Here's the summary of the IRC meeting.



Place: #openvpn-meeting on irc.freenode.net
Date: Wednesday 30th October 2019
Time: 11:30 CET (10:30 UTC)

Planned meeting topics for this meeting were here:


Your local meeting time is easy to check from services such as



cron2, lev, mattock and rozmansi participated in this meeting.


Talked about the OpenVPN 2.4.8. Cron2 will tag it ~13:00 CET after which
mattock will start the release machinery. The release process will
probably be 100% done tomorrow.

Merged the final pending tap-windows6 PRs. This means that the installer
bundled with OpenVPN 2.4.8 will be the same as was released earlier as a
test installer.

Decided to postpone merging the NCP/fragment patch as cron2 has had
issues reproducing the problem. [Earlier today decided to bundle OpenSSL
1.1.0 and upgrade to 1.1.1 in 2.4.9 which is due at / after hackathon]


Mattock will probably make the Windows "buildslave" a real buildslave,
as in: "connected to the buildmaster". And do the same with the Debian
package builder (sbuild_wrapper) instead of using home-grown Git
polling, locking, etc. routines. This should not require major
refactoring of the current Buildmaster configuration.


Noted that pushing to SF.net OpenVPN "master" branch is broken. Cron2's
support request was declined. Mattock will open an official one to get
SF.net support's attention.


Lev sent a small patch to enable Travis-CI for Visual Studio builds.
He'd like it to get review so that we don't accidentally break those builds.


Noted that we should do some brainstorming:

- t_client tests: figure out which branch we are in (they fail on 2.4)
- resolvctl: do we want to bundle scripts for it (e.g. for arch linux)


Noted that (due to hackathon) there won't be a meeting next week.


Full chatlog attached.

