| 00:03.43 | sorphin | hugs/clear |
| 00:03.45 | sorphin | grr |
| 00:03.47 | sorphin | wtf |
| 00:03.48 | kergoth | heh |
| 00:03.55 | sorphin | dunno wtf happened there |
| 00:36.18 | *** join/#elinux eggers (n=eggers@cpe-70-112-81-8.austin.res.rr.com) |
| 00:59.31 | *** join/#elinux noclouds (n=mhfan@60.166.33.78) |
| 01:29.16 | *** join/#elinux Soopaman_ (n=soopaman@CPE00112f7ebeec-CM000a739b3c5c.cpe.net.cable.rogers.com) |
| 02:00.48 | *** join/#elinux weo1 (n=weo@p5499CC4E.dip.t-dialin.net) |
| 02:10.00 | *** join/#elinux Soopaman_ (n=soopaman@CPE00112f7ebeec-CM000a739b3c5c.cpe.net.cable.rogers.com) |
| 02:17.48 | sorphin | kergoth: ping |
| 02:17.57 | kergoth | sorphin: pong |
| 02:18.14 | sorphin | kergoth: what's an easy way to find what endian/byteorder you're operating it? |
| 02:18.16 | sorphin | in even |
| 02:18.38 | kergoth | check out /usr/include/endian.h |
| 02:19.31 | sorphin | right |
| 02:19.38 | sorphin | but the problem herein lies |
| 02:19.43 | sorphin | javascript and php |
| 02:19.50 | sorphin | one's on intel, one's on sparc |
| 02:19.57 | sorphin | and the encryption doesn't end up the same |
| 02:20.13 | sorphin | one obvious difference is possible endian difference |
| 02:22.24 | sorphin | woo |
| 02:22.28 | sorphin | php detection code |
| 02:47.28 | sorphin | kergoth: even converted between endians? :) |
| 02:57.44 | sorphin | nevermind |
| 04:04.06 | fishhead | http://www.isracast.com/tech_news/101005_tech.htm <--- kick ass |
| 04:44.54 | *** join/#elinux jipi (n=jipi@cm191.delta245.maxonline.com.sg) |
| 05:02.36 | fishhead | you guys might want to check out the article on slashdot right now if you own a digital camera or camcorder, apprentely 4 companies including cannon and sony have shipped a ton of cameras with ccd modules that fail resulting in distorted or no picture |
| 05:07.52 | *** part/#elinux joe_bleau (n=joe_blea@adsl-65-68-175-165.dsl.fyvlar.swbell.net) |
| 05:41.26 | *** join/#elinux markl_ (i=mark@dsl093-225-127.slc1.dsl.speakeasy.net) |
| 06:06.21 | *** join/#elinux darkschneider (n=gab@213-140-6-96.ip.fastwebnet.it) |
| 06:20.40 | *** join/#elinux cantona_ (n=cantona@unaffiliated/cantona) |
| 07:30.07 | *** join/#elinux cantona_ (n=cantona@unaffiliated/cantona) |
| 09:25.20 | *** join/#elinux ade|desk (n=adavey@194.200.143.249) |
| 10:40.46 | *** join/#elinux Soopaman_ (n=soopaman@HSE-Montreal-ppp143739.sympatico.ca) |
| 11:22.00 | *** join/#elinux toi (i=pleemans@tagung-233-171.tagung.uni-hamburg.de) |
| 11:51.43 | *** join/#elinux vrm (n=vrm@154.55.101-84.rev.gaoland.net) |
| 12:19.39 | *** join/#elinux kergoth (n=kergoth@mail.astrocorp.com) |
| 12:35.49 | T0mW | I dunno why I read news about linux stuff |
| 12:37.19 | T0mW | I find myself scrolling down the headlines going "yeah, uh huh, yeah, yeah" |
| 12:47.07 | *** join/#elinux prpplague (n=billybob@72.22.146.6) |
| 12:48.08 | prpplague | ~tuesday |
| 12:48.11 | ibot | Tuesday sucks, because it follows Monday (see monday). |
| 13:12.48 | *** join/#elinux GPSFan (n=Ken@65.121.49.208) |
| 13:12.57 | sorphin | T0mW: so does one's brain :) |
| 13:13.05 | sorphin | speaking of pluggin in their brain |
| 13:13.09 | sorphin | prpplague: morning dave |
| 13:15.10 | GPSFan | sorphin: prpplague: mornin.. |
| 13:15.22 | prpplague | hola folsk |
| 13:16.35 | prpplague | http://www.nationnews.com/story/9094130890482.php |
| 13:16.43 | prpplague | hehe island wide fiber soon |
| 13:25.23 | *** join/#elinux Soopaman (n=soopaman@209.82.37.5) |
| 13:25.29 | jbevren | dude I'm moving :) |
| 13:28.03 | prpplague | jbevren: shall we start playing the jeffersons theme song? |
| 13:28.55 | jbevren | dunno, does barbados have tall buildings with apartments in em? |
| 13:29.21 | jbevren | If there is a mistake...well, you should have used the 'Preview' button! |
| 13:29.22 | jbevren | hehe |
| 13:29.24 | jbevren | good ol slashdot |
| 13:29.31 | prpplague | about 8 story is about as tall as you get here |
| 13:29.40 | jbevren | heh |
| 13:30.48 | T0mW | jbevren: barbados cannot have tall buildings, they would act as sails during hurricanes and the island would be blown into the Brazilian coastline |
| 13:30.56 | prpplague | hehe |
| 13:30.59 | prpplague | T0mW: good one |
| 13:31.31 | prpplague | T0mW: cpld's , fast enough to emulate nand flash? |
| 13:32.00 | prpplague | T0mW: i asked yesterday, but don't recall if you answered |
| 13:32.14 | T0mW | prpplague: i did |
| 13:32.23 | prpplague | T0mW: ahh i'll check the log then |
| 13:32.46 | prpplague | T0mW: of course i have no experience with cplds and no equipment |
| 13:32.49 | prpplague | T0mW: :( |
| 13:32.54 | prpplague | T0mW: just a thought though |
| 13:33.20 | T0mW | prpplague: first some background on how a pld works: it works through product terms, each term (layer) takes the access time to propogate the signal throuh it. The more complex your logic terms the longer the delay through the device |
| 13:33.41 | T0mW | prpplague: having said that, you can get devices with < 7ns delay per term |
| 13:34.13 | T0mW | prpplague: having said that, you can get CPLD devices with < 7ns delay per term |
| 13:34.53 | T0mW | prpplague: if they aren't fast enough, you can go to something like the Altera Spartan line of FPGA logics, much faster terms |
| 13:35.22 | prpplague | T0mW: fpga seems alittle overkill |
| 13:35.55 | prpplague | T0mW: just want to convert the nand flash interface to something else, possibly spi |
