1.32+git194.15 tested OK: compared to the 1.32+git194.14 test above, no regression was observed but I cannot pronounce regarding the success of the index update: logs sent by e-mail. Looks good. Thank you for your great work on CLAT, also kudos to the repository manager and other testers! 
1 Like
Hi, I have followed to the letter my own memo-post here Testing CLAT for IPv6-only mobile networks - #146 by canne several times now without any problems. Testing time is about 20 minutes. Record the output of line pkcon update and send it with your logs to make sure that you are testing the right modules 1.32+git194.15.
Lines after cat <<EOF >/etc/sysconfig/connman creates connman-file and the line rm -f /etc/sysconfig/connman removes it when the testing is finished.
1 Like
Thank you! And good that it works. This may be soon-ish merged and placed into next release (when that happens I have no idea yet).
This is common behavior without VoLTE that when you have active mobile data and you receive a call → mobile data gets usually dropped as only 2G and 3G networks support audio calls with the GSM protocols, whereas with 4G calls would use VoLTE, which uses mobile data instead.
These changes made for CLAT in ConnMan do not have any effect on which generation mobile network gets selected. Ofono does selection and passes information down to ConnMan.
Do you know that those networks use CLAT or are they supporting DUAL mode (IPv4 and IPv6) or are they IPv6 only networks? And based on this, which setting you have in the Settings → Cellular network → Data access point → Protocol?
@canne has laid out a nice post of how to gather logs.
Basic idea is that in /etc/sysconfig/connman you setup which files are included in debug output. Just for the usefulness of the logs please remember to make the changes to journalctl as well so the logs are not truncated. And revert the changes after you are done with testing because the logs will eat quite significant amount of space on the system. Rest should be quite self explanatory in that aforementioned post.
Wow just discovered this thread.
I’ve had issues with tethering with SFOS for a while using EE in the UK on an Xperia 10 III.
Is it likely to be merged into SFOS?
The changes have now been merged and will be included in the next release. Schedule on the next release is unfortunately not known to me.
I like to thank all of you who helped us test this new feature. It was a really big help and without it this would not have been completed.
5 Likes
If this means on next release that tethering should work reliably then that may fix a bug that’s been present since the original Jolla phone!
Just updated to 4.5.0.21 ot looks like the update hasn’t been applied?
4.5.0.21 update installs connman 1.32+git194-1.19.2 . Don’t know what it contains but it did not work for me - I restored (upgraded) from the test repository 1.32+git194.15-1.13.1 and now everything works as before. Maybe we need to just wait a bit more.
1 Like
4.5.0.24 still installs connman 1.32+git194-1.19.2 - no CLAT update - VPN leaks (for some - see above). If you are using Jussi’s working CLAT solution, follow the procedure in Testing CLAT for IPv6-only mobile networks - #146 by canne to remove the test repository before the 4.5.0.24 update. After the update, follow the same procedure to reinstall the connman from the test repository in order to get rid off the VPN leaks if you have any. Humbly hoping that Jolla would accept Jussi’s connman PR.
2 Likes
@canne are you able to pin point Jussi’s connman PR that you’re referring to? See Pull requests · sailfishos/connman · GitHub
There’s some regression between 1.32+git194…1.32+git202 that should be sorted before promoting newer version to the public.
Cf. Testing CLAT for IPv6-only mobile networks - #168 by jlaakkonen, it says “The changes have now been merged and will be included in the next release” - so I would rather reformulate my wish 
: please integrate in the next release a connman version which contain the modifications of the PR [connman] Cleanup and fix index uses after CLAT VPN changes. Fixes JB#60897 by LaakkonenJussi · Pull Request #47 · sailfishos/connman · GitHub , thanks!
5 Likes
Hello all who’ve participated in CLAT testing before or are interested in this topic in particular. I’m very sorry that it has been a quite long wait for you but because of the ownership changes mentioned elsewhere on the forum we had a bit of downtime in general.
The CLAT VPN changes weren’t in the last release, unfortunately, and that is why I’d wish to ask your help again. Anyone willing to test how CLAT functions after some fixes done to issues that were detected in regular, i.e., non-CLAT use?
The very comprehensive instructions written by @canne in Testing CLAT for IPv6-only mobile networks - #146 by canne can be used still, just use this repository instead:
https://repo.sailfishos.org/obs/home:/jlaakkonen:/clat-new/aarch64/
It will contain a 1.32+git204 version of the most recent ConnMan that has some fixes on top of the previously mentioned git194. There was some issues that apparently during the waiting period for us has been lost from the internal tester’s minds but I did find at least one non-CLAT use related issue with mobile data:
- Right after boot and unlocking the device mobile data is a behaving a bit badly and needs a simple restart of mobile data to have IPv4 support back - I guess with CLAT this is of no issue ;). I’m thinking it is some sort of DHCP issue that I’m working on. EDIT: This seems to happen only on X10 III and only when SIM is put into slot 2.
But I’d be grateful if someone would see how this works with CLAT after some changes were made to internal index management to support the dual-index approach (per each ConnMan service, required by CLAT) better. I think that after installing this it is required to forcefully install any older version because of versioning checks, this can be done with (after disabling the repository as noted in the instructions):
pkcon --allow-downgrade <package>-<version>
or
zypper install --force <package>=<version>
After this week I’ll be back next year to check on the possible results anyone has had with this version. Same instructions on how to send the logs apply still.
Thank you all for all the help you’ve already given with this.
13 Likes
Welcome back! Meanwhile, your previous work has saved me (or at least my VPN). I will be roaming from the operator’s IPv6 country to a (sane) country with IPv4-only. Let’s see what happens!
2 Likes
Im testing from yestarday and seems my X10 III can be finally used as daily driver.
Mobile data works and suvives reboots, hotspot works. I can see 4G and mobile data icon!
OS version: 4.5.0.24
3 Likes
Does it make sense to try the changes if one did not have problems before?
I can run the packages and report if I got problems now.
1 Like
Nice to hear! And thank you for testing.
Actually I just noticed that the issue reported in my previous post happens only on X10 III and only when the SIM is put into slot 2. It might be really an issue of the older Radio Interface we use on that device, since newer might not have that issue. I need to update the post.
1 Like
Well if you want you can try the new version if you have the time and enthusiasm to help us in finding out potential issues. A wider testing will always be better, that is the version we’ve had on our internal testing and running on the development devices since summer but more testing, especially if you live in a country where CLAT is enabled will give more insight if there is something we should still fix.
You can always install the previous version back. We’ll up the version in one way or another before releasing ConnMan with the next release, so in that sense you can use this version until that happens.
2 Likes
Thank you! I’m glad it works still well. Hopefully this new one didn’t break anything but in that case I guess you have good position in evaluating if something went south with this one :).
Maybe it is too early to judge, but my impression is that since installing the latest connman version on my Xperia 10 III, my network connection is much more stable than before. This is especially true for Android apps which had complained about being offline most of the time before.
I will continue to watch the network’s behaviour in the next days, but for now I want to thank everyone who worked on fixing that issue.
1 Like