Here's the summary of the IRC meeting.



Place: #openvpn-meeting on irc.freenode.net
Date: Wed 13th May 2020
Time: 11:30 CEST (9:30 UTC)

Planned meeting topics for this meeting were here:


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



dazo, krzee, lev, mattock, ordex and plaisthos participated in this meeting.


Talked about MSI/MSM for OpenVPN and tap-windows6. Mattock is now
fighting his way through the MSI/MSM jungle.


Talked about improving the buildbot configuration. Agreed that it makes
sense to:

1) Upgrade buildbot to a more modern version (better webui etc)
2) Migrate lots of semi-manual testing that is currently internal to
OpenVPN Inc. to buildbot
3) Start building openvpn3 on buildbot
4) Testing latest server code through buildbot

After the meeting cron2 reminded us that he has t_server testing
framework already - it is just running outside of buildbot.

Krzee and mattock will work together on these buildbot improvements.


Discussed OpenVPN 2.5 patches. This week dazo will review plaisthos'
three patches and start the epic "man page struggle".


Full chatlog attached

(12:28:46) mattock: hello!
(12:28:56) dazo: hey!
(12:30:12) mattock: cron2 said he could not make it
(12:30:26) mattock: it would be nice if rozmansi was here by accident :)
(12:30:42) mattock: I'm working on tap-windows6 MSM and unsurprisingly there 
have been a number of issues I've had to resolve
(12:30:48) krzee [be50baf1@openvpn/corp/krzee] è entrato nella stanza.
(12:30:53) mattock: probably related to the build environment, but problems 
(12:31:59) ordex: ué
(12:32:25) lev__: hello
(12:33:29) plaisthos: hello
(12:34:03) krzee: heyhey
(12:35:21) mattock: hi all
(12:35:23) mattock: so
(12:35:38) mattock: I pretty much reported what I've been up to :)
(12:35:48) mattock: any topics besides openvpn 2.5?
(12:36:16) dazo: plaisthos: what's the status now on the auth-token patches?
(12:36:35) dazo: Lets have a look here: 
(12:36:37) vpnHelper: Title: StatusOfOpenvpn25 – OpenVPN Community (at 
(12:37:15) dazo: I'm planning to start playing with the man page challenge this 
(12:37:32) mattock: \o/
(12:37:39) plaisthos: dazo: err chek patchwork
(12:37:44) mattock: sounds like e-sports
(12:38:51) krzee: id like to understand a little more about our unit tests 
(using buildbot as i understand), specifically because i think ovpn3 should be 
added to it when possible and i think that maybe corp and community could use 
the same tests so that efforts are not duplicated
(12:41:19) mattock: +1
(12:42:17) mattock: basically we'd just add openvpn3 build dependencies to a 
subset of our buildslaves and add a few builders to build openvpn3
(12:42:55) dazo: plaisthos: ahh, found it ... alright, 3 patches missing formal 
acks ... I'll try to complete that this week then
(12:44:57) dazo: mattock: Hmmm ... can we also please upgrade our buildbot to 
something more up-to-date?  So that it is simpler to get an overview what is 
being run/tested, split out build errors from test errors, etc
(12:46:23) mattock: yes that is the plan
(12:46:30) mattock: after MSI stuff
(12:46:50) dazo: good!  Then I'll keep quite for a bit more :-P
(12:46:51) plaisthos: like jenkins?
(12:46:55) dazo: *quiet
(12:47:06) plaisthos: or newer buildbot?
(12:47:10) mattock: no jenkinses
(12:47:14) mattock: update buildbot
(12:47:16) mattock: :)
(12:47:20) krzee: how many tests are already implemented in buildbot?
(12:47:22) dazo: plaisthos: hehe ....
(12:47:35) dazo: krzee: Only `make check`
(12:47:36) mattock: not sure as those are created programmatically
(12:47:47) plaisthos: no was a serious question. People might have more 
experience with jenkin
(12:48:12) mattock: yes, and it is still a piece of crap especially from 
management perspective
(12:48:37) ordex: :D
(12:48:51) krzee: dazo, i dont understand the answer
(12:49:44) ordex: tests are defined in t_client
(12:49:44) dazo: krzee: buildbot runs 'make check' ... so what is being run is 
defined in tests/Makefile.am via the TESTS variable
(12:50:04) ordex: the buildbot just runs make check (which executes the unit 
tests and the various t_* scripts)
(12:50:07) dazo: t_client.sh is one of the defined tests in TESTS
(12:50:31) ordex: right
(12:51:07) mattock: krzee and I also talked about extending t_client to include 
server-side tests (i.e. connect clients to a server instance built from HEAD)
(12:51:28) mattock: at the moment our servers are static (version <n> of 
(12:51:36) dazo: 
https://github.com/OpenVPN/openvpn/blob/master/tests/Makefile.am#L17  .... oh 
and beware of SUBDIRS too ... so this is also evaluated, recursively
(12:51:38) vpnHelper: Title: openvpn/Makefile.am at master · OpenVPN/openvpn · 
GitHub (at github.com)
(12:51:40) ordex: yeah, that would be a great improvement
(12:52:26) dazo: I would suggest implementing a t_server.sh test for that ... 
to avoid complicating t_client.sh further
(12:52:49) mattock: kind of reverse of t_client.sh?
(12:52:57) mattock: as in: static clients, latest server?
(12:53:24) krzee: as well as latest client
(12:53:30) dazo: kinda ... but it could drive also client connections too, but 
t_client.sh is well defined in what it tests and how it works
(12:53:32) mattock: yeah I was about to add that
(12:53:36) ordex: maybe a coordinated effort of t_server andf t_client ? :D to 
run combined tests
(12:54:35) mattock: something like that, but I'm not sure how it would look like
(12:54:37) dazo: yeah, t_server.sh cloud use t_client to drive the client side 
... but t_client.sh should try to strive to focus on testing the client side 
code - it can be extended to handle other configs than t_client.rc
(12:54:55) dazo: gee .... cloud -> could
(12:55:26) mattock: basically one buildslave could just build the latest server 
and make it available before t_client kicks in
(12:55:44) dazo: yeah
(12:56:01) mattock: the only challenge is enforcing the order afaics, but I'm 
sure that can be done
(12:56:58) krzee: i cant add much to the conversation yet because ill need to 
catch up to how tests work :)
(12:57:01) mattock: anyhow, buildbot upgrade is the first step - our version 
_is_ old
(12:57:09) dazo: The test code is not set in stone ... fortunately ;-) .... I'm 
just advocating keeping t_client.sh to be focused on client configs
(12:57:29) mattock: krzee: once I have MSI/MSM in a reasonable shape (=able to 
make release packages) I can help you with it
(12:57:39) mattock: dazo: +1
(12:57:44) mattock: it works well now
(12:57:47) krzee: mattock, maybe after MSI stuff we can pivot our meetings to 
(12:57:55) mattock: krzee: +1
(12:58:00) krzee: nice
(12:58:17) krzee: i still need to write this up to get approval from above
(12:58:30) krzee: but if we put our personal meetings into this, screw it we 
doing it
(12:58:32) krzee: lol
(12:58:44) mattock: be a cowboy like the rest of us :D
(12:58:55) mattock: "lone ranger who does his own thing"
(12:59:03) krzee: dude im in, this is the only logical way
(12:59:37) mattock: I mean, everyone will benefit from this (company and 
(12:59:41) krzee: the idea of manually testing everything in openvpn, which has 
more potential configs that ANY other app i have used, is just silly
(12:59:58) mattock: that is what our QA does at the moment
(12:59:58) krzee: than*
(13:00:07) mattock: due to mostly historic reasons
(13:00:10) krzee: oh i know, i have become qa :D
(13:00:15) krzee: just ask ordex
(13:00:22) ordex: :D
(13:00:27) ordex: yeah
(13:00:50) mattock: anyhow, it seems we have a plan
(13:01:11) krzee: yep
(13:01:16) krzee: thanks
(13:01:26) mattock: mattock will fight his way through the MSI/MSM jungle (MSM 
is still on the winning side) and then he and krzee will move onto automating 
more testing and making it available via buildbot to the community developers 
as well
(13:01:55) mattock: dazo will check plaisthos' three patches this week, as well 
as start the epic "man page struggle"
(13:02:12) mattock: anything else that needs discussion/coordination?
(13:02:15) krzee: best of luck with msi 
(13:02:51) mattock: I have to clear the jungle first before I can even get to 
the MSI
(13:02:52) mattock: :)
(13:07:18) mattock: anything else or are we done?
(13:09:41) krzee: ...
(13:10:12) mattock: I think we're done
(13:10:35) krzee: i doubt ill make it to your morning meeting
(13:10:43) krzee: :D
(13:10:53) mattock: tomorrow?
(13:11:06) krzee: my tomorrow, your tonight
(13:11:07) mattock: (it's not today, so not all hope is lost)
(13:11:20) mattock: Thursday at 7:30 AM PST
(13:11:25) krzee: OHHH
(13:11:28) mattock: yep
(13:11:30) krzee: sweet im in
(13:11:40) krzee: wednesday was unlikely :D
(13:11:48) mattock: definitely, given what time is at your end
(13:34:32) krzee ha abbandonato la stanza (quit: Remote host closed the 
(13:37:18) cron2: wrt "automated server side tests" - that is *hard*, but I 
have it outside buildbot
(13:37:29) cron2: (just to mention it again)
(13:40:28) cron2: ... and please do not start discussing and sort-of-deciding 
stuff that is not on the agenda
(13:58:54) mattock: that is why we have summaries, so that people can also 
complain after the fact
(13:59:04) mattock: summary which I am about to send :)
(13:59:45) cron2: yeah, but it's a waste of precious time if you discuss 
something where I might have answers already (and I'm sure I mentioned my 
t_server setup like 100 times in the past)
(14:01:47) mattock: I can waste 10 minutes discussing it, but yes, we need to 
take your t_server setup into account before actually implementing it
(14:02:06) mattock: implementing anything of sorts

Attachment: signature.asc
Description: OpenPGP digital signature

Openvpn-devel mailing list

Reply via email to