg kernel[]: [30838.250634] unregister_netdevice: waiting for lo to become free. Usage count = 1
2017-12-14T15:34:47+01:00 emerg kernel[]: [30848.440637] unregister_netdevice: waiting for lo to become free. Usage count = 1
2017-12-14T15:34:57+01:00 emerg kernel[]: [30858.640633] unregister_netdevice: waiting for lo to become free. Usage count = 1

Having a few issues since updating:

  • Can not connect to vsftpd - not really sure what’s going on here, can’t find anything useful in the logs.
  • nano exits immediately with “received signal SIGTERM or SIGHUP”

Tried a few things like restarting the vsftpd service, completely reinstalling it. Finally did a rollback to the pre-update snapshot and it works now. So it definitely feels like it’s something with 3.9.

Let me know if you need any more info.

The same here - High load and “unregister_netdevice: waiting for lo to…” in kernel log.
Load before update <1 and after update load is >3, but I think that it is related to LXC containers - I have 2 lxc containers with Debian autostarting - if I autostart only 1 container, load is only about 2 and if I want to stop lxc container it hangs and “unregister_netdevice: waiting for lo to…” starts to showing in kernel log.

Now I have changed lxc-auto config not to start any of the containers a I have load 0 and no “unregister_netdevice: waiting for lo to …” in kernel log

1 Like

I would test with disable autostart lxc conteiners but propably is the issue with lxc.
In luci can’t stop/start lxc conteiner.
Maybe version 3.9.1 it will bring changes

You have to change /etc/config/lxc-auto file using ssh not to autostart containers … In Luci it freezes …

EDIT: The android app shows 100% load. I don’t think it is real.
Top commands shows 99% idle.

Mem: 291076K used, 1779676K free, 3508K shrd, 27308K buff, 100296K cached
CPU:  11% usr   3% sys   0% nic  84% idle   0% io   0% irq   0% sirq
Load average: 1.03 1.06 1.01 2/226 26767

I wasn’t reading top output correctly.

OLD MESSAGE:
My CPU load is now always 100%. I don’t have containers.
Using top command, the load is cause by “nice” processes.
Not sure how to identify them.

1 Like

I noticed this as well.

Try ps w and take a look what are the arguments to nice - nice shouldn’t do anything by itself, it should run something else and that something else is specified via arguments.