=== Trixar_za is now known as Trix[a]r_za [05:12] morning [05:25] morning guys [05:29] good morning early birds [05:29] hey magespawn [05:29] whats up Kilos [05:30] ? [05:30] same old here, and there? [05:31] pretty much, have a transfer today so going to be on the road for awhile [05:31] transfer of aminals [05:31] no people, from richards bay to thanda [05:32] thanda? [05:32] or Thanda, a private game reserve [05:32] aha [05:32] you like a taxi service as well [05:33] yes sometimes [05:33] hehe [05:33] hope you got better road manners than the rest of the taxis around [05:33] but i think it costs a little more for somebody like me that your average taxi [05:33] well i hope so to [05:33] too [05:34] go safe lad and enjoy [05:38] when you leaving magespawn ? [05:38] only at 12:00 [05:38] ah thats good [05:39] have you any idea how to tell ubuntu to allow another wired pc to use your 3g connection [05:39] do they get linked here or what [05:40] not sure, you would have to share the connection somehow [05:40] maybe look in network manager [05:41] ya i made wired shared to others but 3g dont go with [05:41] i will try find more info from you guys friend [05:42] https://help.ubuntu.com/community/Internet/ConnectionSharing [05:42] ty [05:43] have had problems sharing a connection, seems a bit flaky [05:43] but i might be doing something wrong [05:43] lol [05:43] oh well well get there sometime [05:44] http://learnedstuffs.wordpress.com/2012/06/09/internet-connection-sharing-on-ubuntu-12-04/ [05:44] you are using 12.04 now? [05:44] gracias [05:54] hrm [05:54] does anyone know how accurate ftime() is under linux? [05:54] lo Squirm [05:55] or more specifically, does anyone know how to get down to millisecond (and preferably nanosecond) clock accuracy in ANSI C in Linux if ftime() isnt the way to do it [05:55] morning Symmetria [05:55] and I don't know [05:55] morning squirm, kilos etc :) sorry for the early questions, just getting frustrated at something [05:56] sjoe Symmetria you really wanna cut it fine [05:56] kilos heh, busy writing a network testing app [05:56] what you wanna check in nanosecs [05:56] that requires a high degree of accuracy [05:56] ah [05:57] and lol, even nanoseconds isnt that accurate :) you know they have managed to get atomic clocks synched to an accuracy of 10^16th of a second? :) [05:57] *THATS* accurate [05:57] wow [05:57] lol, they wanna get it to down to 10^18th of a second but the fiber distances you need to accomplish that are... insane :) [05:58] hmm they dont believe in " Môre is nog n dag" [05:59] maybe they do, they just want to know exactly when it starts [05:59] haha [05:59] heh kilos there are actually pretty good reasons for high accuracy clocks, mostly in the astronomy field [05:59] but they hit a problem after they managed the 10^16th thing [05:59] aha [06:00] under standard time definitions, you can only drop to 9 billionths of a second [06:00] :P so now they are trying to get the definition of a second redefined to cater for the 10^16th of a second [06:00] :p [06:00] cause thats outside of conventional definitions [06:02] but right now I'll be happy if I can get down to the millisecond level, and exstatic if I can get down to the nanosecond level [06:02] what are you measuring Symmetria? [06:02] dont you know anyone at nasa [06:04] magespawn the theory behind what Im developing is almost a type of ping tool, except, its dual ended and measures both forward and verse path latencies, as well as a host of other things based on packet size and number of packets on the wire and itterations through the loop it creates [06:05] heh, it gets a little tricky though, and it requires that the servers also have properly synched clocks :p [06:05] (though clock synching is easy, we're busy putting up some GPS based NTP servers for testing purposes, a number of them) [06:09] i think i understood maybe 1 word in 2 in that paragraph [06:11] magespawn lol basically, Im injecting a number of packets on one side, they send to the other side, the other side (provided the clocks are properly