(12:32:03) cron2: heelow
(12:32:06) mattock: hello
(12:32:15) lev__: guten tag
(12:32:23) rozmansi: hi
(12:33:19) mattock: https://community.openvpn.net/openvpn/wiki/Topics-2019-10-30
(12:33:20) vpnHelper: Title: Topics-2019-10-30 – OpenVPN Community (at 
(12:34:45) cron2: updated
(12:36:09) mattock: okeli dokeli
(12:36:18) mattock: 2.4.8 - what is missing
(12:37:11) cron2: I have one patch on my list, and that's the NCP/fragment fix 
- but I currently can not reproduce the problem at all ("release/2.4/HEAD 
without the patch behaves *exactly* the same as master with the patch") - so I 
might postpone that
(12:37:15) cron2: everything else is in
(12:37:32) cron2: version.m4, changes.rst, ChangeLog, tag, push :)
(12:37:57) cron2: any other patches that need to go in that I forgot about?
(12:38:20) mattock: did the other tap-windows6 PR get ack?
(12:38:59) rozmansi: not yet AFAIK
(12:39:08) cron2: which one?
(12:39:21) mattock: https://github.com/OpenVPN/tap-windows6/pull/84
(12:39:24) vpnHelper: Title: Introduce TAP adapter as a virtual device by 
rozmansi · Pull Request #84 · OpenVPN/tap-windows6 · GitHub (at github.com)
(12:39:47) rozmansi: thou JJK and Rogier tested it and reported it fixes HP 
laptop issue
(12:39:54) mattock: yep
(12:40:51) cron2: I thought the HP laptop issue was in something we already 
(12:41:09) rozmansi: cron2: it was a combination
(12:41:28) rozmansi: it's not until this last PR is in that it fixes it finally
(12:43:14) cron2: well, it got some testing, nobody complained, so let's put it 
(12:43:17) rozmansi: Surfnet would like to deploy OpenVPN (via eduVPN client) 
to a university in Netherlands that is using this kind of HP laptops 
exclusively and they would love to get this PR merged and published ASAP.
(12:44:56) cron2: seems that just happened... so we now need to bump the 
version numbers in the tap6-windows repo accordingly
(12:45:03) rozmansi: I don't see what could possibly go wrong. But it's a big 
world out there and TAP-Windows6 adapters has been marked as physical Ethernet 
NIC for a long long time.
(12:45:26) mattock: we'll know soonish :)
(12:45:50) cron2: mattock1: the driver out there is exactly what is in the repo 
now, right?
(12:45:52) mattock: the to-be-released driver will be identical to the test 
driver then
(12:45:55) mattock: yes
(12:46:04) lev__: cron2: that fragment/ncp stuff was fairly easy to reproduce, 
let me try now
(12:46:06) cron2: so we can keep the version number?
(12:46:09) mattock: yes
(12:46:28) cron2: lev__: that is what I thought :-) but 2.4 and master are 
behaving absolutely identically in my --fragment test setup
(12:47:24) lev__: cron2: have you looked at those steps 
(12:47:26) vpnHelper: Title: #1140 (mssfix appears to be broken with NCP) – 
OpenVPN Community (at community.openvpn.net)
(12:47:41) cron2: lev__: yes
(12:48:03) cron2: mattock1: please push the version number update to the repo, 
then :-)
(12:48:21) mattock: cron2: will do
(12:48:55) cron2: this somewhat sounds like the totally anticlimatic end of a 
long journey...
(12:51:15) cron2: ok, tap6-windows done :-) - anything else for 2.4.8?
(12:51:24) cron2: mattock1: when do you need the tagging?
(12:52:54) mattock: is around 13:00 your time ok?
(12:53:44) mattock: the sooner I can start the various builds the better of 
(12:54:20) cron2: 1300 is doable if I either can get lev__s patch sorted out 
quickly or we decide to postpone
(12:56:47) mattock: ok, no rush - it is unlikely that I could do the full 
release process today (lots of stuff in the evening plus toddler duty after 
(12:56:57) mattock: but if I could build the stuff today that'd be nice
(12:58:00) cron2: working on Changes.rst
(12:58:03) cron2: anything noteworthy?
(12:58:23) mattock: not from my end - there's lots of windows stuff but that 
goes elsewhere
(12:58:44) lev__: cron2: I just reproduced it
(12:59:02) lev__: (and ncp-disable fixes those)
(12:59:24) cron2: lev__: so what did you do, exactly?
(12:59:45) lev__: actually, I reprodeuced the problem (IP fragmentation), 
haven't tried the patch yet
(12:59:46) cron2: as I said, my release/2.4/HEAD build produces exactly the 
same MSS values as master
(13:00:34) cron2: my test is "look at the MSS values in tcpdump and in openvpn 
log output", and I cannot see a difference there
(13:07:11) mattock: ok, so how shall we proceed with this?
(13:07:33) cron2: we agree that you shall have your tag at 1300, in 53 minutes 
(13:07:50) cron2: Lev__ is testing the MSS stuff, but we merge that afterwards 
for 2.4.9 if we better understand what is happening
(13:07:51) mattock: ok, was referring to the discussion between you and lev
(13:07:59) mattock: that is enough for me
(13:08:00) cron2: that got moved to query
(13:09:06) mattock: off one thing I want to add completely unrelated to 2.4.8
(13:09:29) ***cron2 listens :)
(13:09:30) mattock: I'm thinking that instead of bringing back the "Windows 
buildslave" as-is, I will make it a buildslave, but just with a different build 
(13:09:45) mattock: same with sbuild_wrapper which I use to build deb release 
(13:09:56) mattock: buildbot would handle queuing, git monitoring, etc.
(13:10:03) cron2: all I care is "if I break mingw building, I want to know 
right away" - so how the machine works, I do not need to know
(13:10:13) cron2: (and we should have the windows snapshot builds back on the 
download site)
(13:10:17) mattock: yes, it would just trigger openvpn-build just like 
openvpn-windows-buildtest did
(13:10:47) mattock: we'd see build status in the buildbot ui as a bonus
(13:10:57) cron2: nice
(13:11:15) mattock: should be doable without major refactoring of buildmaster 
config (knock on wood)
(13:12:21) mattock: anything else for today?
(13:12:25) cron2: oh, something else I want to mention while I have people's 
(13:12:41) cron2: pushing to sourceforge is broken - I can push to release/2.4 
but not to master because a broken git lock is lying around
(13:12:58) cron2: their support tells me "please contact the project if you 
have questions about openvpn"... silly dimwits
(13:13:06) mattock: ugh
(13:13:16) cron2: no problem for 2.4 release, but something to keep in mind 
when fetching master
(13:13:58) mattock: do you have console output I could see and copy to the 
SF.net support ticket "from the project"?
(13:14:07) mattock: I can open one right away
(13:14:19) cron2: sure
(13:14:49) lev__: "anything else for today" I've sent a small patch which adds 
MSVC to travis build, would be nice to get it reviewed
(13:15:02) lev__: this will help us not to break VS builds
(13:15:08) cron2: coming over query
(13:15:10) cron2: yes
(13:17:17) cron2: pushing 2.4.8 to the buildbot repo... if they are happy, I'll 
tag and push to the official repos
(13:17:51) cron2: ah, mattock1: have you seen my request in #openvpn-devel to 
remove 4 dead buildslaves?
(13:18:25) mattock: yes
(13:18:33) mattock: I'll tackle that as well
(13:19:01) cron2: thanks.  Buildbot is lazy today :)  (sometimes it picks up 
new pushes "right away", sometimes it takes a few minutes)
(13:19:18) mattock: lazy is by design
(13:19:31) mattock: there is a "wait if new commits arive" timeout there
(13:19:31) cron2: anyway - 2.5 "nothing much has happened", except that I'm 
glueing my test server infra back together, so I can use that for TAP/VLAN 
tests "soon"
(13:20:02) cron2: mmmh, it's a pity dazo is not here
(13:20:54) cron2: we have t_client tests currently that fail on 2.4 ("not 
implemented yet") but pass on master - and I want to make things conditional in 
t_client.rc... so an easy way to figure out "which git branch are we in" would 
be appreciated...
(13:22:37) cron2: but maybe I can just tackle that next week
(13:22:43) cron2: so
(13:22:50) cron2: pull-resolv-conf/client.up
(13:22:58) cron2: this is a script for linux to get DNS info into place
(13:23:10) cron2: and it fails on Arch linux because they do not use 
"resolveconf" anymore but "resolvectl"
(13:23:22) mattock: the joys of new systemd components
(13:23:28) cron2: which seems to be the new and all improved systemd way to do 
things in a new way
(13:23:31) cron2: yes
(13:24:14) cron2: I could care less about all these newfangled Linux 
inventions, but unfortunately two colleagues @ customer have decided to use 
Arch for their home laptops, and now my openvpn configs I give them are failing 
(no DNS, and split DNS in use)...
(13:24:49) cron2: so, brainstorming session :-) - do we want to have this in 
our repos?  eworm says "arch does not add scripts, they just pack what openvpn 
upstream ships"
(13:27:08) mattock: ok noted
(13:29:38) mattock: 1 minute left :)
(13:29:43) mattock: done?
(13:30:00) cron2: yeah... no meeting next week thursday
(13:30:19) mattock: good point
(13:30:39) cron2: but long meeting next week friday :-) - looking forward to 
see you all again
(13:30:51) rozmansi: same here :)
(13:31:09) mattock: likevise!
(13:31:13) mattock: summary almost out

Attachment: signature.asc
Description: OpenPGP digital signature

Openvpn-devel mailing list

Reply via email to