04:08.16 | *** join/#gllug Cope-zh_ (n=sanelson@cpc2-ando2-4-0-cust168.sot3.cable.ntl.com) |
05:00.45 | *** join/#gllug Leeds (n=richardc@202.82.163.139) |
05:43.47 | new2unix | moin all |
05:45.48 | Leeds | afternoon |
05:47.01 | new2unix | Terra? |
05:52.52 | *** join/#gllug formi (n=formi@82-35-232-163.cable.ubr01.camd.blueyonder.co.uk) |
05:52.54 | Leeds | Hong Kong |
06:17.04 | *** join/#gllug morsing (n=morsing@emil.morsing.cc) |
06:17.09 | morsing | 'morning Leeds |
06:22.41 | Leeds | hey morsing |
06:23.22 | morsing | How was lunch? |
06:27.17 | Leeds | uh, fine... why do you ask? |
06:29.08 | Leeds | you really don't appreciate decent bandwidth until you've done without it for a few weeks... |
06:29.15 | morsing | What else am I supposed to ask at this time of the morning. Sorry |
06:29.29 | morsing | You on a 14400? |
06:31.41 | Leeds | was on shared 128k in Sri Lanka for 3 weeks, with GPRS http-only out-of-hours |
06:31.52 | Leeds | now on shared 4mbit (I think) :-) |
06:35.14 | Leeds | "wget http://www.hk.kernel.org/pub/linux/kernel/v2.6/linux-2.6.13.tar.bz2" - my standard speed test - gives a sustained 414KB/s |
06:36.05 | Leeds | oh, and apparently my HK visa is "basically approved" :-) |
07:25.50 | mozrat | Hey Leeds, morsing |
07:25.59 | mozrat | have a good trip back Leeds? |
07:27.18 | Leeds | not bad... didn't sleep much, but read, played DS, watched Family Guy and Simpsons on the laptop... :-) |
07:27.44 | mozrat | :) |
07:33.43 | mozrat | that sounds good |
07:34.20 | Leeds | get paid a daily stipend for being in Sri Lanka... my total expenses for the month have come to less than the stipend total, so my actual paycheck is untouched |
07:34.33 | mozrat | you lived for free! |
07:36.01 | Leeds | when I was in SL by myself, I paid for lunch (about $10-$12 per day) and for 'supermarket'-type stuff, while the housekeeper pays for 'market'-type stuff, and travel is covered - plus minimal other expenses because frankly there isn't much else to spend it on |
07:36.12 | Leeds | when the cousins were there with me, they paid for lunch and supermarket |
07:38.02 | Leeds | and about the only expense I have in HK which applies when I'm not here is my mobile rental |
07:39.31 | mozrat | or lack thereof |
07:39.34 | mozrat | see you later |
07:40.03 | Leeds | enjoy |
07:52.33 | *** join/#gllug z00dax (n=z00dax@kbsingh.plus.com) |
07:52.59 | Leeds | afternoon z00dax |
08:03.28 | morsing | z00dax! |
08:06.46 | Leeds | do outlook and outlook express do nntp, with threading and so forth? |
08:58.14 | morsing | Is snoop better than tcpdump? |
08:58.47 | Cope | dunno how to measure it - I've used snoop a lot and found it to be very powerful. |
08:59.04 | Cope | is there something in particular you need to do? |
08:59.27 | Erwin | I usually use ethereal (or tcpdump -w + ethereal on another machine) |
08:59.57 | Cope | that's really excellent; |
09:00.12 | morsing | Cope: I need to prove to my weird Solaris colleagues that snoop is stupid |
09:00.26 | Cope | lol |
09:00.37 | Cope | then I guess you're asking the wrong person. |
09:03.09 | Leeds | I like tcpdump for simple stuff - most of the time when I'm in a position to use it ethereal is either not available or not appropriate |
09:03.39 | Leeds | oh, and will you all please make sure you're subscribing to LWN? |
09:08.39 | morsing | Why? |
09:08.50 | morsing | So what about iptrace vs. snoop then? |
09:11.11 | *** join/#gllug kbsingh (n=kbsingh@nat1.lon1.uk.xinit.com) |
09:11.45 | Leeds | why what? |
09:20.55 | Leeds | seems that some nice person has been trying a long list of possible usernames to ssh in with... |