synched), can read the time stamp at time of sending outta the packet, figure out how long it took to get there, then reverse the source/destination address, change the time stamp and send it back [06:11] and you get a forward and reverse path latency reading [06:12] then, it flips back and forth through a set number of times to get a baseline [06:12] double ping? multiple ping? with accurate synced time stamps [06:13] magespawn, ping reads RTT [06:13] (round trip time) [06:13] I dont want round trip time, I want forward time and reverse time [06:13] to check for path a-symmetry [06:14] and I want to be able to check that through a range of packet sizes and with a set number of packets on the wire similtaneously [06:14] this only does it in one direction until it gets sent back? [06:15] magespawn basically, the theory behind what Im looking at is this [06:15] sorry lagged a bit there [06:15] machine a creates a packet, it puts a payload in the packet including the starting time right before transmit [06:15] it sends it to machine b [06:16] machine b looks at the timestamp in the packet and compares it to its own clock [06:16] and says "it took X number of milliseconds for me to get this packet" [06:16] thats the forward time [06:16] then it flips the source and destination address and sends it back with a new timestamp [06:16] and machine a goes "it took X number of milliseconds for it to get from machien b to machine a" [06:16] thats the reverse path time [06:18] then, it repeats that process X number of times until either the packet goes missing on the wire or until it finishes its itterations in a loop which gives it a baseline [06:19] its tricky though, because the servers have to have very accurate clock synchs for it to work [06:20] or else you could add/subtract to the transit time just because of a diffrence in time [06:21] exactly :) [06:24] heh, looking at putting down 4 stratum-1 servers for synch in africa though [06:32] o [06:33] and you're going to keep them in sync with gps? [06:33] it would be really interesting to see the result of your pings [06:36] hi Kerbero [06:36] hi kilos [06:42] Kerbero yeah [06:44] Symmetria: do you have any technical specs on how that is done [06:44] as one can't use a normal gps/rs232 port to get that accurate times from the gps [06:45] ^ nmea strings [06:45] *yawn* [06:45] what a wet day [06:45] Kerbero Im trying to figure that out at the moment :) doing some reading [06:45] but i guess one would be able to use the rts etc lines on the serial port to input a clock line from the gps [06:47] Kerbero you can actually buy a gps that is a timing specific gps that is a PCI board [06:48] gps's from what I read are calbirated either for navigation OR for time, they arent the same, and the timing specific gps's you can get as seperate boards [06:49] would you need one per server? [06:50] magespawn if you had the servers both synched to multiple stratum-1 ntp servers, they should be in sync enough to allow this [06:50] since ntp compensates for latency between themselves and the ntp servers [06:50] ahh right that answers the nest question [07:07] good mornings [07:08] hey inetpro [07:08] no meetings? [07:10] eh [07:10] magespawn: at the moment? no [07:10] time for a cuppa coffee [07:11] Maaz: coffee on [07:11] * Maaz washes some mugs [07:11] well at least things are looking up on that front then [07:11] lol [07:11] Maaz coffe please [07:11] magespawn: Go get it yourself! [07:11] hah [07:11] Maaz coffee please [07:11] magespawn: Alrighty [07:11] Maaz, coffee please [07:11] Kilos: Okay [07:11] hi inetpro [07:12] Kilos: hi [07:12] at least you guys are still awake [07:12] lol [07:12] i go sort sheep [07:12] or should I rather say alive and kicking [07:12] we need more rain [07:13] Kilos: remember to check for the ticks [07:13] im not catching every one and lifting tails to look [07:13] :-) [07:14] hehe [07:15] Coffee's ready for inetpro, magespawn and Kilos! [07:15] Maaz, ty [07:15] You are welcome Kilos [07:16] Maaz ty botsnack [07:16] thanks, magespawn [07:34] Maaz botsnack [07:34] Yay, at