| 13:35.58 | jbevren | altera didnt make spartan |
| 13:36.00 | jbevren | xilinx did :) |
| 13:36.08 | T0mW | prpplague: generally, the smaller the device in the number of gate terms, the more delay. When you go to a larger device, the compiler will/should optimize for speed by using adjacent terms to avoid excess prop delays |
| 13:36.17 | T0mW | jbevren: phffft |
| 13:36.25 | T0mW | :) |
| 13:36.27 | jbevren | prpplague: if youre looking for interface conversion, that's different from pretending to be flash :) |
| 13:37.17 | T0mW | I think what he wants to do is take something other than flash (EEPROM / RAM) and make it look like a flash media device |
| 13:37.24 | prpplague | jbevren: how so? |
| 13:40.06 | jbevren | prpplague: pretending to be flash requires onchip storage |
| 13:40.13 | jbevren | adapting an interface doesnt :) |
| 13:40.25 | jbevren | T0mW: yeah |
| 13:40.39 | *** join/#elinux eggers (n=eggers@pixpat.austin.ibm.com) |
| 13:42.35 | jbevren | prpplague: fyi, xilinx has a nice starter kit for $100 |
| 13:44.32 | jbevren | heheh |
| 13:44.59 | jbevren | http://www.xilinx.com/xlnx/xebiz/designResources/ip_product_details.jsp?sGlobalNavPick=null&sSecondaryNavPick=null&category=&iLanguageID=1&key=DO-SPAR3-DK |
| 13:45.05 | jbevren | their website suddenly sucks |
| 13:45.08 | prpplague | jbevren: too much for this project |
| 13:45.14 | jbevren | prpplague: its a devkit :P |
| 13:45.30 | jbevren | with enough gates to have some serious fun |
| 13:45.39 | prpplague | jbevren: yea i;ve looked at a couple of them before |
| 13:45.47 | jbevren | I may get one soon. Ive been holding out on the 3e kit (500k gates) |
| 13:45.53 | jbevren | but it was to be available in july |
| 13:45.56 | jbevren | and again in sept |
| 13:45.58 | jbevren | now its dec |
| 13:46.01 | prpplague | jbevren: cost wise they are kinda out of the league i need |
| 13:46.04 | CosmicPenguin | Ok.... so anybody know the punchline numeric series in todays Foxtrot? |
| 13:46.07 | jbevren | nod |
| 13:46.08 | CosmicPenguin | I'm math stupid |
| 13:46.21 | jbevren | CosmicPenguin: me too |
| 13:46.31 | T0mW | I recall when Motorola realized that people were taking their $200 dev boards for the 68328 processor and using them for production units. Mot was selling the dev kits at a loss to developers |
| 13:46.43 | prpplague | jbevren: i was thinking a small avr possible, just not sure if it can keep up with the speed i need |
| 13:47.07 | jbevren | ah, but an avr isnt a pld :) |
| 13:47.10 | T0mW | prpplague: probably not |
| 13:48.14 | prpplague | T0mW: the problem is that the JB otp rom doesn't appear to suffer from bitflipping so no checking is done |
| 13:48.14 | T0mW | prpplague: problem with a processor in that application that you would need a register file / fifo to interface with the outside world. CPU can't respond fast enough to I/O requests to make it appear as an I/O device |
| 13:48.32 | prpplague | T0mW: this is a problem since i want to use regular nand flash that does have bitflipping problems |
| 13:48.44 | T0mW | yeah |
| 13:48.49 | prpplague | T0mW: bummer |
| 13:48.49 | T0mW | get rid of the smartmedia? |
| 13:48.54 | jbevren | T0mW: that said, the 75mips ubicom sx is -almost- fast enough to soft-ulate a 6502 |
| 13:48.55 | jbevren | :) |
| 13:49.02 | jbevren | (said 6502 is at 1mhz) |
| 13:49.09 | prpplague | T0mW: can't find anything else with a serial nand interface |
| 13:49.25 | jbevren | prpplague: playstation memory cards? :-D |
| 13:49.36 | T0mW | jbevren: yeah, was thinking of using the SPI of the LPC2106 to directly drive a QVGA |
| 13:49.50 | prpplague | jbevren: those are nand flash as well iirc |
| 13:51.14 | T0mW | hmm, Mandriva-2006 DVD download finally finished, now to md5sum it |
| 13:51.56 | jbevren | mandrivia? |
| 13:52.17 | T0mW | mandrake |
| 13:52.21 | jbevren | ahh |
| 13:52.26 | jbevren | I thought maybe so |
| 13:52.56 | T0mW | this is some amalgamation of Connectiva, Mandrake and some other distro |
| 13:53.12 | jbevren | ahh |
| 13:53.24 | jbevren | a conglomeration of three distributions from groups that cant make their own distro? |
| 13:53.25 | jbevren | :) |
| 13:53.52 | CosmicPenguin | heh |
| 13:54.14 | T0mW | heh |
| 13:54.15 | jbevren | the last several times I tried an rpm based distro, it ended up being another redhat :) |
| 13:54.31 | jbevren | the last few times I tried a debian-based distro, I ended up dist-upgrading to debian standard |
| 13:54.34 | jbevren | hehe |
| 13:54.48 | T0mW | jbevren: I hated what redhat turned into, mandrake has been very easy to work with for the past 3 years |
| 13:54.48 | prpplague | jbevren: have you got experience with the ps2 mem card? |
| 13:54.58 | jbevren | prpplague: no :( |
| 13:55.03 | jbevren | Ive only read docs on the ps1 card |
| 13:55.10 | CosmicPenguin | I have NCAA 2006 saved on one, if that helps |
| 13:55.27 | prpplague | jbevren: ahh, i assume you did since you suggested it |
| 13:55.41 | jbevren | I just know its a serially-operated flash |
| 13:56.48 | jbevren | fun |
| 13:56.52 | jbevren | spartan3 usb module |
| 14:01.22 | prpplague | jbevren: ahh, nand serial is a little misnamed as its not a true serial interface |
| 14:03.08 | jbevren | it's a narrow parallel interface? :P |
| 14:03.53 | prpplague | hehe |
| 14:04.04 | prpplague | that would be a little more accurate |
| 14:04.16 | jbevren | nod |
| 14:04.24 | jbevren | I'm aware of the presence of word-banging |
| 14:04.26 | jbevren | :) |
| 14:04.31 | jbevren | (as compared to bit-banging) |
| 14:04.45 | jbevren | JiffyDOS for the commodore 64 does this. Asynch wordbanging at high speeds |
| 14:05.10 | jbevren | dumping 2 bits at a time instead of 1, and using an increased rate to provide 'up to' 16x faster data transfer |
| 14:05.23 | jbevren | which is fantastic, considering a 2400bps modem can outrun the floppy drive |
| 14:09.46 | prpplague | i just find it odd that you can't find some small nor based devices with nand style interfaces |
| 14:14.43 | prpplague | seems like some e2proms should have the same style of interface |
| 14:16.53 | *** join/#elinux Genesis (n=genesis@dev.aimao.org) |
| 14:18.20 | Genesis | bonjour |
| 14:18.49 | *** join/#elinux TorbaX2 (n=TorbaX@host-84-222-204-240.cust-adsl.tiscali.it) |
| 14:19.18 | prpplague | Genesis: morning |
| 14:19.43 | Genesis | lo prpp |
| 14:34.36 | chouimat | morning |
| 14:54.41 | chouimat | prpplague: LOL http://www.askmen.com/jokes/current/index.html |
| 15:00.02 | CosmicPenguin | ex-chat? |
| 15:00.09 | *** join/#elinux kergoth (n=kergoth@mail.astrocorp.com) |
| 15:01.54 | *** mode/#elinux [+o kergoth] by ChanServ |
| 15:04.56 | prpplague | chouimat: cute |
| 15:16.15 | *** join/#elinux kergoth_ (n=kergoth@c-24-118-219-25.hsd1.mn.comcast.net) |
| 15:16.39 | *** mode/#elinux [+o kergoth] by ChanServ |
| 15:42.04 | *** join/#elinux TimRiker (n=timr@12.110.19.201) |
| 15:42.05 | *** mode/#elinux [+o TimRiker] by ChanServ |
| 16:14.45 | jbevren | heh |
| 16:15.00 | jbevren | my new router draws .75a, and I'm getting an 8ah battery for it :) |
| 16:15.03 | jbevren | mm 10h battery backup |
| 16:16.27 | sorphin | ~lart php and javascript encryption that either doesn't work together or doesn't have a version in the other language |
| 16:19.37 | sorphin | lessee.. blowfish.. output from one doesn't match the other.... javascript xxtea, php xtea... incompatible w/ eachother.. |
| 16:19.39 | sorphin | grr |
| 16:22.35 | Soopaman | jbevren, your new router is for heathens |
| 16:22.36 | Soopaman | :) |
| 16:23.17 | sorphin | speaking of routers |
| 16:23.25 | sorphin | i still need a small case for this flexatx board |
| 16:23.41 | sorphin | i't 3/4" too big for mini-itx cases |
| 16:24.18 | jbevren | Soopaman: hah. |
| 16:24.39 | Soopaman | i still need to find time to hack this bloody router of mine :/ |
| 16:24.42 | jbevren | Soopaman: I'm using the artis ppc board sorphin gpsfan and I got ages ago |
| 16:24.49 | jbevren | I want to phase it out so i can get back to playing/toying with it |
| 16:25.11 | jbevren | and what better than a p3/400 with builtin switch that sips power? :) |
| 16:25.15 | prpplague | T0mW: hey you know of any cheap ccd camera modules |
| 16:30.38 | prpplague | CosmicPenguin: soon to run out of names for storms |
| 16:30.49 | prpplague | CosmicPenguin: i see a large one brewing off the coast of africa today |
| 16:35.03 | CosmicPenguin | alpha beta gamma |
| 16:36.22 | *** join/#elinux andersee (n=andersee@codepoet.org) |
| 16:36.22 | *** mode/#elinux [+o andersee] by ChanServ |
| 16:48.25 | *** join/#elinux zinosat (n=Davide@host182-92.pool8256.interbusiness.it) |
| 17:04.29 | sorphin | jbevren: my new router/firewall will be a PIII 1.13ghz tualatin ($50 for the board and cpu) |
| 17:04.34 | sorphin | just need a damn case |
| 17:04.35 | sorphin | heh |
| 17:04.45 | sorphin | and it'll of course be running JunOS :) |
| 17:05.01 | jbevren | sorphin: nice. |
| 17:05.06 | ade|desk | sorphin: you could have mice , but the shipping costs :( |
| 17:05.12 | jbevren | where does one (ahem) acquire junos? |
| 17:05.27 | sorphin | jbevren: a service contract w/ juniper ;) |
| 17:05.44 | sorphin | we only have a crapload of M160s and M40s :P |
| 17:05.50 | jbevren | of course, that'd be way too much for a residential gateway :) |
| 17:06.06 | sorphin | jbevren: junos Olives work nicely tho ;) |
| 17:06.13 | jbevren | heh |
| 17:06.17 | sorphin | and it just so happens i'm an expert at making them ;p |
| 17:06.23 | sorphin | ade|desk: dimensions? |
| 17:06.27 | jbevren | my box is configured with a dual eepro100 ;-) |
| 17:06.37 | sorphin | jbevren: funny, that's what junos uses |
| 17:06.40 | sorphin | eepro100 |
| 17:06.46 | jbevren | <PROTECTED> |
| 17:06.46 | jbevren | <PROTECTED> |
| 17:06.51 | sorphin | right |
| 17:07.00 | sorphin | it handles 82559 no issues |
| 17:07.08 | sorphin | pii won't work tho |
| 17:07.16 | sorphin | it'll think it's a real router |
| 17:07.19 | jbevren | ah |
| 17:07.21 | jbevren | this is a p3-m |
| 17:07.23 | sorphin | and require a hdd and flash |
| 17:07.26 | sorphin | ah |
| 17:07.30 | sorphin | it'd work then |
| 17:07.30 | jbevren | model name : Pentium III (Coppermine) |
| 17:07.31 | jbevren | cpu MHz : 397.333 |
| 17:07.34 | sorphin | ah |
| 17:07.35 | sorphin | cool |
| 17:07.38 | sorphin | that'd work good then |
| 17:07.51 | sorphin | actually that'd make a perfect olive |
| 17:07.51 | sorphin | heh |
| 17:08.07 | sorphin | i have my crystals (one is an olive already), but footprint is too big |
| 17:08.25 | jbevren | ah |
| 17:08.33 | sorphin | my only issue now |
| 17:08.35 | jbevren | this is the p3, 512M ecc sdram, and an 8port switch in a shallow 1U |
| 17:08.41 | jbevren | slightly more shallow than a cisco 2500 |
| 17:08.42 | sorphin | is i'll need to use a dual port intel card |
| 17:08.53 | sorphin | nod |
| 17:09.05 | sorphin | notehr toy? |
| 17:10.29 | jbevren | its the compact pci box off ebay I showed you recently |
| 17:10.35 | jbevren | the one the lucent 48v power system's for |
| 17:10.43 | jbevren | so.. how does someone like me acquire junos ;-) |
| 17:11.09 | sorphin | hehe |
| 17:11.28 | sorphin | any more of em left? |
| 17:11.36 | jbevren | I think so actually |
| 17:11.39 | jbevren | ebay for 'kontron' |
| 17:11.46 | sorphin | k |
| 17:11.51 | jbevren | they come with some linux distro on em |
| 17:11.53 | jbevren | very redhat-like |
| 17:12.10 | jbevren | I reinstalled debian to the swap partition by loading the debian installer from /boot :0 |
| 17:12.11 | jbevren | <PROTECTED> |
| 17:12.27 | sorphin | uh |
| 17:12.30 | sorphin | 9 results |
| 17:12.34 | jbevren | yeah |
| 17:12.34 | sorphin | wanna help me narrow it down? |
| 17:12.36 | jbevren | you want the um |
| 17:12.38 | jbevren | cp604 I think |
| 17:12.44 | sorphin | ow |
| 17:12.50 | sorphin | the $79/$120 one |
| 17:13.07 | jbevren | yeah its pricey, but its a nice 2slot cpci 1u chassis |
| 17:13.28 | kergoth | sorphin: you know anything about multilink ppp? |
| 17:13.39 | jbevren | sorphin: the 512M ecc is soldered on the board. heh |
| 17:13.44 | jbevren | it does have a pmc slot too |
| 17:13.52 | sorphin | pmc? |
| 17:14.02 | sorphin | jbevren: no vga option... |
| 17:14.02 | sorphin | heh |
| 17:14.11 | sorphin | serial consoled? |
| 17:14.15 | sorphin | kergoth: somewhat |
| 17:14.44 | kergoth | i need to know if theres a way to control the algorithm used to distribute the multilink fragments over the ppp links, or if i'm stuck with a non-weighted round robin |
| 17:14.50 | jbevren | sorphin: yeah |
| 17:14.53 | jbevren | the bios has a serial agent |
| 17:14.55 | jbevren | it works pretty well |
| 17:15.01 | kergoth | i.e. for multilink across links of unequal bandwidth |
| 17:15.08 | jbevren | press tab to enter the bios setup, and vt100 terminals work fine |
| 17:15.16 | jbevren | com 1, 19200n8 |
| 17:16.02 | sorphin | jbevren: i don't think i'm gonna BIN this one.. i think if i get it i'll let the auction run out :P |
| 17:16.09 | sorphin | $79 is cheaper than $120 P |
| 17:16.15 | CosmicPenguin | usually thats the case |
| 17:16.48 | sorphin | heh |
| 17:17.36 | jbevren | well |
| 17:17.39 | jbevren | its a lot cheaper |
| 17:17.46 | jbevren | and they havent been selling since he doesnt konw the specs |
| 17:18.26 | sorphin | so it's a PIII-M 400? 512M, intel 82559 |
| 17:18.28 | sorphin | anything else? |
| 17:20.19 | kergoth | sorphin: so, no idea on the multilink fragment distribution algo? |
| 17:22.03 | file | what's selling? eh? |
| 17:25.12 | jbevren | sorphin: sec |
| 17:25.24 | jbevren | the switch is vlan capalbe, but not managed fyi |
| 17:25.39 | jbevren | you program it via bitbang serial or some crap on its 'parallel port'. |
| 17:25.52 | jbevren | -[00]-+-00.0 Intel Corp. 440BX/ZX/DX - 82443BX/ZX/DX Host bridge |
| 17:25.52 | jbevren | <PROTECTED> |
| 17:25.55 | jbevren | <PROTECTED> |
| 17:25.55 | jbevren | <PROTECTED> |
| 17:25.55 | jbevren | <PROTECTED> |
| 17:25.55 | jbevren | <PROTECTED> |
| 17:25.56 | jbevren | <PROTECTED> |
| 17:25.57 | jbevren | <PROTECTED> |
| 17:26.06 | jbevren | has a 10g hdd in my case |
| 17:26.28 | file | okay I'm going to ask, what are you two talking about now? |
| 17:27.04 | kergoth | TimRiker: purl has me ignored for 'abusing forget', since i was using listkeys and listvalues and making it forget some of the collected cruft. if you could unignore me, thatd be good |
| 17:29.36 | jbevren | China has decided that if you are spending too much time online, you must be an addict. |
| 17:29.36 | jbevren | yay |
| 17:29.48 | jbevren | I wonder if those of us who are occupationally online count as addicts |
| 17:32.39 | CosmicPenguin | probably |
| 17:32.55 | CosmicPenguin | I like the term occupationally online |
| 17:34.01 | *** join/#elinux Crofton|laptop (n=balister@hc6521ca9.dhcp.vt.edu) |
| 17:38.34 | jbevren | sorphin: were you going to take a stab at one? :P |
| 17:47.49 | *** join/#elinux T`2 (n=total@pdpc/supporter/student/T) |
| 17:51.31 | *** join/#elinux andersee (n=andersee@codepoet.org) |
| 17:51.31 | *** mode/#elinux [+o andersee] by ChanServ |
| 18:34.34 | TimRiker | kergoth`food: you're unignored |
| 18:34.39 | kergoth`food | thanks |
| 18:47.13 | sorphin | heh |
| 18:47.26 | kergoth`food | sorphin: no idea then? |
| 18:47.34 | kergoth`food | methinks its hardcoded in the linux kernel to a round robin algo :( |
| 18:47.39 | sorphin | kergoth`food: sorry, was at lunch |
| 18:47.46 | kergoth`food | ah |
| 18:47.47 | kergoth`food | np |
| 18:48.01 | sorphin | kergoth`food: umm unfortunately, i'm not aware of one |
| 18:48.27 | kergoth`food | i want to use mlppp for my pptp vpn links, to do load balancing at the packet rather than session level |
| 18:48.34 | kergoth`food | but they're going over unequal bandwidth links |
| 18:48.41 | kergoth`food | so i want weighted round robin, not round robin |
| 18:48.43 | kergoth`food | :\ |
| 19:02.07 | *** join/#elinux hufnus (n=slonsiki@DSL135-071.LABridge.com) |
| 19:11.34 | MonMotha | kergoth`food: you should be able to play with tc |
| 19:12.17 | MonMotha | use routing rules to separate the two links (so the kernel won't consider them a multipath) then you can classify and route accordingly |
| 19:12.33 | kergoth`food | i want load balancing between the links, by _packet_ |
| 19:12.52 | kergoth`food | one high speed ftp connection across two wan links with 2 different ips, to a remote site also within our control |
| 19:12.56 | MonMotha | kergoth`food: that's exactly what this allows you to do |
| 19:13.23 | MonMotha | you can actually route based on any criteria iptables can match on, and you can use the traffic control stuff as well |
| 19:13.23 | kergoth`food | i dont see how. the remote site would get half the packets from one ip and half from the other. |
| 19:13.32 | MonMotha | oh, these have separate IPs? |
| 19:13.47 | kergoth`food | site A with 3 t1 lines, site b with one dsl line and one t1 line |
| 19:13.57 | MonMotha | that's a lot harder, since that's not the way you're "supposed" to multihome |
| 19:13.57 | kergoth`food | i need to turn that into one high speed tunnel between the sites |
| 19:14.04 | kergoth`food | with correct balancing between them |
| 19:14.08 | MonMotha | so you have control of both endpoints? |
| 19:14.17 | kergoth`food | i can load balance by session across the links easily, thats no problem, just a multipath route |
| 19:14.19 | kergoth`food | but this.. |
| 19:14.19 | kergoth`food | yeah |
| 19:14.24 | kergoth`food | they're both the same device, one at each end |
| 19:14.45 | MonMotha | I wasn't aware that multipath routing took TCP sessions into account |
| 19:14.50 | MonMotha | that sounds way too smart |
| 19:14.53 | kergoth`food | so i can establish, in this case, like 6 ppp tunnels. one from each link on one side to each link on the other, then balance based on weights assigned to the tunnels based on the bandwidth they're getting from the links |
| 19:15.03 | kergoth`food | MonMotha: it does. connection tracking is our friend |
| 19:15.18 | MonMotha | yeah, I just didn't know the routing stuff took that into account at this point |
| 19:15.22 | MonMotha | that's kinda neat |
| 19:15.33 | MonMotha | at one poitn,t he multipath router was very dumb, round robining on the packet level |
| 19:15.34 | kergoth`food | you know how iptables can mark packets? |
| 19:15.37 | MonMotha | yeah |
| 19:15.45 | MonMotha | and then the router can classify based on that |
| 19:15.49 | kergoth`food | you can even store a packet's mark in the connection tracking entry |
| 19:15.56 | kergoth`food | then route the entire connection based on that out a specific interface |
| 19:15.57 | kergoth`food | etc |
| 19:16.03 | MonMotha | ok, so you're using iptables to do this? |
| 19:16.15 | MonMotha | not just defining two default routes and letting the kernel fly with it? |
| 19:16.15 | kergoth`food | using policy routing with iproute2 and netfilter |
| 19:16.21 | MonMotha | ok, that makes a lot more sense |
| 19:16.39 | kergoth`food | the linux advanced routing and traffic control howto was my best firend when i got hired here :) |
| 19:16.43 | kergoth`food | didnt know about any of this fancy stuff |
| 19:17.08 | MonMotha | hum, I wonder if Kurdel is still around |
| 19:17.50 | kergoth`food | so anyway, i basically want to establish 6 pptp tunnels covering all possible paths between the WAN lines for both sites, then weight them based on the max bandwidth each can have, givne the # of tunnels going over the line, and the line's bandwidth |
| 19:17.59 | kergoth`food | but right now afaik mlppp is purely round robin |
| 19:18.06 | MonMotha | hum, he's not on EFNet right now, but the guys is a genius with this stuff |
| 19:18.24 | kergoth`food | i could do something really sick and establish multiple tunnels per wan link, more tunnels for a line with more bandwidth, but consider the overhead.. icky |
| 19:18.32 | MonMotha | naw, that seems silly |
| 19:18.37 | kergoth`food | yeah |
| 19:18.52 | MonMotha | one option would be universal tun/tap and do the weighting in userspace, but I'm guessing the policy routing can handle this stuff |
| 19:18.57 | kergoth`food | i hear EQL, the linux serial line load balancing, can be weighted by the speed of the serial link, i.e. 115200 |
| 19:19.15 | kergoth`food | i could use eql with ppp with pptp and give it fake values to weight |
| 19:19.18 | kergoth`food | but that seems ugly too |
| 19:19.24 | kergoth`food | mlppp is what i need, afaict |
| 19:19.39 | MonMotha | well, there are better ways to tunnel than pptp |
| 19:19.55 | kergoth`food | nod, i could just do ip over ip with gre, or whatever else |
| 19:20.04 | kergoth`food | i just picked pptp or l2tp because they use ppp, for mlppp |
| 19:20.13 | MonMotha | hum, iptables nth match? |
| 19:20.20 | MonMotha | just match every nth packet |
| 19:20.22 | kergoth`food | hmm, thats a good point, i forgot about that |
| 19:20.35 | kergoth`food | i've seen people doing that instead of using a multipath route |
| 19:20.39 | kergoth`food | for balancing |
| 19:21.55 | kergoth`food | MonMotha: the problem comes back to the fact that i cant put half the packets over one link and half over another if the two links have different ips, unless the other end knows how to put the mess back together, and account for the packets not necessarily being in order |
| 19:22.02 | kergoth`food | MonMotha: thats what mlppp did for me, behind the scenes |
| 19:22.09 | MonMotha | you should be able to do that with GRE as well |
| 19:22.21 | *** join/#elinux hufnus (n=slonsiki@DSL135-071.LABridge.com) |
| 19:22.25 | MonMotha | and since IP doesn't guarantee in-order delivery anyway, that's not a problem |
| 19:23.07 | kergoth`food | well, with the tcp packets arriving in a sequence that doesnt match the sequence numbers, its hard for us to know when we've lost one, isnt it? |
| 19:23.40 | prpplague | cbrake2: next time you talk to ccole tell him i'm almost done with the JuiceBox mod howto |
| 19:23.50 | MonMotha | kergoth`food: tcp tolerates out of order delivery |
| 19:23.56 | MonMotha | that's the whole point of the sequence numbers |
| 19:24.15 | MonMotha | it doesn't care about order, it just waits for a sequence number to time out before it assumes it was lost and retransmits |
| 19:24.27 | kergoth`food | yeah, thats right. misunderstood |
| 19:24.28 | kergoth`food | hmm |
| 19:24.50 | MonMotha | you should be able to set up two GRE tunnels and give both terminations at the same end the same IP |
| 19:25.18 | MonMotha | other option is to forge the IP source address :) |
| 19:25.24 | kergoth`food | ah, then use nth packet to distribute the packets across them.. thatd be perfect |
| 19:25.29 | kergoth`food | yeah, that could work too |
| 19:25.39 | kergoth`food | hmm |
| 19:26.06 | MonMotha | actually, the addresses of the tunnel endpoints are inconsequential |
| 19:26.08 | MonMotha | they can be different |
| 19:26.22 | MonMotha | there's no guarantee with IP that two packets will take the same route from one place to another |
| 19:27.00 | MonMotha | as long as both transport mediums (both tunnels) have connectivity between them and represent a valid route, you can do whatever |
| 19:27.26 | kergoth`food | yeah, thats right. i just need to use the nth in netfilter to set the nexthop to the remote end of whichever tunnel i'm sending it over, then the remote side decides what to do from there |
| 19:27.28 | MonMotha | then just use nth to mark packets according to bandwidth (or the random match, with a percent that represnts the capabilities of the link) and policy route accordingly |
| 19:27.53 | MonMotha | or you can use policy routing based on netfilter mark. No need to explicitly set next hop in netfilter. Your policy router handles that anyway |
| 19:28.10 | MonMotha | there is the -j ROUTE target as well, if you find that useful (though I believe that's only on -t raw) |
| 19:29.15 | kergoth`food | okay, lets say i've got a user on a machine on the lan behind this device, he does an ftp to the remote device at the other site, not to a machine on the remote network. both devices are doing SNAT for their private network on the lan |
| 19:29.34 | kergoth`food | so.. i need to decide whether to SNAT from the wan link ip or the local tunnel ip |
| 19:30.37 | kergoth`food | in other words.. user A ftp's to live internet ip of remote box. i'd normally snat that from the wan ip of this device and send it out to the internet to it |
| 19:31.37 | kergoth`food | i just want to make sure that the remote device isnt confused by packets coming from a live ip, to one of its wan link ips, but coming in from a tunnel with a completely different network, doesnt cause problemsl |
| 19:31.48 | MonMotha | don't do NAT unless you have to |
| 19:31.57 | kergoth`food | these arent my environments. they're customer environments. |
| 19:32.01 | kergoth`food | and they're doing nat |
| 19:32.02 | MonMotha | I know |
| 19:32.07 | MonMotha | but if you aren't hitting the internet, don't do NAT |
| 19:32.16 | MonMotha | it's perfectly legal to route site local IPs over a tunnel |
| 19:32.56 | kergoth`food | i cant connect the lans. i cant let them see one anothers real local addresses. they might be two lans on the same network, and have machines on the same ips. i cant let those conflicts occur |
| 19:33.12 | MonMotha | oh, so they're completely unmanaged wrt each other? |
| 19:33.19 | kergoth`food | site A might have an ftp server on 192.168.1.12, and the other site might have an exchange server on 192.168.1.12 |
| 19:33.31 | kergoth`food | its a remote site that has no direct connection (vpn, or whatever else) to the main site |
| 19:33.36 | MonMotha | (as an aside, this is another reason I fscking hate NAT) |
| 19:33.43 | kergoth`food | yeah, no kidding. its a mess |
| 19:33.50 | MonMotha | so how is this tunnel we just discussed involved? |
| 19:34.16 | kergoth`food | its only being created to give the user a single tcp session to the nat box on the remote end that uses all their wan lines |
| 19:34.33 | kergoth`food | the tunnel is only a means to get packet based load balancing across the wan links to the remote device |
| 19:34.57 | MonMotha | I'm not sure I understand the toplogy of the second problem here |
| 19:35.28 | kergoth`food | L1 - F1 - P1 --- internet --- P2 - F2 - L2 |
| 19:35.32 | kergoth`food | where F is the firewall |
| 19:35.33 | kergoth`food | L is the lan |
| 19:35.42 | kergoth`food | P is the nat box, our product, that load balances across the wan links |
| 19:36.00 | kergoth`food | we want a high speed, packet based load balanced tcp session between P1 and P2 only |
| 19:36.18 | kergoth`food | i.e. user on L1 ftps to P2, which port forwards to whatever., we want that link to be high speed, all the wan links |
| 19:37.13 | MonMotha | I see what you're saying |
| 19:37.24 | kergoth`food | no direct connection between L1 and L2, though thats what i wanted to do initially |
| 19:37.29 | MonMotha | this topology doesn't really work well for this |
| 19:37.29 | kergoth`food | since its easier, just a vpn implementation |
| 19:38.02 | MonMotha | the only reason the pairing up non-shared-ip-space links like that works here is because of the tunnel. Once you start throwing real addressing back into the equation, stuff starts breaking |
| 19:38.09 | MonMotha | does F1 and F2 have a real internet address? |
| 19:38.24 | kergoth`food | nope, the P boxes are the nat boxes, in this proposed situation |
| 19:39.31 | MonMotha | ok, give them both a link-local address (one of those 169 things). Add a static route for their peer at each end using those 169 addresses. P boxes NAT to that link local address for tunneled connections ONLY |
| 19:39.52 | MonMotha | as usual, NAT leads to more NAT, leads to more NAT, leads to... |
| 19:39.58 | MonMotha | this is why we all need IPv6 |
| 19:40.13 | MonMotha | that and I want usable multicast :) |
| 20:03.17 | mindfunk | does anyone here work at montevista? |
| 20:03.50 | mindfunk | montavista |
| 20:06.44 | *** join/#elinux Wingnut (i=mcrist@mknod.org) |
| 20:07.00 | sorphin | mindfunk: nope, no one's that desperate ;) |
| 20:07.11 | *** join/#elinux ade|desk (n=adavey@213-232-83-137.dsl.prodigynet.co.uk) |
| 20:07.22 | mindfunk | he |
| 20:07.30 | mindfunk | seemed like a cool place to me |
| 20:07.33 | sorphin | cept maybe ade|desk |
| 20:07.43 | sorphin | ade|desk: oh hey adrian, didn't see you come in ;) |
| 20:07.46 | mindfunk | I guess the real question is: |
| 20:08.07 | mindfunk | does anyone know where montavista puts their open source distros? |
| 20:08.19 | sorphin | mindfunk: i've wondered that myself |
| 20:08.31 | ade|desk | ~whaleslap sorphin |
| 20:08.36 | ibot | ACTION slaps sorphin upside and over the head with one freakishly huge killer whale named hugh |
| 20:08.36 | sorphin | they seem to have made everything hidden now |
| 20:08.53 | mindfunk | they claim that it is available |
| 20:09.05 | sorphin | mindfunk: heh |
| 20:09.19 | mindfunk | ade|desk: do you work at MV? |
| 20:10.27 | ade|desk | mindfunk: no sorry |
| 20:10.41 | mindfunk | ~botmail for TimRiker: do you know where MV puts their open source'd distros? :) |
| 20:10.47 | *** join/#elinux TimRiker (n=timr@pdpc/supporter/bronze/TimRiker) |
| 20:10.47 | *** mode/#elinux [+o TimRiker] by ChanServ |
| 20:10.57 | mindfunk | TimRiker: do you know where MV puts their open source'd distros? :) |
| 20:11.05 | jbevren | mindfunk: nice timing |
| 20:11.17 | mindfunk | heh |
| 20:12.10 | sorphin | mindfunk: personally, just to get a buildtree |
| 20:12.16 | sorphin | i had to fill out a form |
| 20:12.23 | sorphin | get sent a key to unencrypt something |
| 20:12.30 | sorphin | real "open" |
| 20:12.33 | sorphin | let me tell ya |
| 20:12.42 | mindfunk | they claim to be open |
| 20:12.45 | mindfunk | hrm |
| 20:12.47 | Wingnut | sorphin: I've done the same thing, but that only gets you the 2.4 kernel |
| 20:12.51 | mindfunk | I was suspicious |
| 20:13.07 | sorphin | Wingnut: yeah, this got be gcc et al |
| 20:13.11 | sorphin | s/be/me/ |
| 20:13.15 | sorphin | funny thing was |
| 20:13.16 | T`2 | yea i remember searching for montavista's sources.. couldn't find zip on their site |
| 20:13.20 | sorphin | it wasn't even anything special |
| 20:13.23 | mindfunk | Wingnut: I was physically at MV a couple weeks ago, and they claimed EVERYTHING was available |
| 20:13.29 | sorphin | just normal gcc-cross |
| 20:13.47 | sorphin | mindfunk: if you wanna be hassled by sales drones, sure |
| 20:14.04 | sorphin | i tried searching for what they used to have out in the open |
| 20:14.09 | sorphin | all gone |
| 20:14.18 | sorphin | all under key now :/ |
| 20:14.28 | mindfunk | TimRiker: you listening to this ^^ |
| 20:14.39 | sorphin | mindfunk: heh |
| 20:14.40 | sorphin | broadcom-bcm98245cpci-ppc_82xx-encrypted.iso |
| 20:14.45 | sorphin | that's from montavista |
| 20:14.57 | sorphin | note the word encrypted |
| 20:15.21 | jbevren | theyre lame. |
| 20:15.28 | jbevren | they need to release their source according to the gpl |
| 20:15.29 | jbevren | :) |
| 20:15.33 | sorphin | yep |
| 20:15.35 | mindfunk | a lot of companies release the code but almost hide it on their site somewhere... so they are GPL complient |
| 20:15.37 | jbevren | wait, the gpl said available, not useable |
| 20:15.44 | ade|desk | its to stop people from ripping off their gpl code ... erm whoops |
| 20:16.04 | mindfunk | encrypted == not available |
| 20:16.13 | jbevren | :) |
| 20:16.18 | jbevren | that depends on whose lawyer you are |
| 20:16.23 | T`2 | well.. sorphin u didn't sign anything saying you aren't gonna fork the tree and host it on your servers did u? |
| 20:16.23 | mindfunk | right :) |
| 20:19.40 | Wingnut | http://source.mvista.com/git/gitweb.cgi?p=linux-omap-2.6.git;a=log may be it |
| 20:20.33 | jbevren | heh |
| 20:20.35 | jbevren | read the logs all you want |
| 20:20.38 | jbevren | you cant view the tree |
| 20:20.55 | Wingnut | yeah, once I get home today I'll see if I can go 'git' it |
| 20:21.03 | jbevren | hehe |
| 20:21.33 | jbevren | http://source.mvista.com/git/gitweb.cgi?p=linux-omap-2.6.git;a=tree;h=eed337ef5e9ae7d62caa84b7974a11fddc7f06e0;hb=88d7bd8cb9eb8d64bf7997600b0d64f7834047c5 |
| 20:21.35 | jbevren | 2.6.12 |
| 20:21.35 | jbevren | :) |
| 20:21.50 | jbevren | rc4 |
| 20:22.05 | jbevren | wtf is 'git' anyway |
| 20:22.21 | Wingnut | the new kernel revision control system |
| 20:22.25 | jbevren | ohhh |
| 20:22.29 | Wingnut | since the kernel moved off bitkeeper |
| 20:22.33 | jbevren | I see |
| 20:23.57 | mindfunk | Wingnut: source.mvista.com/pub |
| 20:24.03 | mindfunk | Wingnut: www.muru.comn |
| 20:24.18 | jbevren | pub not found |
| 20:24.24 | jbevren | ftp? :P |
| 20:24.33 | jbevren | yep |
| 20:25.01 | jbevren | btw, think I should move my amd64 to linux64? |
| 20:25.02 | sorphin | well |
| 20:25.08 | sorphin | this was when i was doing the A3000 |
| 20:25.10 | Wingnut | mindfunk: last time I looked at muru there was only h2 and perseus stuff |
| 20:25.20 | jbevren | sorphin: a3k = back burner? :) |
| 20:25.38 | mindfunk | Wingnut: dig around there are a bunch of kernel patches there |
| 20:25.43 | mindfunk | Wingnut: old stuff though |
| 20:27.22 | mindfunk | ~lart montavista |
| 20:27.36 | kergoth`food | MonMotha: got it. no NAT in the tunnels themselves at all. set up the tunnels with link local addresses, or something from one of those reserved blocks, so they wotn conflict with anything. nexthop the traffic destined to P2 to the ip of the remote end of the tunnel. the remote end will be the tunnel interface with the link local address, which normally wouldnt accept that traffic destined for the public ip of its other int |
| 20:27.36 | kergoth`food | erface.. but if we enable forwarding for htat interface, and just forward it back to ourselves on our other interface, it should just work. |
| 20:27.37 | Wingnut | hmm...sounds like mvista anyway... |
| 20:27.45 | mindfunk | heh |
| 20:28.18 | sorphin | jbevren: yeah, in line w/ 5000 other things |
| 20:29.01 | sorphin | woo |
| 20:29.06 | sorphin | hope we win the powerball tomorrow |
| 20:29.09 | sorphin | pool here at work |
| 20:29.11 | kergoth`food | sorphin: how much do you know about gre tunnels? is it possible to have 3 tunnels between the same two machines, and in the tunnels have the source ip be the same on all 3, and the remote ip be the same on all 3 as well? |
| 20:29.18 | jbevren | sorphin: Nod. :) |
| 20:29.25 | sorphin | jbevren: $240 mil |
| 20:29.25 | jbevren | sorphin: Life keeps getting in my way heh |
| 20:29.31 | jbevren | sorphin: 240mil? |
| 20:29.36 | sorphin | powerball |
| 20:29.38 | jbevren | oh |
| 20:29.42 | jbevren | maybe I should get a ticket eh? |
| 20:29.45 | sorphin | hehe |
| 20:29.46 | jbevren | not that I'd win |
| 20:29.54 | jbevren | but if I did, that's some serious investment money |
| 20:29.56 | sorphin | jbevren: indeed |
| 20:30.17 | jbevren | I'd have to invest it, or I'd go wild and get a mazaratti or some other lame car |
| 20:30.19 | ade|desk | i'd buy a new mouse for that |
| 20:39.22 | jbevren | I like my SGI/Logitech 3button PS/2 mice |
| 20:39.29 | jbevren | they track better than my intellimouse ;-) |
| 20:39.44 | kergoth | ~pester sorphin |
| 20:39.45 | ibot | sorphin: Are we there yet? .. Are we there yet? .. Are we there yet? see http://beverlys.net/LJ/BuggingYou.swf |
| 20:45.44 | CosmicPenguin | we beat the street |
| 20:45.46 | CosmicPenguin | yayyyyyy |
| 20:48.38 | kergoth | sorphin: nevermind, i got it |
| 21:02.54 | jbevren | lol I love that swf |
| 21:03.38 | kergoth | !llama |
| 21:03.45 | kergoth | ~llama |
| 21:03.46 | ibot | i guess llama is the incarnation of god on earth, or http://images2.jokaroo.net/flash/llamasong.swf |
| 21:04.24 | jbevren | sigh |
| 21:04.40 | jbevren | hehe |
| 21:04.51 | jbevren | site seems offline |
| 21:05.12 | jbevren | 3200x1024 |
| 21:07.30 | chouimat | nice resolution |
| 21:08.05 | jbevren | 2x 1600x1024 |
| 21:08.10 | jbevren | http://jbevren.net/screens/sshot.jpg |
| 21:08.47 | jbevren | sgi 1600sw |
| 21:08.49 | jbevren | I have 3, 2 on this system |
| 21:13.50 | jbevren | sigh |
| 21:14.05 | jbevren | #9 cards for 1600sw's are still 150 freakkin bucks |
| 21:22.24 | sorphin | jbevren: installed the mini iso yet? :P |
| 21:24.30 | fishhead | http://www.digitimes.com/Backgrounders/ArtReview.asp?datePublish=2005/10/11&pages=PR&seq=207&Cat=1 <--- uNF! |
| 21:38.09 | CosmicPenguin | sigh |
| 21:38.16 | CosmicPenguin | I so don't want to go to german class today... :) |
| 21:42.49 | sorphin | CosmicPenguin: they moving to you dresden? ;) |
| 21:52.52 | fishhead | jbev is a anime fan |
| 21:52.57 | jbevren | I am? |
| 21:53.42 | fishhead | Looking at your background |
| 21:53.49 | jbevren | heh |
| 21:53.51 | jbevren | Belldandy? |
| 21:53.55 | fishhead | yes |
| 21:54.12 | jbevren | I was using her for a background long before I'd ever seen anime |
| 21:57.25 | sorphin | jbevren: wake that hunk of junk up |
| 21:57.38 | jbevren | wake that hunk of junk up? |
| 21:58.20 | sorphin | jbevren: starbase |
| 22:00.47 | sorphin | bbl. |
| 22:08.21 | *** join/#elinux T`2 (n=total@pdpc/supporter/student/T) |
| 22:22.06 | Genesis | bonne nuit |
| 22:53.22 | *** join/#elinux joe_bleau (n=joe_blea@adsl-65-68-175-165.dsl.fyvlar.swbell.net) |
| 22:53.55 | fishhead | quick question. when you wash a circuit board and you end up with white stuff on the board espically around pins and pads, should you be concerned, according to one doc I read no, but want to make sure |
| 23:06.25 | joe_bleau | I've seen 'no-clean' flux turn white when exposed to high humidity. I've run SMT boards through the dishwasher, and they come out really clean...except for the nasty no-clean flux ring around each pad. The CM told me it was harmless, but I still would rather have clean boards with no flux residue. |
| 23:10.24 | *** join/#elinux T`2 (n=total@pdpc/supporter/student/T) |