09:29.20 | *** join/#gllug wethrin (n=dankolb@ipv6.eco.li) |
09:30.29 | Leeds | evening wethrin |
09:30.43 | wethrin | Evening |
09:34.37 | Leeds | hometime! :-) |
09:36.58 | *** join/#gllug clyphox (n=clyphox@82-35-127-97.cable.ubr01.enfi.blueyonder.co.uk) |
09:38.10 | morsing | 10:01 < Leeds> oh, and will you all please make sure you're subscribing to LWN? |
09:38.17 | morsing | wethrin! |
09:40.31 | wethrin | HELLO |
09:41.15 | morsing | How are you? |
09:41.46 | morsing | Weird Qmail message: |
09:41.48 | morsing | <PROTECTED> |
09:41.50 | wethrin | Need breakfast and coffee |
09:41.56 | morsing | Mmm... Cawfy |
09:42.07 | wethrin | Yar. What's SMTP error 451? |
09:42.40 | wethrin | 451 - Requested action aborted: local error in processing |
09:42.43 | eye69 | Requested action aborted: local error in processing |
10:26.39 | *** join/#gllug Leeds (n=richardc@ipvpn095205.netvigator.com) |
11:02.12 | clyphox | Moin |
11:02.17 | clyphox | eek noon |
11:04.37 | wethrin | yes |
11:14.16 | *** join/#gllug ebbeyes (n=ebb@82-44-186-111.cable.ubr06.haye.blueyonder.co.uk) |
12:06.49 | Leeds | gronda gronda, your majesty |
12:37.56 | gregj | :> |
13:13.57 | JAV | afternoon! |
13:23.27 | rhowe | mm, quiche |
13:44.33 | *** join/#gllug ebbeyes (n=ebb@82-44-186-111.cable.ubr06.haye.blueyonder.co.uk) |
13:44.51 | wethrin | rhowe is not a real programmer |
13:45.12 | stephen | heh |
13:52.06 | rhowe | wethrin: indeed not |
13:52.12 | wethrin | Ha |
13:52.19 | wethrin | Real programmers can write Fortran in any language |
15:22.38 | Leeds | hmm... uknova has a "which browser" poll running... firefox is leading IE by 58% to 27% |
15:28.11 | mozrat | does anyone else see a problem with http://planet.gllug.org.uk/ ? |
15:35.15 | kbsingh | mozrat, a proxy error ? |
15:35.29 | mozrat | kbsingh: aye |
16:09.52 | Leeds | working for me now |
16:11.36 | mozrat | and me now |
16:15.02 | *** join/#gllug George (i=george@kde/developer/gwright) |
16:16.04 | wethrin | Ha |
16:21.35 | *** join/#gllug George (n=Test@kde/developer/gwright) |
16:22.24 | *** join/#gllug kbsingh (n=kbsingh@nat1.lon1.uk.xinit.com) |
16:23.26 | wethrin | Very RedHat-ish. In a not-wonderfully-polished sort of way |
16:27.46 | mozrat | BASTARDS |
16:28.19 | wethrin | What's the command to remove leading and trailing whitespace? |
16:30.52 | rhowe | sed -e '^ *' -e ' *$'? |
16:31.00 | wethrin | pah :-P |
16:31.00 | rhowe | er.. fill in s// as appropriate :) |
16:31.15 | wethrin | I thought there was a command to do it |
16:34.33 | rhowe | apropos whitespace returns nothing |
16:34.40 | rhowe | most programming languages have a trim() function |
16:34.54 | Leeds | strip() in python |
16:35.11 | wethrin | Yarr, but nothing in shell. Poo. |
16:35.23 | wethrin | I've done what I want, anyway |
16:36.08 | Leeds | you could map that sed or something similar to a shell function if you were particularly worried |
16:37.11 | wethrin | I'm not that worried |
16:42.31 | rhowe | wethrin: Bash'll let you chomp trailing spaces with ${var% *} |
16:42.41 | rhowe | wethrin: There will be a similar thing for initial spaces, I expect |
16:45.46 | wethrin | Fine, fine. But I've done what I needed |
16:46.03 | wethrin | I could have run it through perl -e "chomp($_)" too |
16:47.43 | Leeds | www.zimbra.com |
16:53.22 | Leeds | it's something along the lines of an open source gmail system |
16:58.59 | Leeds | just downloaded their source tarball - all 150MB of it - to hosted machine in Texas. |
16:59.00 | wethrin | They should be more innovative! |
