Hi Michael,
Tested out creating a new n2n openvpn connection with CU197 on both ends.
Connection was created without any issues and ping worked both ways.
Also checked the commit
initscripts: Fix process check for processes with PID file
and can confirm that now when a reboot is done with a running openvpn server
the shutdown of it is shown as OK and no longer showing a red FAIL due to the
old pid file still being present although the daemon had been stopped. So that
commit resolved that issue also.
Regards,
Adolf.
On 08/08/2025 19:13, Michael Tremer wrote:
Hello Adolf,
Thanks again for your extensive testing.
I have merged next into master and the builders should push the update into
testing tonight.
Let’s all give this another round of testing to release another solid update.
I also wrote the change log which I will probably post on Monday:
https://www.ipfire.org/blog/ipfire-2-29-core-update-197-is-available-for-testing
Have a nice weekend!
-Michael
On 8 Aug 2025, at 14:17, Adolf Belka <adolf.be...@ipfire.org> wrote:
Hi Michael,
Good news.
On 04/08/2025 17:42, Michael Tremer wrote:
Hello Adolf,
Thank you for doing all this testing and apologies for replying so late.
On 25 Jul 2025, at 16:52, Adolf Belka <adolf.be...@ipfire.org> wrote:
Hi All,
So the openvpn-2.6 issues with the update process from 196 to 197 have been
solved by the last patches that were applied.
So now I have been evaluating existing connections and new connections.
There is good and bad news but more good than bad so progress is being made.
First thing was to test out my existing rw and n2n connections with CU197.
The existing rw connection from my Linux laptop worked without any issues and
could ping a machine on the green network.
The existing rw connection from my Android phone showed it was connected but
ping to a machine on the green network no longer gave any response except 100%
packet loss.
Based on a suggestion from @Michael I tried to connect to the IPFire WUI using
the IP for the IPFire green interface.
It worked. I was able to login and check through WUI pages.
So the connection is definitely working but for some reason the ping command no
longer works to a machine on the green network although it works with a CU196
system with the same client connection settings. However ping on the laptop
works for both the CU196 and CU197 versions with the same client connection.
Good to know that this has worked!
I don't think this is a breaking issue just a puzzling one.
Mostly seems to be a client issue.
Yes.
I then tested out the existing n2n connection between a CU197 system at one end
and a CU196 system at the other. Connection worked and ping worked in both
directions.
Then I created a new completely new client connection for the Linux Laptop.
Connected via it and the connection was made successfully. I think I tried the
ping and it worked but I am not 100% certain so I will do the test of creating
a new connection again as I deleted the old rw connections when I was testing
out the n2n connections.
I will also do a test of a new client connection with my android phone.
I then created a new n2n connection between a CU197 acting as the server and a
CU196 acting as the client.
Where did you create the connection and where did you import it? I created a
connection on c197 and them imported it on c197 again and it seemed to have
worked.
This gave a peculiar result in the WUI as it provided two lines for the
connection. I attach a screenshot of this as it is a bit difficult to explain.
Hopefully the screenshot is accessible. If not let me know and I will put it in
the paste system.
Yes, I could see the screenshot. Something went wrong with creating the
configuration line in the CSV file. Once there is something off, a lot seems to
break at the same time.
I repeated making a new n2n connection, both with an empty set of client
connections and also with existing rw & n2n connection in place.
I could not reproduce the issue with getting that double line shown in my
screenshot. It looks like there was some hiccup or something when I did my
initial test.
Have now created three new n2n connections and in all three cases the entry
went as expected.
Additionally in all three cases I was able to ping in both directions with no
problems.
I then deleted the line that had (Expired) in the Name column and enabled the
other line.
The connection then showed up as connected at both ends but doing a ping in
either direction gave 100% packet loss, whereas the version with CU196 at both
ends gave a good ping result in both directions.
I then reviewed the n2n logs for one end of the tunnel between the ping working
version and the ping not working version.
Basically the contents were the same, resulting in an "Initialization Sequence
Completed" message, so it looked like it was fully working.
So I then tried accessing from one end of the tunnel the WUI of the other end
via the IP URL. That worked. I could successfully log into the WUI of the far
end of the tunnel.
Okay, this sounds all very good. But I think we need to probably invest a lot
of time to bring N2N to a good standard and I currently don’t have the
resources for this. I am not sure if we broke things in this changeset of if
they had been broken for a long time already.
So with 2.6 at one end of the tunnel and 2.5 at the other a new n2n connection
is working in terms of actual data traffic, except for the ping traffic not
seeming to work and the creation of an additional line in the Connection Status
and -Control table of the OpenVPN WUI page.
I will also try and find some time to test out a new n2n installation with 2.6
at both ends.
Perfect!
Can we use the matrix on the wiki page to show what is working well and what
isn’t?
https://www.ipfire.org/docs/roadmap/openvpn-26
We might want to create a second table for N2N.
I will see what I can do.
So most critical things seem to be working but there are a couple of puzzling
things to be dealt with.
Cool. Hopefully we should be able to get it all done very soon!
So currently I am able to confirm that existing n2n connections and newly
created n2n connections with CU197 at the end creating the new connection and
CU196 at the end receiving it are all working as previously with CU196 at both
ends.
Regards,
Adolf
Best,
-Michael
Regards,
Adolf.
<Screenshot_2025-07-25_14-12-40.png>