last someone thinks about me as well [07:38] hrm [07:40] anyone awake who can do me a favour? [07:40] http://pastebin.com/AL8p6YFw <== someone take that code, compile it on a linux box, and run it, and lemme know how many itterations per second its claiming and what cpu you are using [07:40] depends on what, so ask away? [07:40] (I only have one box I can test on, and I wanna make sure that on another box Im getting different results) [07:41] this old server Im testing on is giving me like, 15k runs per millisecond [07:41] he pastebins are sloow [07:41] that or our it broke something again [07:42] (and yeah, I know that code is not particularly accurate in what it does, but *shrug* its an ok test of the function) [07:43] Tonberry lol I can email you the code if you want [07:43] it loaded fine the second time [07:43] strange intermittent issue we seem to be having [07:44] looks like some tcp connections just never connect or do anything [07:44] heh [07:44] browser will just keep loading until timeout [07:44] what sorta results re you getting from that code [07:44] but stop and refresh and it shows instantly [07:44] feels like gprs.. [07:45] Executed 27620 increments in 1 milliseconds [07:45] Executed 28366 increments in 1 milliseconds [07:45] Executed 28041 increments in 1 milliseconds [07:45] End time: 1348213520.331 [07:46] seems to stay more or less around that increment count [07:46] model name : Intel(R) Core(TM) i5 CPU 760 @ 2.80GHz [07:46] for my cpu [07:49] hrm [07:49] ok well, if its pretty consistant thats good [07:50] you are testing how many times you can increment a number in a millisecond? [07:50] tonberry was actually just screwing with millisecond timers :) [07:50] ah [07:54] you can get microsecond time with gettimeofday btw [07:55] tonberry heh various documents on why that isnt accurate though [07:55] i assumed as much [07:55] i have seen a setting in my bios about fine precision timers [07:56] *high [07:56] question then is if the kernel knows or cares about them [07:56] indeed [08:03] oh shucks, I forgot me coffee [08:03] thanks Maaz [08:03] hehe [08:09] heh Tonberry I still have no clude how to get down to below millisecond timing [08:09] would be kinda cool if I could get to nanosecond :p [08:10] nano on a pc will be a massive pain i think [08:10] lol, I still wanna know how those guys got to 10^16th of a second [08:10] and actually understand it [08:12] with the right hardware anything is possible [08:16] Symmetria: tell us if you find out [08:18] inetpro lol, there is a paper on it online somewhere, will find it [08:18] but its so far beyond my comprehension maths wise [08:18] they were using optical crap to do it [08:19] I just would not know why you'd get into that much detail, so it's beyond m comprehension as well [08:19] inetpro heh, I can explain that fairly easily, do you know what eVLBI is? [08:19] no [08:19] old peeps dont mind being milliseconds late [08:19] (its just one example of where you need it) [08:20] umm ok, eVLBI is whats known as extra very long baseline inteferometary (sp?) [08:20] its a branch of astronomy science [08:20] basically, what they do is point a bunch of radio telescopes to the same point in the sky and collect white noise [08:20] then, they send the white noise to a supercomputer that overlays the signals over each other [08:20] now, white noise cancels white noise [08:21] so, whatever is LEFT is an image of sorts [08:21] but that noise has to be overlaid very carefully, and that requires timing to figure out the overlay [08:21] the more accurate the timing, the more accurate the overlay, the better definition on the image [08:23] heh there are numerous branches of science that use that kinda timing [08:24] ok gotta take doggy to the vet quick :) back in a bit [08:27] Symmetria: ahh, makes sense [08:29] basically what the SKA will do [08:29] and some other old dishes throughout africa [08:41] sounds like an interesting challenge === Trix[a]r_za is now known as Trixar_za [09:28] Start time: 1348219641.760 [09:28] Total of 152840601 itterations