16:59.07 | wethrin | That's rather a lot |
16:59.16 | Leeds | it came down in 20 seconds |
16:59.45 | Leeds | akamai with the nearest node apparently being on the same LAN :-) |
17:00.02 | wethrin | eek :) |
17:00.07 | rhowe | lePHP? |
17:00.16 | rhowe | Leeds: PHP? |
17:00.27 | Leeds | my machine is less than 1 ms ping away from their download server :-) |
17:00.42 | wethrin | RAR. Dan's cashed my cheque to j-colo |
17:00.44 | Leeds | rhowe: dunno - not unpacked the tarball yet... still marvelling at the download speeds |
17:00.48 | Leeds | wethrin: ah, good! |
17:00.49 | wethrin | Now I owe you lot nothing :-) |
17:02.07 | Leeds | rhowe: appears to contain at least java, perl, php and javascript (for the client) |
17:04.58 | Leeds | whoa, built on lots of existing projects - the tarball contains at least Perl, Postfix, amavis, clamav, cyrus-sasl, jakarta-tomcat, jdk 1.5 (!), mrtg, mysql, openldap, rrdtool, sleepycat-db, and some snmp thing |
17:06.02 | Leeds | might explain why it's a 150meg tarball :-) |
17:07.45 | Leeds | also got nice looking calendar and contact manager... and it's enough of a show-off to make "next Wednesday" in a plain text mail into a link to that day in the calendar |
17:12.05 | Leeds | ah well, bedtime |
17:23.02 | wethrin | Heh |
17:31.16 | Cope | :) |
17:33.49 | George | hey Cope |
17:41.27 | mozrat | cope george! |
17:43.00 | George | I'm SO HUNGRY |
17:43.01 | George | :'( |
17:46.17 | mozrat | george, eat something |
17:46.40 | George | yes |
17:46.50 | George | I'm pondering going downstairs and frying a couple of eggs/sausages |
17:47.38 | mozrat | do so |
17:51.05 | mozrat | next stop home! |
17:51.22 | mozrat | good evening all |
18:25.36 | George | <- fed |
18:26.08 | rhowe | Cope: heh, I don't pretend to be an expert, but fire away |
18:26.16 | rhowe | Cope: murb'd know better than I |
18:26.54 | Cope | well i'm just trying to understand it better; I understand the idea of taking a class c network, and extending the netmask etc |
18:27.12 | Cope | let me back track |
18:27.56 | Cope | forgetting class stuff altogether - suppose you come to cope.isp and say you need 70 ip addresses |
18:27.56 | rhowe | Cope: Well to get a grasp of CIDR notation, you need to think of the netmask in binary. Then it becomes easy. CIDR allows for compact notation for a subset of netmask values |
18:28.06 | Cope | yeah I'm thinking in binary |
18:28.38 | Cope | now suppose you want 70 ip addresses - well 2^7 is 128 - 2 = 126 |
18:28.58 | Cope | so you need a /25 |
18:29.00 | rhowe | OK, so you need 7 bits of address space |
18:29.05 | rhowe | Yes |
18:29.11 | Cope | fine - I understand that |
18:29.20 | rhowe | /25 or larger, anyhow |
18:29.28 | Cope | so now you have 126 ip addresses |
18:29.50 | rhowe | Yes, but you don't know what they are without specifying the prefix |
18:29.59 | Cope | let's call it 213.225.136.0/25 |
18:31.04 | rhowe | OK |
18:31.04 | Cope | so that's 1-126 |
18:31.04 | Cope | with 127 as the broadcast |
18:31.04 | rhowe | yep |
18:31.04 | Cope | fine |
18:31.04 | rhowe | and .0 as the network address |
18:31.05 | Cope | yeah |
18:31.05 | Cope | now this is where i get confused - suppose you want to subnet your own address space |
18:31.20 | Cope | you want 3 separate networks of 25 machines each |
18:31.58 | rhowe | OK.. well for 25 addresses the smallest block you could allocate with CIDR notation would be a 30-address one |
18:32.03 | Cope | yep |
18:32.13 | stephen | you'd want 136.0/27, 136.32/27 and 136.64/27 |
18:32.23 | rhowe | So that's 5 bits of addressing, a /27 |
18:32.30 | *** join/#gllug Erwin (n=erwin@pdpc/supporter/sustaining/Erwin) |
18:32.34 | Cope | ok hang on a sec, let me do the maths |
18:32.46 | Cope | yep 5 bits = 27 |
18:32.48 | Cope | fine |
18:32.56 | rhowe | 32 - 5 :) |
18:33.32 | Cope | ok - so we've reduced our number of ips to 90 useable |
18:33.38 | Cope | that's fine |
18:33.50 | rhowe | 90? |
18:33.59 | Cope | yeah - 3 subnets with 30 ips |
18:34.13 | rhowe | I think you'll find you can fit 4 subnets in there |
18:34.34 | Cope | ah ok - because we could throw in a /29 |
18:35.01 | rhowe | well let's see. |
18:35.31 | rhowe | 213.225.136.0/25 is 213.225.136.[0-127] |
18:35.38 | Cope | 64+32 = 96 |
18:35.46 | stephen | you could have .96/27 too, surely? |
18:36.06 | Cope | we've got 126 addresses in our allocation |
18:36.17 | Cope | so yes, we could actually fit another /27 |
18:36.40 | Cope | ok - all that goes to show is that I feel ok on this sort of thing |
18:36.56 | rhowe | 213.225.136.0/27 is .[0-31] .32/27 is .[32-63], .64/27 is .[64-95] and .96/27 is .[96-127] |
18:37.04 | Cope | where it gets complicated is when we sraer allocating a bunch |
18:37.27 | Cope | suppose rhowe.isp gets 32 class C addresses |
18:37.44 | rhowe | OK... are they all contiguous? |
18:37.51 | Cope | 213.225.0.0 - 213.225.31.0 |
18:37.52 | Cope | yeah |
18:38.12 | rhowe | OK, so that's 213.225.0.0/19, right? |
18:38.16 | rhowe | 213.225/19 for short |
18:38.17 | Cope | now as I understand it these can be aggregated |
18:38.28 | rhowe | (though I dislike that notation myself) |
18:38.33 | Cope | um /21? |
18:39.06 | rhowe | I thought we established 32 addresses were 5 bits. A class C is /24 and 24-5 = 19 |
18:39.19 | rhowe | Or, the other way.. a class B is 16 |
18:39.36 | rhowe | half a class B has 128 class C's, which is /17 |
18:39.42 | rhowe | half that has 64, /18 |
18:39.45 | rhowe | and half that has 32, /19 |
18:40.48 | rhowe | (regarding string comment) |
18:41.55 | Cope | hmm ok |
18:42.27 | rhowe | <PROTECTED> |
18:42.28 | rhowe | <PROTECTED> |
18:42.28 | rhowe | <PROTECTED> |
18:42.33 | rhowe | heh |
18:42.39 | Cope | well the next question is, we could use that internally as 1 x /19 |
18:42.41 | rhowe | keep dreaming, Steve (not you, Cope) |
18:42.59 | Cope | ie 1 network of 8190 hosts |
18:43.11 | rhowe | Cope: Yes. The only reason you would ever need to think "32 class C's" is if you're using old routers or switches which don't understand CIDR |
18:43.27 | rhowe | Cope: I think some older routing protocols only handled class A/B/C/D/E too |
18:43.31 | Cope | sure |
18:43.33 | rhowe | Cope: Maybe old BGP or something |
18:44.26 | Cope | well what I'm getting at is that I've read that somehow (and I don't see how) we might be able to have many networks of /26 (eg) with 62 hosts each |
18:44.28 | rhowe | Cope: So, with modern BGP and route aggregation/summarisation you could advertise a 213.225.0.0/19 route to the world instead of annoucing a seperate route for each of the subnets you've defined |
18:44.40 | Cope | in a way that we've somehow gained more ips than we were given |
18:44.40 | rhowe | Cope: Sure you can |
18:44.48 | Cope | ok - this is the bit where I get confused |
18:44.52 | rhowe | Cope: Hm.. I don't see that that's possible |
18:45.11 | rhowe | Cope: Draw 32 squares in a row, to form a long rectangle |
18:45.24 | Cope | rhowe: yes - I know - this is what I don't follow |
18:45.33 | rhowe | Cope: It may be possible with routing trickery, but unless you use NAT you can't have two hosts with the same IP address |
18:45.56 | Cope | ok - here - from a book: |
