[04:57] good morning [05:25] gutem morgen [05:25] guten [05:26] Good evening [05:26] hows the family superfly [05:26] and you? [05:27] hi theblazehen inetpro magespawn MaNL paddatrapper and other lurkers [05:28] Kilos: we go on [06:48] goeie middag [07:01] dag se inetpro [07:29] system restart needed [07:43] Morning Kilos, everyone [07:54] good evening paddatrapper [08:02] Mornings all [08:07] I'm sitting with a conundrum. We have a 10Meg fibre line at the office with Neotel. Then at Neotel DC we a cabinet, with a 100Meg connection. Between the sites we have a pptp tunnel. Speedtests we get 8Meg dl and about 7Meg ul. Yet when we rsync between the two sites through the VPN tunnel we barely reach 1Meg. Even mysql replications don't even reach 1Meg. But the lines are maxing out. At my wits end with Neotel as their QoS is wrongly configured [08:07] somewhere. Having a meeting with them today. [08:12] ai! [08:12] when it warms up your head will think more clearly [08:15] Been coming since summer, so ja will see [08:15] lol [08:15] thats tricky [08:16] yip, even with the 2% extra overhead on the packet we should still at least hit 5Megs on the vpn tunnel [08:17] vpn is private network right? [08:17] not a virtual network [08:17] yip [08:18] encrypted network [08:18] will mtr not show you bottlenecks [08:19] nope mrtg is reporting maxing the lines out when we copy a few megs over, no bottlenecks. [08:19] sjoe, so what is adding the brakes [08:20] inetpro read and explain why and where prob is please [08:20] thats it I don't know on Neotel side. We have a managed fw on the DC side [08:21] the pptp on our side is compressed with mtu of 1450 [08:22] Our fibre at the office is 100Meg line, but QoS to 10Meg. My biggest suspicion is that QoS config [08:23] is the prob the same both ways [08:23] or cant you test from other side [08:23] same results which ever way [08:24] not perhaps something to do with the encryption [08:25] disbaled encryption on both sides and still same results [08:25] the nsa is slowing it down while they decrypt it [08:25] oh my [08:25] lol could be [08:25] so both sides connect to neotel [08:25] yip [08:26] have you tried talking to one one their techies [08:26] bahwahahahaha, the lot from India [08:26] prob must surely be by them [08:26] oh [08:27] you first have to sy salaam [08:27] say [08:27] its that holiday atm isnt it [08:28] theblazehen ideas man [08:28] all they do is deable the config so we get the full 100Meg at the office then all is working 100%. Then they implement it then their reason is that it is bottlenecking and loosing packets. Yet they can't proof to me that it is and I don't see any packet being dropped on our interfaces [08:28] im a greeter bot [08:29] Dumela [08:29] so its by them [08:29] yip [08:29] hi divansantana [08:30] oh wait got a reply from umteenth ticket [08:30] twice in one year w000t [08:31] wtf??? [08:31] Dear Anton [08:31] [08:31] This is a 10 Mb service the speed that is hitting is fair as per the contention ratio of the product. So we need to close this ticket. [08:32] wow [08:32] Contention ratio = NeoInternet Standard 25% [08:32] 1m isnt 25% of 10m [08:32] Nope [08:33] ask them can you send them a calculator to work out the percentages [08:35] hahaha [08:35] talk to someone in charge there not their help desk [08:36] Ja we have "customer representative" [08:41] its sad how peeps get away with shoddy service [08:42] Yea, but I think Neotel has had the wrong leadership, like the country. [08:50] neotel, what a disaster [08:50] You telling me [08:50] pretty sure there was some behind the table non-compete stuff going on there [08:51] we are in a country where the entire market hates telkom - the market was there to take for any competent second national operator [08:51] bungling on that level can only be deliberate [08:52] deliberae or not, it's shocking to say the least === hateball_jnr is now known as zipper [10:31] anton_may: pptp goes over TCP right? [10:32] Dropping packets is a normal way of slowing down tcp, but the double changing window size fucks things up [10:32] http://sites.inka.de/bigred/devel/tcp-tcp.html [10:32] anton_may: can you do some UDP speed tests? [10:41] tcp and gre [10:42] just finish lunch then will do [11:08] serverside - https://bin.snyman.info/mmmjatw8 [11:09] clientside - https://bin.snyman.info/mmmp6sdt [11:09] while we moaning about neotel. I'll def agree. Using neotel uncapped 4mbps. And their service is shocking and staff clueless. :( Anyway, at least FTTH is nearby for me. [12:04] * theblazehen likes crystal web, even if security isn't the best [12:06] yea they have a very stable network [12:23] Sorry, didn't do in sudo mode, previous result. Correct result here [12:23] serverside - https://bin.snyman.info/mmmwvy3b [12:28] clientside - https://bin.snyman.info/mmmc8wfh [12:37] Docker doesn't wanna work right in a lxd container :( [12:37] hi andrewlsd [12:38] hi theblazehen [12:39] layers. [12:39] hmm. [12:39] It should work. Could be a xenial kernel issue [12:39] what isn't working? the port forwarding stuff? [12:39] 0/ andrewlsd [12:39] https://github.com/opencontainers/runc/issues/769 [12:39] \o anton_may [12:39] Docker compose of https://github.com/mirumee/saleor [12:40] @ theblazehen udp speedtest result are in [12:40] aah. hmm. docker *compose*... that might be the issue. [12:40] ty anton_may. Will look more this evening [12:41] andrewlsd: Problem with the build script or whatever it is you think? Or what [12:41] https://gist.github.com/336f1e5f60e0ae8a8db22001c26d1ac3 [12:42] yea don't worry, already seeing abysmal results [12:42] Yup, build hasn't failed yet on desktop. Sucks. Might need to set up a kvm for docker then :/ [12:42] I applied the docker profile to the lxd container [12:43] I've actually never used docker inside LXD, although I know it is *supposed* to work [12:43] The test docker image worked [12:44] is your LXD container perhaps needing to run specially privileged? [12:44] in order to run a privileged docker image? [12:44] I think docker-compose should complain? And it failed at an apt-get install, so I doubt it [12:45] https://en.wikipedia.org/wiki/Coda_(file_system) seems interesting [12:48] * theblazehen is doubtfull of the 'S' in my SSHD :( [12:55] brb [12:55] * theblazehen is starting to regret buying 32 GB ram for current vm host. Should have gona new mobo + cpu with 64 GB maybe.. [12:55] http://bellard.org/linmodem/ :D I hope this works [12:56] > last update: Mar 6, 2000 [12:56] Well then [12:57] Eh. lxd can live migrate. Why not both? :D 96 GB ram ought to be enough for everybody [12:59] As can kvm [12:59] But then I need shared storage.. [12:59] Guess I could then get *another* box running dragonfly bsd with hammerfs :D [12:59] But I think 10 gbit hardware is expensive :( [13:00] And I'd need another box for storage. Current motherboard was decently expensive but only 6 sata ports.. [13:05] Does anyone know how to get the web results from a search using DuckDuckGo API? I can only seem to find help with its instant answers which isn't what I'm looking for... [13:08] paddatrapper: https://github.com/thibauts/duckduckgo [13:09] theblazehen: Thanks [13:13] theblazehen: get lots of gigabit nics, and bond them [13:13] theblazehen: also "mooseFS" [13:17] andrewlsd: How much would a switch be that does that? Or directly connect storage to both the hosts? [13:18] andrewlsd: Opioion on lizard vs moose fs? [13:19] And currently latency between desktop to host is 0.8 ms (wow, quite slow..), so don't really need local storage on each host [13:19] And why mose over gluster? [13:20] no idea. [13:20] theblazehen: any desktop switch could do it [13:21] unmanaged if you use the right "mode=..." bit for the bond [13:21] Ah, yes. was thinking of bonding 10 gbit.. [13:21] :D [13:21] if you want link-aggregation 802.3ad, then you need a managed switch [13:22] but for xor or round-robin load-balancing you don't. [13:22] theblazehen: I can't remember between Lizard and Moose. Lizard is a newer fork of Moose, IIRC.but I also don't know what benefits/requirements they have as compared to Ceph, or Gluster [13:23] Gluster seems to have quite VM hosting performance. [13:23] why not just use NFS share? [13:24] as long as the VM's image file isn't being written to in two places simultaneously you should be good to go with NFS. [13:24] andrewlsd: Yeah, was thinking nfs share with dragonfly bsd as host. But need to bond or use 10 gbit nics [13:25] even if using Gluster internal disks shared between two hosts you'd need at least 3 GBit in order to be able to run more than a couple of VMs [13:25] Because ssd > 1gbit [13:26] Ok [13:26] so I'd recomend looking for a quad-port gigabit LAN card. [13:26] and then just back-to-back cable them [13:26] then bond, and watch those lights flash [13:27] Yeah :D [13:27] Then I just need to find VMs to host.. [13:28] lol [13:28] not much you can do with 1 mbit upload [13:28] true dat [13:28] imagine how much I can do with less than half of that. [13:28] * theblazehen wants to upgrade to 20 mbit adsl. Then 2 mbit up :D [13:29] Oh come on.. If both your normal install instructions and your docker-compose instructions don't "just work" then I'm kind of less willing to use your product. [13:31] theblazehen: you just need a gentle 50Mbit each way fibre link [13:32] Yes, that would be *sufficient* [13:32] theblazehen: you referring to the Python Storefront. [13:32] andrewlsd: Yeah [13:32] theblazehen: at least, it would be sufficient initially [13:33] telkom sucks. They say the uncapped lte won't come to our area. Now it's sold out [14:44] `via socket in "/var/run/postgresql" at port "5432"` Now is it a socket or a port I'm connected to? Make up your mind.. [14:51] FFS!!! I swear Neotel...urg [14:54] * has quit (Ping timeout: 256 seconds) [14:54] and nothing was running on the line now [14:54] everybody has left [14:55] all servers are synced up [14:56] hopefully everything stays up untill i get home [14:59] anton_may check if they have a twitter account and rev them there then they fix things quick because the whole world sees [16:16] Maaz coffee on [16:16] * Maaz puts the kettle on [16:16] later all. home time [16:20] Coffee's ready for Kilos! [16:20] Maaz ty [16:20] You are welcome Kilos [16:46] Maaz: coffee on [16:46] * Maaz starts grinding coffee [16:50] Coffee's ready for superfly! [17:45] Maaz: ag no man, now my coffee's cold [17:45] superfly: Excuse me? [17:48] Maaz: Strong coffee please [17:48] theblazehen: Huh? [17:49] Maaz: 2x ocfee please [17:49] theblazehen: Sorry... [17:49] Maaz coffee please [17:49] theblazehen: There isn't a pot on [17:49] Maaz coffee on [17:49] * Maaz washes some mugs [17:49] * theblazehen clearly needs a "ocfee" [17:53] Coffee's ready for theblazehen! [17:57] ty Maaz [18:05] hi all [18:07] hey magespawn [18:08] Hello theblazehen magespawn Kilos Kilos! [18:09] and everyone! with the chan admin, :p [18:09] hey paddatrapper [18:09] Oh hello paddatrapper , and theblazehen :p [18:10] Oh, whoops, pavlushka* [18:10] I know what happened, theblazehen [18:10] lol [18:11] can any body help me with bug 1590771 [18:11] lp bug [18:12] * pavlushka beckoning inetpro ;) [18:40] too cold to sit and chat and cough [18:42] what bug is that pavlushka [18:43] Kilos: get a blanket, and usb modem bug in Xubuntu. [18:43] network manager? [18:47] yes [18:47] what isnt it doing [18:49] https://www.reddit.com/r/sysadmin/comments/4nakyq/what_are_the_chances_fitbit_while_changing_a_ups/d42b3fg not a bad idea [19:07] night all. sleep tight [19:14] good night [19:14] \exit [19:23] Goodnight everyone! [19:36] sup all :) [20:12] hey nlsthzn [20:12] what's new theblazehen ? all well I hope? [20:13] nlsthzn: Not much, just working.. Yeah, good ty :) And you? [20:13] all good thanks :) [21:18] Well, time for bed. Night all. [21:18] My alarm for tomorrow: `echo while:; do sudo beep; done | at 5 AM tomorrow` [21:18] On phone I can just swipe it off [21:19] Now I'll need to log in to ps [21:19] pc* [21:19] then ps fax, find the pid and kill it [21:19] Hopefully it'll wake me up enough to know that coffee will help [21:25] good night theblazehen [21:25] and I hope your alarm works/worked