completed in 10000ms, average of 15284 per ms [09:28] End time: 1348219651.760 [09:28] lol [09:28] there, I cleaned up my code [09:30] paste? [09:31] http://pastebin.com/PnMT3r0Q [09:31] you can change the MILLISECONDS line to specify how long you want it to run for === Trixar_za is now known as Trix[a]r_za [10:30] hi Trix[a]r_za psyatw [10:32] hi Kilos [10:32] hi Trix[a]r_za [10:32] and everyone else [11:20] hrm [11:20] anyone here who can do me a favour and compile something and run it to send me 100 small packets, Im trying to test something [11:21] http://pastebin.com/nbE6nYuh [11:21] that code [11:22] :P and preferably make sure that your box has a synched clock first [11:22] else its not gonna work well :p [11:40] hi Symmetria [11:40] it appears to work on Ubuntu 10.04.4 x64 [11:49] Symmetria, i can [11:49] have only 10mins though [11:54] well i need to go show the undergrads how to program\ [11:54] Symmetria, i'll be back at 17h if you need help [11:57] hi Kerbero [11:58] o [11:58] een nederlander [11:58] goedemiddag [11:58] ik moet nu gaan [11:58] doei [11:59] doei [12:32] http://pastebin.com/chfCHqjC [12:32] for anyone interested [12:32] note: that code will only work if the boxes you are using are both properly ntp synched [12:32] if the timing is out on the boxes that code will break [13:39] anybody used the mechanize module in python [13:48] heh wheeeeeeeee [13:48] http://pastebin.com/kD3WvEpY [13:48] bug fixed code :) [13:48] lol havent COMPLETELY forgotten how to code it seems [13:51] "BREAKING: I've been told that an arrest warrant for Julius Malema was issued this morning. Very reliable info. Appearing in court next week." [13:51] http://www.news24.com/SouthAfrica/News/Breaking-Arrest-warrant-issued-for-Malema-20120921-2 [13:55] heh [13:55] that money laundering charge [13:55] is interesting [13:56] will be interesting to see what happens [13:56] lol, he will go in, and if the courts have any sanity, be denied bail [13:56] :p [13:59] hah [14:02] heh [14:02] it also means by next week [14:03] his assets will be frozen [14:03] as is standard procedure [14:03] at that point, if he continues to travel around, the courts will have grounds to ask who is funding it [14:09] hrrrmmmm [14:09] Zuma is funding his travelling tendencies [14:12] not a chance [14:12] zuma wants malema at the bottom of the limpopo river [14:12] :p [14:12] lol [14:13] trial will go on forever [14:13] will prob be found uilty [14:13] malema will claim sudden illness [14:14] *guilty [14:14] NOT 1 day in jail [15:28] ooooh, New Mumford and Sons :D [16:50] hrm [16:50] anyone awake? [16:53] yes? [16:56] Kerbero [16:56] do me a favour [16:56] make sure your box is ntp synched [16:56] then... [16:56] download this: [16:56] http://pastebin.com/KrncXZTW [16:56] compile it [16:56] 1. duh [16:56] and run it with ./client 196.32.210.12 500 50 [16:57] :) [16:58] :~$ sudo service ntp stop [16:58] * Stopping NTP server ntpd [ OK ] [16:58] :~$ sudo ntpdate ntp.sun.ac.za [16:58] 21 Sep 18:57:57 ntpdate[1038]: adjust time server 146.232.128.1 offset 0.001488 sec [16:58] :~$ sudo service ntp start [16:58] * Starting NTP server ntpd [ OK ] [16:58] did it run? [16:59] you might be firewalled sending that shit :( [16:59] cause I aint seeing packets :p [16:59] need to compile still [16:59] aaahh [16:59] lol [16:59] that would be why [16:59] :) [16:59] it should compile clean under linux, its broken under fbsd :P [16:59] because fbsd doesnt include ftime() [16:59] ok here we go [17:00] *hrm* ok, do me a favour, ping that ip address [17:00] ran [17:00] and tell me what times you getting [17:00] jip it pings [17:00] what latency? [17:00] 207-209 [17:00] *HRM* interesting [17:01] lower latency from you -> me then from me -> you [17:01] your forward latency is around 90 - 100 [17:01] wow [17:01] hmm [17:01] just a sec [17:01] Received packet. Packet delay calculated at 91 ms [17:01] Received packet with 82 bytes of data with sequence number 498, padding size is 50 [17:01] Received timestamp 1348246822:998 [17:01] Received packet. Packet delay calculated at 91 ms [17:01] Received packet with 82 bytes of data with sequence number 499, padding size is 50 [17:01] Received timestamp 1348246823:48 [17:01] Received packet. Packet delay calculated at 99 ms [17:01] can i run again? [17:01] go for it [17:02] heh, whatever you did stabilized the latency / dropped it slightly [17:02] 89 - 90 pretty constant [17:02] indeed [17:02] :P you can dig through that code and figure out how it works, though I wanna put code in both sides to actually do some clock negotiation [17:03] ye [17:03] rather than relying on the ntp clocks being right [17:03] but I have no idea how to code that clock negotiation (yet) [17:03] but you see my clock is not out more than 2ms [17:04] *nod* [17:04] the problem is, mine may be, because the closest stratum-1 servers I have are at least 30 ms away [17:04] hmm [17:04] +ntp1.tdc.fi .PPS. 1 u 31 64 377 33.237 -1.958 1.661 [17:04] +ntp2.tdc.fi .PPS. 1 u 26 64 377 31.939 -1.853 1.694 [17:04] *ntp.meerval.net .GPS. 1 u 30 64 377 23.410 -7.253 2.462 [17:04] ntp.sun is 2ms away [17:05] what stratum is ntp.sun? [17:05] no idea [17:05] definitely not 1 [17:05] :P [17:05] do an ntpq -p [17:05] :p [17:05] *maties1.sun.ac. 196.25.1.1 3 u 45 64 177 0.977 -0.459 0.508 [17:05] ouch [17:05] lemme give you some .za 1's [17:06] ntp server 146.64.8.7 [17:06] ntp server 146.64.58.41 [17:06] use those 2 [17:06] they are both in .za and both stratum-1s [17:06] oh, and stratum1.neology.co.za has address 41.73.40.11 [17:06] there :p you can run 3 [17:06] :) [17:08] well i doubt that ntpd will sync with them immediately [17:08] almost immediately :) [17:08] it take under 5 seconds :) [17:08] ok [17:08] if you run a ntpq -p you will see if they are synched :) [17:08] if they arent they will say .INIT. [17:09] ahh [17:09] delay offset and jitter are all 0 [17:09] ntpd will build that up over time i guess [17:09] it still saying INIT on them? [17:09] or what? [17:10] yes [17:10] ok it aint inited yet [17:10] should change in a coupla seconds [17:10] unless your firewalling blocks it [17:10] it's stellenbosch.... [17:11] yes [17:11] LOL so still on init? [17:42] hi smile [17:42] hi Kilos :p [17:43] brb :) [17:59] kilos wanna help me test something? I got some tests from kerbero, but I want some more tests :) [17:59] ok [18:00] just dont ask me to compile stuff [18:00] http://pastebin.com/KrncXZTW <=== haha you need to compile that :p [18:00] tell me how and lets go [18:00] ok, get that shit off pastebin, and stick it in a text file [18:00] i dunno what compile is [18:00] then type gcc -o client textfile [18:00] :p [18:00] ok sec [18:00] just make sure you copy it from the raw paste so you dont get all the line number bullshit in it :) [18:02] wow that pastebin is slow [18:02] paste it to slexy.org rather [18:02] nm it opened [18:03] :) [18:05] in place of textfile i gotta put path hey [18:05] well, whhat file did you save the stuff in [18:05] cause its that filename you change textfile to [18:05] its on my desktop [18:05] then yeah you put the full path to the filename [18:07] grr [18:08] /home/miles/Desktop/textfile: file not recognized: File format not recognized [18:08] collect2: ld returned 1 exit status [18:08] nono, gimme the full command you are typing [18:08] to compile that textfile [18:08] it should be gcc -o client /home/miles/Desktop/textfile [18:08] which should make a file called client [18:09] gcc -o client ~/Desktop/textfile [18:10] ok did that work [18:10] it should have made a file called client [18:10] nope [18:10] gcc -o client /home/miles/Desktop/textfile [18:10] /home/miles/Desktop/textfile: file not recognized: File format not recognized [18:10] collect2: ld returned 1 exit status [18:11] is it just the text from the pastebin in that textfile [18:11] yeah i copied the bottom window [18:11] *HRM* lemme email you the file as an attachment :) [18:12] whats your email address [18:12] msdomdonner@gmail.com [18:13] when i open your file it has coloured text in it [18:13] Symmetria perhaps have Kilos cd to where the file is [18:13] hey Kilos [18:13] yes [18:13] kilos I emailed you client.c [18:13] :) [18:14] ok i got client.c on desktop [18:15] ok, now try gcc -o client /home/miles/Desktop/client.c [18:15] what now [18:15] lol [18:15] and tell me what it says [18:15] /home/miles/Desktop/client.c: In function ‘main’: [18:15] /home/miles/Desktop/client.c:58: warning: format ‘%s’ expects type ‘char *’, but argument 2 has type ‘char (*)[20]’ [18:16] thats fine, you can ignore that :) [18:16] did it make a file called client? [18:16] where? [18:16] in the directory where you ran that gcc command [18:16] there should be a file called client [18:17] on Desktop? [18:17] if there is, type ./client 196.32.210.12 50 50 and it should start spewing out a bunch of stuff [18:17] well, what directory where you in when you typed the gcc command? [18:18] cli on desktop [18:18] if you ran that from terminal, just go back to teh same terminal window and type the ./client 196.32.210.12 50 50 command [18:18] yeah go back to cli [18:18] no client file here [18:18] to the same place [18:18] *HRM* [18:18] ok, go back to the cli in desktop [18:18] and type gcc -o client client.c && ls -l client [18:18] and tell me what it says [18:18] :) [18:19] miles@P4:~$ gcc -o client client.c && ls -l client [18:19] gcc: client.c: No such file or directory [18:19] gcc: no input files [18:19] *HRM* [18:19] oh you arent in desktop [18:19] hehe [18:19] oh where am i [18:19] you're in your home directory :) [18:19] type cd ~/Desktop [18:19] and then rerun the gcc -o client client.c && ls -l client [18:19] :p [18:20] heh the ~$ says you were in your base home directory :) [18:20] client.c: In function ‘main’: [18:20] client.c:58: warning: format ‘%s’ expects type ‘char *’, but argument 2 has type ‘char (*)[20]’ [18:20] -rwxr-xr-x 1 miles miles 7795 2012-09-21 20:20 client [18:21] aha [18:21] now type ./client 196.32.210.12 50 50 [18:21] :) [18:21] its green [18:21] in the cli window in the ~/Desktop directory [18:21] :) [18:21] holy moly [18:21] how much of that you wanna see [18:22] heh, wait, need you to run it again :) then will see it on my side, lemme just get back to the server [18:22] ok run the same command again [18:22] and tell me when you've run it :) [18:22] aahh there we go [18:22] I see it [18:22] done [18:22] :) ok, firstly, your clock is out [18:23] uumm [18:23] type ntpdate stratum1.neology.co.za [18:23] give command to sync it [18:23] that should sync it at least temprorarily [18:23] :) [18:23] then rerun the command :) [18:23] I recon you're about 700ms out before the ntpdate command [18:24] (700ms slow) [18:24] ok synced here comes the run thing [18:24] once you've run that ntpdate command, it should sync, and when you rerun the client command, the readings should be more accurate [18:24] k [18:24] bingo :) [18:24] Received packet. Packet delay calculated at 75 ms [18:24] Received packet with 74 bytes of data with sequence number 48, padding size is 50 [18:24] Received timestamp 1348251881:858 [18:24] Received packet. Packet delay calculated at 64 ms [18:24] Received packet with 74 bytes of data with sequence number 49, padding size is 50 [18:24] Received timestamp 1348251881:908 [18:24] Received packet. Packet delay calculated at 75 ms [18:24] you have a pretty low outbound latency to europe :) [18:25] dont tell me [18:25] fix it [18:25] heh, the problem with this is the clock shift between yourself and myself [18:25] kilos lol, actually low is good [18:25] high would suck :) [18:25] oh haha [18:26] heh, what thats basically saying is its taking 75 1000ths of a second [18:26] for a packet to go from you to london [18:26] whew [18:26] providing the clock syncs are accurate on both sides :) [18:27] heh, you have some jitter on your client though which isnt so nice [18:27] Received packet. Packet delay calculated at 68 ms [18:27] Received packet with 74 bytes of data with sequence number 37, padding size is 50 [18:27] Received packet. Packet delay calculated at 77 ms [18:27] Received packet with 74 bytes of data with sequence number 38, padding size is 50 [18:27] heh ideally, you want something that has a very consistant delay [18:28] cause that variation in delay is whats called jitter [18:28] (It wont have a negative effect on you, but in certain situations that can be nasty) [18:28] kilos heh, basically, what that software you ran does [18:28] what is causing it [18:28] is get the time your machine [18:28] then stick it into a udp packet [18:28] and then the udp packet to me [18:28] I then compare that timestamp to the timestamp on my server [18:28] and calculate the difference [18:29] is your server in london [18:29] your machine sends one packet every 50 milliseconds for however many number of packets you tell it (50 is what you ran on the cli) [18:29] yeah [18:29] then the other setting (the second 50 you specified) is the packet size [18:29] we basically pad the packets by 50 data bytes just to make it a more realistic test [18:30] (well, its a pad of 50, actual packet size is 74 bytes) [18:30] magespawn, you watching this [18:30] as I say, the PROBLEM with that code is that it relies on the clocks being identically synched [18:30] and thats extremely difficult to do because of asymmetrical routing to the ntp servers [18:31] (ntp works by getting a time from the ntp server, working out the latency between you and the ntp server, dividing the latency by 2, adding that to the time its got and setting the time) [18:31] (though using multiple servers it kinda averages them and it gets better accuracy) [18:32] so where did we sync my time [18:32] to johannesburg [18:32] ah [18:32] ntpdate will synch it but not very accurately, to get high accuracy you need to install ntpd [18:32] if you want time synced more accuratley why not sync to the same server on both sides [18:32] apt-get install ntp, then edit the config in /etc/ntp.conf and replace the server lines in there [18:33] * doei! :) [18:33] kilos heh, because the latency between the time server and your ntp client has a noticeable effect [18:33] ah [18:33] and if you're using stratum-1 servers, the clocks on those servers are probably within 2 1000ths of a second anyway [18:33] (a stratum-1 server = its setting its clock via a GPS receiver) [18:33] eish i cant even think that small man [18:34] you cant snap your fingers that fast [18:34] the only thing really more accurate than a stratum-1 server is an atomic clock and lol, if you think a thousandth of a second is small, thats still huge [18:34] good night, Kilos :) [18:34] kilos, lol, they managed to sync 2 atomic clocks to an accuracy of 10^16 of a second [18:34] night smilsleep tight [18:35] thats 100000000000000000th of a second :P [18:35] that is rather on the accurate side hey [18:35] lol, they wanna get it to 10^18 but I dont see them doing it any time soon [18:36] the complexity of getting to the 10^16th is already crazy and requires like, 900 kilometer long optical fiber links to work with [18:36] (they know the exact speed of light through optical fiber so they can use long lengths of fiber and time delays and dispersion etc to get accuracy) [18:36] how can that work on the internet [18:37] lol, well, thats something people are working on :P [18:37] there are entire papers on this [18:37] cant work i think [18:37] heh, on the general internet no, it doesnt work 2 well [18:37] but they have uses for it [18:37] service providers with mess up in between [18:38] wait lemme find this [18:40] http://www.ces.net/events/2012/cef/p/Comparison%20of%20clocks%20using%20optical%20fiber%20links%20in%20recent%20results%20and%20future%20projects.pdf [18:40] http://www.ces.net/events/2012/cef/p/Time%20transfer%20by%20using%20SONET%20connections%20between%20SP%20and%20Mikes%20in%20experiences%20and%20results.pdf [18:40] both of those are interesting :) [18:40] the first one you probably need a maths degree to understand half of it though :P [18:40] its way over my head [18:41] oh, they got to 10^18 [18:41] aint we got any wiz kids here [18:42] big storm here dunno how long power will hold out [18:42] lol not sure we have any at that level :p [18:42] i dont see where such accuracy can be used [18:42] stars move in years [18:43] if you see it tonight its in the same place tomorrow night [18:43] kilos heh, space time distances, overlaying of data etc [18:43] kilos, if you put 10 radio telescopes at the same point in the sky with the telescopes spread around the globe [18:43] you then take the white noise they collect [18:43] and you overlay it ontop of each other in a supercomputer [18:44] the white noise cancels itself out [18:44] and what you're left with is an image [18:44] that requires incredibly accurate time calculations to do the overlay properly [18:44] image of what [18:44] and the more accurate the overlay is, the higher the resolution of what you're looking at [18:44] same as mixing colours on a tv to make others [18:44] kilos well, its not really an "image" as you'd imagine, but basically, they are looking for stars that are very far away :p [18:45] and radio signatures and crap to find them [18:45] and tell how far away they are etc :) [18:45] and ever spoken to an astronomer? their "accuracies" can be a few million kilometers out [18:45] and sometimes a few hundred million kilometers [18:45] so they refine the clock accuracies etc to get more accurate in other measurements [18:46] ah [18:46] time is a rather key component of physics from what I understand (though my understanding of physics is almost non-existant and I openly admit that) [18:47] yes Kilos just read the logs [18:47] sorry we missed you help there magespawn [18:48] things go too fast for me [18:48] cd to desktop methinks it was [18:48] no worries [18:48] yup [18:48] i always battled to find the path to it [18:48] heh here is an interesting one [18:49] that ~ goodie works kiff [18:49] apparently the speed of light through fiber changes dependant on the temp of the fiber [18:49] :p [18:49] isnt the speed of light a constant [18:49] only in a vacuum [18:50] the density will affect the speed [18:50] ah [18:50] density is affected by temperature [18:51] if we cant get data to move that fast how they gonna get spaceships to go faster even [18:51] heh kilos speed of light through glass for example [18:51] is significantly slower than the speed of light through a vaccum [18:52] einstein's theory says it is impossible [18:52] aha [18:52] (if we didnt have to deal with the slowdown through glass, the internet would run better) [18:52] i thought fiber was hollow [18:52] not if vodacom and company are involved [18:52] magespawn heh, yes, einstein's theory says moving faster than light is impossible, but he doesnt discount the possibilities of bending space/time [18:53] magespawn *shake* fiber is glass, the light moves through glass, which means your latency is actually worked on about 2/3rds the speed of light [18:53] so we what is called einstien space [18:53] thats part of the reason why if you bend fiber 2 far [18:53] it breaks and cant be repaired [18:53] the glass shatters [18:53] and your fiber is fubar :) [18:53] i always thought it was more a tube than a solid [18:54] no, its glass :) very very very thin glass (strands are no thicker than a human hair) [18:54] if you could make it a tube you could make a partial vacuum and speed it up [18:55] :) maintaining a vaccum through it would be rough though [18:55] lol one of the universities whined at me once about the latency to london [18:55] :P I told them they had an awesome physics department, go figure out how to bend space/time [18:55] hehe [18:55] and when they had, come tell me, and I'd reduce their latency using their research [18:55] :p they werent impressed with me [18:56] he lol [18:57] i am off, good night all [18:57] night magespawn [18:57] sleep tight [18:58] later man [18:58] :P Im pondering hw to do DWDM through a vaccum [18:58] cause that would be awesomenes [18:59] lol [18:59] you on your new pc? [18:59] lol back later, Im gonna go contemplate that one over a beer :P and hehe yeah I got my new pc [18:59] its fast as hell :) [18:59] thats nice [19:00] enjoy your beer [19:02] night all . sleep tight === Trix[a]r_za is now known as Trixar_za === Trixar_za is now known as Trix[a]r_za