18:45.58 | rhowe | Cope: You may be able to avoid wasting 2 addresses on broadcast + network for each subnet, but otherwise, I don't see how it can be done |
18:48.01 | Cope | Imagine a site 192.144.0.0/21; with CIDR we can aggregate several networks - hence CIDR is sometimes called supernetting. Our multiple class C networks do not require a separate routing table entry. The site could also be allocare a subspace of class A or B addresses. Internally the site could use their allocation as: |
18:48.23 | Cope | 1 network of length /21 = 2046 hosrs, netmask 255.255.224.0 |
18:48.47 | rhowe | OK, fine |
18:48.47 | Cope | 32 networks of length /24, 254 hosts each, netmask 255.255.255.0 |
18:49.00 | rhowe | Cope: Note the missing "or" between those sentences :) |
18:49.15 | Cope | 64 networks of length /25, 126 hosts each. netmask 255.255.255.0 |
18:49.42 | Cope | 128 networks of length /26, 62 hosts each, netmask 255.255.255.192 |
18:49.47 | Cope | </end> |
18:50.06 | Cope | sure i understand this is an OR |
18:50.26 | Cope | but to me 64 networks x 126 is much bigger than 2046 |
18:50.45 | rhowe | well.. let's see.. |
18:50.58 | rhowe | A /24 is 254 hosts. We can understand that trivially |
18:51.08 | Cope | I make that 8064 hosts |
18:51.22 | rhowe | So, a /23 is 2*256-2 = 510 hosts |
18:51.42 | rhowe | a /22 is twice as big, 1022 hosts, and a /21 is twice as big as that, 1046 hosts |
18:51.45 | rhowe | err, 2046 |
18:51.55 | rhowe | OK, so their maths is right for the /21 |
18:52.14 | rhowe | I don't think their maths is right for the /24 though |
18:52.30 | rhowe | Try this.. draw 32 squares next to each other, horizontally in a row. |
18:52.35 | Cope | ok |
18:52.38 | rhowe | So that you have a long stripy rectangle |
18:52.47 | rhowe | This is your IPv4 address space |
18:53.23 | rhowe | hm, actually we can do this quite trivially |
18:54.03 | rhowe | A CIDR network of size /n can contain 2 networks of size /(n-1) |
18:54.13 | rhowe | err, I got that wrong - s/-/+/ |
18:54.29 | rhowe | A CIDR network of size /n can contain 2 networks of size /(n+1) |
18:54.49 | rhowe | Or, conversely, a CIDR network of size /n is half the size of a network of size /(n-1) |
18:55.00 | rhowe | e.g. a /25 is half the size of a /24 |
18:55.15 | rhowe | Taking the size as all addresses in the network, including network and broadcast |
18:55.41 | Cope | ok |
18:55.52 | rhowe | So, we have a /21 |
18:55.57 | rhowe | which can contain 2 /22's |
18:56.04 | rhowe | which can each contain 2 /23's |
18:56.09 | rhowe | which can each contain 2 /24's |
18:56.15 | rhowe | that's 2*2*2 = 8 |
18:56.25 | rhowe | So in a /21 we can only have 8 /24's, not 32 like the book says |
18:57.04 | rhowe | We can generalise this... a network of size /n can contain 2^i networks of size /(n-i) |
18:57.25 | rhowe | dammit, +i |
18:57.39 | rhowe | A network of size /n can contain 2^i networks of size /(n+i) |
18:57.50 | rhowe | So, in our /21 we can have 2^5 = 32 /26's |
18:57.57 | rhowe | Not 128 |
18:58.25 | rhowe | And the book is also wrong about the netmask |
18:58.29 | Cope | ahaha! |
18:58.38 | rhowe | hm.. no it isn't |
18:58.40 | rhowe | well, it sort of is |
18:58.50 | Cope | wait a moment! |
18:58.56 | rhowe | a /25 isn't 255.255.255.0, it's 255.255.255.128! |
18:59.02 | rhowe | Or, it's 255.255.255.0 |
18:59.08 | rhowe | Depending on the prefix! |
18:59.12 | rhowe | hm, ignore that |
18:59.16 | rhowe | It's .128, always |
18:59.39 | Cope | i think the book has a consistent error |
18:59.44 | rhowe | I refer you to the shell script I posted once about using bash to calculate CIDR from netmask |
19:00.06 | Cope | remember my example 213.225.0.0 - 32 class c's? |
19:00.18 | Cope | and I said that was a /21 - but I was wrong |
19:00.25 | I | <PROTECTED> |
19:00.25 | I | <PROTECTED> |
19:00.25 | I | <PROTECTED> |
19:00.25 | I | <PROTECTED> |
19:00.27 | I | <PROTECTED> |
19:00.30 | I | <PROTECTED> |
19:00.32 | I | <PROTECTED> |
19:00.35 | I | <PROTECTED> |
19:00.39 | I | <PROTECTED> |
19:00.40 | I | <PROTECTED> |
19:00.42 | I | <PROTECTED> |
19:00.45 | I | <PROTECTED> |
19:00.47 | I | <PROTECTED> |
19:00.50 | I | <PROTECTED> |
19:00.52 | I | <PROTECTED> |
19:00.55 | I | <PROTECTED> |
19:00.57 | I | <PROTECTED> |
19:00.58 | Cope | okok |
19:01.01 | I | nick rhowe |
19:01.04 | Cope | now look! |
19:01.05 | rhowe | argh |
19:01.11 | Cope | 20:10 < I> 255.255.224.0) return "19";; |
19:01.14 | Cope | 20:10 < I> 255.255.224.0) return "19";; |
19:01.19 | Cope | see |
19:01.30 | Cope | the book has typo'd /21 for /19 |
19:01.43 | Cope | well not typos |
19:01.47 | Cope | just got it wrong |
19:01.49 | rhowe | Ah, yes |
19:01.51 | Cope | look: |
19:01.58 | Cope | 1 network of length /21 = 2046 hosrs, netmask 255.255.224.0 |
19:02.07 | Cope | that should be /21 |
19:02.10 | Cope | NOT |
19:02.13 | Cope | should be 19 |
19:02.28 | rhowe | Yeah, that's where it's gone wrong |
19:02.33 | rhowe | Should be /19, not /21 |
19:02.38 | rhowe | Then it's right |
19:02.48 | Cope | sanelson@kotov:~$ ipcalc 192.144.0.0/19 |
19:02.49 | Cope | Address: 192.144.0.0 11000000.10010000.000 00000.00000000 |
19:02.49 | Cope | Netmask: 255.255.224.0 = 19 11111111.11111111.111 00000.00000000 |
19:02.49 | Cope | Wildcard: 0.0.31.255 00000000.00000000.000 11111.11111111 |
19:02.49 | Cope | => |
19:02.51 | Cope | Network: 192.144.0.0/19 11000000.10010000.000 00000.00000000 |
19:02.54 | Cope | HostMin: 192.144.0.1 11000000.10010000.000 00000.00000001 |
19:02.56 | Cope | HostMax: 192.144.31.254 11000000.10010000.000 11111.11111110 |
19:02.59 | Cope | Broadcast: 192.144.31.255 11000000.10010000.000 11111.11111111 |
19:03.01 | Cope | Hosts/Net: 8190 Class C |
19:03.30 | Cope | So all that confusion because of a mistake in the book! |
19:03.38 | rhowe | yep |
19:04.11 | rhowe | If you have a network you want to subnet, it helps to draw it out like I described with the 32 squares |
19:04.27 | rhowe | Starting from the left, fill in the bits you aren't allowed to mess with |
19:04.31 | Cope | ok |
19:04.36 | Cope | let's make up an example |
19:04.38 | rhowe | e.g. if you have a /19, fill in the first 19 squares |
19:04.49 | Cope | nah - I have a /27 |
19:05.12 | Cope | ok - so 27 filled in |
19:05.14 | Cope | what now |
19:06.17 | rhowe | OK, sop you have 5 empty squares, 5 bits of address space to play with |
19:06.32 | rhowe | Now, decide on the smallest subnet you want to allocate |
19:06.51 | Cope | ok 8 machines |
19:07.07 | rhowe | OK, so 3 bits |
19:07.15 | *** join/#gllug ebbeyes (n=ebb@82-44-186-111.cable.ubr06.haye.blueyonder.co.uk) |
19:07.38 | rhowe | Draw a thick line between box 29 and 30 |
19:08.01 | rhowe | call it 'subnet boundary' or something |
19:08.04 | Cope | hmm |
19:08.10 | Cope | yep |
19:08.26 | rhowe | That leaves you 2 bits |
19:08.39 | rhowe | Those 2 bits are your network numbers, and with them you can have 4 networks of /29 |
19:08.53 | Cope | why 4? |
19:09.02 | rhowe | 00, 01, 10, 11 |
19:09.03 | rhowe | 4 |
19:09.12 | Cope | oh right yah sure |
19:09.20 | Cope | but we shouldn't use 00 and 11 |
19:09.26 | rhowe | We should |
19:09.35 | Cope | not according to the rfc |
19:09.45 | rhowe | Oh I'd ignore that :P |
19:09.53 | