Hi, Here's the summary of the previous community meeting.
--- COMMUNITY MEETING Place: #openvpn-devel on irc.freenode.net List-Post: openvpn-devel@lists.sourceforge.net Date: Thursday, 16th Sep 2010 Time: 18:00 UTC Planned meeting topics for this meeting were on this page: <https://community.openvpn.net/openvpn/wiki/Topics-2010-09-16> Next meeting next week, same place, same time. Your local meeting time is easy to check from services such as <http://www.timeanddate.com/worldclock> or with $ date -u SUMMARY Discussed the "Windows route + hibernate/suspend issue": <https://community.openvpn.net/openvpn/ticket/56> D303k and jamesyonan debugged this issue during the meeting and got mixed results on Windows 7 VM running on VMWare ESX: sometimes OpenVPN recovered properly, sometimes not. Mattock promised to ask the original bug reported for log files with "verb 7" log level. D303k promised to try reproducing this problem on real, hardware Win7 laptop. This issue seems to be Windows-related, as similar problems have not been reported for *NIX platforms. Also, mattock and frk are constantly using OpenVPN with hibernation on Ubuntu 9.10 / 10.04 / 10.10 without any issues. -- Discussed the use of a relatively powerful Xen DomU (VM) which was offered to the project. Potential uses for this server were discussed in this meeting two weeks ago. Decided to use VM as the public test server (for "make test") and as a buildslave. FreeBSD was chosen as the OS, other alternatives being Debian, Ubuntu, CentOS and Win2008k. The reason was that we already have buildslaves for the other OSes (except Win2008k) and in test server use the OS does not really matter. -- Mattock gave an update on recently released 2.2-beta3. Looking at the webserver logs 2.2-beta3 there have been total of 6842 downloads from unique IPs. Out of these 6593 were windows installers. As 2.2-beta3 has without a doubt been downloaded from computers behind NAT firewalls, the real download figure is probably between 7000 and 10000. -- Mattock gave an update on Buildbot and Debian packages. Yesterday he got four buildslaves (debian-5-i386/amd64 and ubuntu-10.04-i386/amd64) connected to the buildmaster server through the community VPN. These buildslaves will then build packages for their respective OS/architecture. Building the Debian packages under buildbot has been tested already, but the debian control files need some work, as does buildmaster configuration. If all goes well, first Debian packages based on "allmerged" branch should be available in the evening of 17th September. -- Discussed the possibility of hosting future OpenVPN (open source) file releases on SF.net instead of openvpn.net servers. The main download page with download links, release notes etc. would still be at openvpn.net: <http://openvpn.net/index.php/open-source/downloads.html> This would give a few benefits: * More visibility for the OpenVPN project through higher ratings at SF.net * Free mirroring and bandwidth * Very detailed download statistics The only downside is the potential reduction of traffic going through openvpn.net. An alternate approach would be to host releases primarily on openvpn.net servers, but offer them through SF.net also as a secondary option. Mattock promised to look into this issue. --- Full chatlog as an attachment -- Samuli Seppänen Community Manager OpenVPN Technologies, Inc irc freenode net: mattock
(21:01:41) jamesyonan [~jamesy...@c-76-120-71-74.hsd1.co.comcast.net] è entrato nel canale. (21:01:55) mattock: hi james! (21:02:20) modalità (+o jamesyonan) da ChanServ (21:02:20) jamesyonan: Hi (21:02:20) fkr: hey james. (21:02:34) fkr: <- the guy who maintains openvpn on openbsd (among other things) (21:02:49) mattock: dazo: did you find any stuff worth discussing? (21:03:08) mattock: perhaps the thread Alon started... (21:04:28) dazo: mattock: No, I haven't managed all I wanted ... but it's been a lot of interesting discussions on the ML lately (21:04:32) ecrist: raidz: I'm thinking that the best thing to do would be to put up a webser that can respond with something as simple as 'Service Unavailable' (21:04:42) ecrist: not worth the effort to replicate the db (21:04:53) ecrist: if it's going to be long-term, we can restore the db from backups (21:05:14) mattock: ok, time to start meeting... (21:05:17) ecrist: if the forum gets 'really' active, we can even start doing backups more frequently than once a day. (21:05:19) raidz: ecrist: no problem, that works (21:05:24) ***ecrist shuts up (21:05:46) mattock: ok, so topic list is short and here: https://community.openvpn.net/openvpn/wiki/Topics-2010-09-16 (21:05:48) vpnHelper: Title: Topics-2010-09-16 â OpenVPN Community (at community.openvpn.net) (21:06:21) mattock: I dug into the Windows route + hibernate/suspend issue today: https://community.openvpn.net/openvpn/ticket/56 (21:06:24) vpnHelper: Title: #56 (WinXP: route setup broken after standby) â OpenVPN Community (at community.openvpn.net) (21:06:31) d303k [~heiko@88.130.197.223] è entrato nel canale. (21:06:33) mattock: seems like similar issues have been around since 2004 (21:06:38) mattock: hi d303k! (21:06:43) d303k: hello (21:07:11) mattock: jamesyonan: what's you're take on the Windows suspend/hibernate issue? Is that a known problem and something we could fix? (21:08:09) d303k: mattock: what are the symptoms? (21:08:28) ***dazo understands it as VPN routes are missing (21:08:29) mattock: d303k: check this bug report: https://community.openvpn.net/openvpn/ticket/56 (21:08:30) vpnHelper: Title: #56 (WinXP: route setup broken after standby) â OpenVPN Community (at community.openvpn.net) (21:08:45) mattock: there are several similar reports from 2004, 2006 and 2010 (21:09:12) mattock: I'm wondering if this is an issue with Windows suspend/resume cycle or OpenVPN (21:09:13) jamesyonan: why not just re-add the routes after awakening from standy? (21:09:44) dazo: Without knowing how this stuff really works on Windows, I think it is related to that the routes are taken down automatically when the device is suspended (21:09:45) jamesyonan: does it occur when "persist-tun" is used? (21:09:50) d303k: openvpn-gui does tear down openvpn when windows is suspended, however the bugreport doesn't mantion the gui (21:10:13) mattock: I'll check the latest ml thread again... (21:10:33) mattock: it's here: https://sourceforge.net/mailarchive/message.php?msg_name=L8D646C6A35154b2399CD5E4EB944E66D.1284482670.email.pmi.phoenixmi.com%40MHS (21:10:34) vpnHelper: Title: SourceForge.net: OpenVPN: (at sourceforge.net) (21:10:46) bGatti [~b...@cpe-024-074-150-231.carolina.res.rr.com] è entrato nel canale. (21:10:56) mattock: btw, better replace https with http (21:11:03) mattock: hi bGatti! (21:11:12) mattock: topic list is here: https://community.openvpn.net/openvpn/wiki/Topics-2010-09-16 (21:11:14) vpnHelper: Title: Topics-2010-09-16 â OpenVPN Community (at community.openvpn.net) (21:12:04) jamesyonan: hi ben (21:12:09) bGatti: Hi. (21:12:10) mattock: jamesyonan: it seems using "persist-tun" makes no difference (21:12:54) jamesyonan: it's completely normal for the openvpn client to lose the connection with the server after awake from standby (21:13:06) jamesyonan: but it should reestablish it on wakeup (21:13:33) mattock: the issue seems to be that openvpn client does not reconnect properly after waking up (21:13:34) dazo: agreed ... but could it be that routes are processed in this case? (21:13:55) dazo: good point, mattock ... I took that for granted (21:14:15) mattock: on Windows 7 at least, although one guy reported that routes _are_ re-established after wakeup on WinXP (21:14:26) jamesyonan: is the issue a failure to reconnect, or does the reconnect occur but routes are not added? (21:15:17) dazo: we should probably request log files from the clients, with --verb set to minimum 5 ... (21:15:48) mattock: so we may be looking at two different issues: the old one (from 2004-2006) and the new one (21:15:51) dazo: unless anyone else got access to a Win7 box and can manage to reproduce this issue (21:15:53) mattock: dazo: agreed (21:16:13) d303k: i have a virtual win7 at work# (21:16:20) mattock: I can ask for log files from the reporter (21:16:29) mape2k [~mape2k@2001:6f8:133b:0:21f:3bff:fe27:21a9] è entrato nel canale. (21:16:33) mattock: d303k: could you try to reproduce this issue? (21:16:52) dazo: I would say we should have a close look at this new Trac ticket first. A 6 year old report won't give too much right now (21:16:56) mattock: on Ubuntu 9.10 I've had zero problems with OpenVPN recovering from hibernation (21:16:57) d303k: i'll try (21:17:07) dazo: d303k: cool! thx! (21:17:39) fkr: mattock: can second that on 10.04 and 10.10 (which is beta) (21:17:46) mattock: fkr: ok (21:17:58) mattock: so most likely only related to Windows (21:18:25) mattock: ok, so I'll request log files from the reporter and d303k will try to reproduce this on Win7 (21:18:29) mattock: next topic? (21:18:45) mattock: hmm, perhaps output of "route" command would be nice, too (21:18:55) fkr: ack (21:18:56) mattock: what else? (21:18:58) dazo: I think it is isolated to Windows ... and I got a vague feeling it might be connected to the Windows TAP driver not being ready yet when openvpn.exe starts reconnecting (21:19:06) mattock: perhaps "ipconfig" output? (21:19:44) dazo: mattock: nah ... logs with --verb 5 or higher should be sufficient right now ... that will tell if routes are tried to be set or not, and if that fails or not (21:19:49) mattock: ok (21:20:06) dazo: ipconfig output of the tap device might be interesting though (21:20:08) d303k: so the guy had the service start the connection (21:21:01) mattock: d303k: I think so, he didn't mention anything about a GUI (21:21:08) dazo: mattock: make it --verb 7 ... I think that can be good now (21:21:14) mattock: dazo: ack (21:21:27) mattock: next topic: Wayne's VPS server offer (21:21:45) mattock: wayne = praetorian (21:21:54) Praetorian|Mobil: thats me :P (21:22:02) dazo: Cool! (21:22:48) mattock: praetorian: you mentioned the VPS was dual core... do you mean the VPS has two dedicated CPUs or that the host has two CPUs? (21:22:57) dazo: I would say that having a build slave and a public test server, where people (everyone who are interested) can try to connect against the server ... it should not forward anything, but could run a iperf server (21:23:28) Praetorian|Mobil: the server is a 16 core box. (21:23:29) dazo: the build slave would be, secondary in my eyes (21:23:42) mattock: praetorian: ok, that's plenty :) (21:23:54) mattock: so I think we should use it for something that benefits from those 2 cores (21:23:54) Praetorian|Mobil: I bascially got in on an intro offer for 3 vps for a flat yearly fee that was too good to pass up for two vps let alone three (21:24:40) Praetorian|Mobil: we have used them for almost tow years now and the uptime is pretty damn good for vps providers (21:25:11) mattock: dazo: I was thinking along the same lines (21:25:14) Praetorian|Mobil: but I only have uses for two now.. so instead of just letting it sit there and be wasted I would like to let the openvpn community use it (21:25:32) dazo: That's really awesome! (21:26:01) mattock: praetorian: what were the OS choices again? centos, debian, freebsd, ubuntu... what else? (21:26:19) Praetorian|Mobil: win2008 ent I think was hte other one (21:26:23) mattock: oh yes... (21:26:33) mattock: would we have any use for a Win2008 server? (21:27:18) mattock: if not, I think a combined buildslave + test server would be best option (21:28:28) dazo: hmm ... I think it will be better use of it as test server + build slave .... but we need to think about how we do test the windows platform (21:29:45) Praetorian|Mobil: the only request I have to make it not possible to be used as a spam tunnel. (21:29:57) dazo: Praetorian|Mobil: agreed! (21:29:58) mattock: praetorian: I'll make sure it won't be misused (21:30:17) Praetorian|Mobil: :) (21:31:11) dazo: I'd say that's to restrict the firewall on that VM to only allow traffic via the tunnel to specific IPs ... like a iperf server or similar stuff (21:31:21) mattock: what if we use it as a buildslave and a test server, then? what OS to use? Debian/Ubuntu are already covered (21:31:25) dazo: Praetorian|Mobil: are there any bandwidth restrictions on this VM? (21:31:41) Praetorian|Mobil: 5.6tb both ways a month (21:32:04) mattock: what happens if that limit is reached? (21:32:18) mattock: are the connections refused or do you get a really big bill? :) (21:32:21) Praetorian|Mobil: hmm, good question.. let me look and see (21:32:32) dazo: Praetorian|Mobil: I'd consider that "unlimited" for our use case ;-) (21:33:31) dazo: mattock: If installing Linux, we can setup some iptables rules which can block traffic when it hits a certain threshold (21:33:41) Praetorian|Mobil: they dont actually sy on their site.. so not sure, one thing, its capped at 20mbit rate.. (21:34:55) dazo: Praetorian|Mobil: if we're going for an OpenVPN test server, open for "everyone" to run tests against ... would that make you nervous? (21:35:53) Praetorian|Mobil: nope, they have it QOS'd so if they go over that it will get throttled. (21:36:12) dazo: :) sounds good! (21:37:35) mattock: ugh... just got phone call... (21:38:51) mattock: I suggest using FreeBSD for the OS... all other OS'es are/can be covered easily with buildslaves (21:39:02) mattock: and for the "make test" server OS does not really matter I guess (21:39:09) mattock: and Windows is not an option :) (21:39:16) dazo: true enough! (21:39:42) mattock: I wouldn't want to contaminate forums.openvpn.net (freebsd VM) with buildslaves, either (21:40:08) mattock: praetorian: can we select the DNS name for the VPS? (21:40:33) Praetorian|Mobil: sure, that should be possible. (21:40:48) Praetorian|Mobil: you are speaking of reverse dns and host anme right (21:41:13) mattock: yep (21:41:25) Praetorian|Mobil: should not be a problem. (21:41:32) Praetorian|Mobil: just let me know what you want (21:41:35) mattock: ok (21:41:55) mattock: I guess that topic is covered, then :) (21:42:02) mattock: praetorian: thanks a lot for the VPS! (21:42:09) mattock: much appreciated! (21:43:05) mattock: jamesyonan, raidz: I'm interested in download statistics from 2.2-beta3 and 2.1.3... are any available? (21:43:20) mattock: s/from/for/1 (21:44:06) mattock: so far I've heard no complaints about 2.2-beta3 (21:44:06) Praetorian|Mobil: can I inquire about the updated debian builds? with gerts ipv6 payload in them? (21:44:15) mattock: praetorian: sure (21:44:52) mattock: in fact, today I finally got my four buildslaves (debian-5-i386/amd64 and ubuntu-10.04-i386/amd64) connected to the buildmaster server (21:44:54) raidz: I can parse our logs (21:45:01) mattock: raidz: thanks! (21:45:20) mattock: so, these buildslaves will build packages for their OS+architecture (21:45:46) mattock: I've already tested building debian packages using a buildslave and it seemed to work ok (21:46:08) mattock: the debian control files need some work, as does buildmaster configuration, but otherwise things are looking bright (21:46:33) mattock: I intend to continue working on these buildslaves tomorrow (21:47:02) mattock: If I'm don't run into big issues we'll have some debian/ubuntu packages based on "allmerged" branch available tomorrow evening (21:47:24) mattock: and "allmerged" has ipv6 support (21:47:33) Praetorian|Mobil: mattock: cool, I have a setup that I would like to test with the new packages when they are ready. (21:47:51) Praetorian|Mobil: 1 server with 6 clients all using ip6 (21:48:13) Praetorian|Mobil: inside the tunnel (21:48:22) mattock: great, IPv6 _and_ my debian packages are begging for testing :) (21:48:29) dazo: ipv6 in the tunnel ... that will make cron2 happy! (21:50:20) dazo: Praetorian|Mobil: your setup is using tun or tap devices? (21:50:54) dazo: if using tun ... that will be the new big thing for ipv6 support (21:51:26) Praetorian|Mobil: heh, i think its tun.. its been working so well I have not had to ess with it (21:51:46) mattock: I'll check download statistics for 2.2-beta3 now (21:51:58) dazo: Praetorian|Mobil: now IPv6 is supported in TAP mode only, with manually configuring IPv6 addresses (21:52:28) Praetorian|Mobil: dazo: correct, I setup both ends with addresses. (21:53:12) dazo: Praetorian|Mobil: yeah, that explains ... if it is possible to trick you into running 'allmerged' with tun mode ... that would be very much interesting :) (21:54:10) Praetorian|Mobil: Ill give it a shot.. (21:55:07) Praetorian|Mobil: build me a package (I'm not very good at building packages.) more of a generic network admin that does way to many things. (21:56:29) dazo: Praetorian|Mobil: cool! :) (21:58:26) d303k: ok, i could reproduce the thing on win7 (21:58:50) dazo: d303k: lovely! could you parse something sensible out of the log files? (21:59:01) d303k: after coming back from hibernation two services use up 100% cpu (21:59:04) mattock: ok, so unless my command-line was screwed up, 2.2-beta3 has been downloaded 27515 times (21:59:12) mattock: I'll see how many of those are windows installers (21:59:19) dazo: mattock: that's a lot! (21:59:23) mattock: yep (21:59:33) mattock: in 6 days (22:00:02) d303k: when i killed the openvpn.exe child of openvpnserv.exe the load got less but was not at 100% idle, yet (22:00:03) mattock: and no complaints so far... 2.2-beta3 is probably pretty stable (22:00:20) d303k: stopping the service brought back 100% idle again (22:00:20) dazo: mattock: over 4500 downloads per day, that sounds too good to be true ... for a beta release ... (22:00:31) mattock: 22638 of those downloads are windows installers (22:00:57) mattock: dazo: we should not forget that 2.2-beta3 is the _first_ release on the page, and people never pay attention to what they're doing :P (22:01:16) dazo: mattock: true :) (22:02:20) d303k: the verb 3 log didn't show anything after init seq completed (22:02:20) mattock: I'll check how many unique IPs have downloaded 2.2-beta3 (22:02:33) d303k: openvpn*.exe did not use any cpu (22:02:51) d303k: the two services were started by svchost (22:03:12) dazo: jamesyonan: ^^^ do you pay attention to d303k findings? (22:03:39) mattock: ok, now we should get more realistic download statistics (22:03:44) mattock: filtered by source IP (22:03:58) jamesyonan: d303k: which process was using 100% cpu? openvpn.exe or openvpnserv.exe ? (22:04:01) d303k: C:\Windows\System32\svchost.exe -k LocalServiceNetworkRestricted was the one taking less cpu (22:04:02) mattock: although if there are several clients behind the same NAT, then that does not work (22:04:41) mattock: ok, windows installer of 2.2-beta3 has been downloaded from 6593 unique IPs (22:04:51) d303k: C:\Windows\system32\svchost.exe -k LocalServiceNoNetwork was the other one (22:05:24) dazo: mattock: what about to put future community releases on sf.net? ... they got mirroring and download statistics ... and it we might get some advertisements via sf.net announcements if it scores high up on the lists ... (22:05:28) d303k: jamesyonan: none of the openvpn processes used a recognizable amount of cpu (22:06:38) mattock: dazo: I think we could put the releases to SF.net and Freshmeat as well as openvpn.net (22:06:44) jamesyonan: d303k: which process was using most of the CPU then, when usage was spiking to 100%? (22:07:03) d303k: the two service processes mentioned above (22:07:17) d303k: the latter one was the greater hog, though (22:07:26) dazo: mattock: agreed ... if I understand you right, having downloads only from sf.net, but pointers from openvpn.net and freshmeat.net to the sf.net download page ... (22:07:29) jamesyonan: svchost.exe? (22:07:41) d303k: nd they calmed down only after stopping openvpnserv.exe (22:08:12) d303k: well, svchost ist just the starter pay attention to the -k parameters (22:08:33) d303k: i don't know what they are for yet (22:08:59) d303k: but it might indeed have to do with tap behaving badly on resume (22:09:20) mattock: dazo: I was thinking about keeping openvpn.net the primary download page, but augmenting it with sf.net and freshmeat to increase visiblity further. And we could have links to alternate downloads pages (=sf.net and freshmeat) (22:09:42) mattock: personally I would not mind having the downloads in SF.net only, but that's also a political decisions (22:09:57) d303k: jamesyonan: any specific thing i should pay attention on the upcoming next try (22:10:59) dazo: mattock: I understand the concept for the closed source release ... but for the open source/community edition, I think it would be better to move it ... also as they have pretty good download statistics, and the new beta interface is very much interesting in that regards .... plus good mirroring (22:11:52) mattock: dazo: I can suggest that... I agree that SF.net's download statistics are pretty good (22:11:57) mattock: and fun to watch :) (22:12:44) dazo: yeah :) And if we're able to get it up on the "top 20 downloads" list ... openvpn gets even more attention through their community mails as well (22:12:50) mattock: plus with this many downloads we would be one of the most active projects (22:13:24) jamesyonan: d303k: not sure -- svchost.exe is a windows app (22:13:52) dazo: mattock: yup! (22:14:20) jamesyonan: openvpnserv.exe defines two dependencies -- the TAP driver and the DHCP client. svchost is supposed to make sure these dependent services are started before starting openvpnserv.exe (22:14:31) ecrist: mattock: did you have source download numbers? (22:14:36) jamesyonan: that's the main interaction (22:15:29) d303k: jamesyonan: svchost.exe is used to start "normal" programs as services iirc (22:15:31) mattock: ecrist: so total of 6842 downloads from unique IPs... out of these 6593 were windows installers. (22:15:44) mattock: so ~300 source downloads (22:15:46) dazo: that sounds more likely (22:16:08) d303k: i found instructions on how to find out which functions are called when it eats up cpu (22:16:10) mattock: although I'm pretty sure many of those connections are coming from behind corporate NAT firewalls (22:16:41) d303k: i'l try to find out tomorrow at work (22:16:53) d303k: http://www.msfn.org/board/topic/140264-how-to-get-the-cause-of-high-cpu-usage-caused-by-apps/ (22:16:54) vpnHelper: Title: How to get the cause of high CPU usage caused by apps? - MSFN Forums (at www.msfn.org) (22:17:20) jamesyonan: d303k: are you saying that you're running openvpn.exe as a service, but not using the windows service control manager to do so? (22:17:50) mattock: James' signatures have been downloaded from 216 unique IPs... (22:18:59) dazo: mattock: agreed, some of the downloads have come through some NAT's ... but probably not over 16000 ;-) (22:19:15) d303k: jamesyonan: no i'm saying svchost is not the scm process (22:19:35) mattock: I'd say more than 7000 but less than 10000 downloads (22:19:46) dazo: mattock: agreed (22:19:52) d303k: the scm process is called services.exe (22:20:07) d303k: and openvpnserv.exe is a child of it (22:20:26) ecrist: mattock/raidz: any word on getting a section onto the forums for access server? (22:20:29) d303k: and so are the two offending svchost.exe processes (22:20:34) ecrist: https://forums.openvpn.net/viewtopic.php?f=6&t=7104 (22:20:35) vpnHelper: Title: OpenVPN Support Forum View topic - No Access Server (at forums.openvpn.net) (22:22:08) mattock: ecrist: raidz is probably the best person to answer that (22:22:32) raidz: I will work on that today, we can plan on launching it Monday. (22:23:04) mattock: I think the official part of the meeting is over by now... (22:26:00) dazo: :) (22:26:27) d303k: jamesyonan: just tried hibernation without openvpn runninf and the "svchost.exe -k LocalServiceNetworkRestricted" process used up some cpu for a while (22:26:52) ***dazo need to head for bed ... still feeling the jetlag a little bit :) (22:27:10) mattock: dazo: good night! (22:27:22) d303k: dazo: night (22:27:34) dazo: c'ya! (22:27:54) dazo: d303k: thx a lot for looking into the suspend stuff! Great work there! (22:27:55) d303k: jamesyonan: trying again with ovpn and will be more patient (22:28:37) d303k: dazo: nah (22:29:29) dazo è ora conosciuto come dazo_afk (22:32:51) mattock: I'll write the summary tomorrow (22:34:55) d303k: jamesyonan: hm, this time coming back from hibernation worked, the routes are there and the LocalServiceNoNetwork service doesn't go nuts (22:35:47) d303k: and cpu is idle now after a while of setteling down (22:38:05) jamesyonan: maybe the issue is triggered by the order in which services come out of hibernation (22:38:55) d303k: jamesyonan: i'm trying again (22:39:04) Praetorian|Mobil ha abbandonato il canale (quit: ). (22:40:20) d303k: worked again (22:58:12) d303k: jamesyonan: ok i can reproduce the issue again, but i'm not sure if it's really the one the user has (23:03:41) d303k: the win7 runs in a VM on a ESX server. when i try to sleep (which does not work - win7 comes back instantly) before hibernating the LocalServiceNoNetwork service loops (23:03:53) d303k: trying without openvpn again now (23:25:19) cron2: re (23:26:20) d303k: final words for today: LocalServiceNoNetwork uses 100% cpu after a couple of the non-working sleeps on the VM as well as after hibernate after sleeping. that's when the route are not set (23:27:27) d303k: however, i don't trust the ESX, i'll try to steal a laptop with win7 on it from a coworker tomorrow and verify my findings (23:27:40) d303k: cron2: hi and good night to all (23:27:54) cron2: d303k: hi and good night :-) (23:28:47) d303k: oh and as soon as i kill openvpn.exe cpu goes down to idle again (23:30:17) d303k: ok that's it then for tonight
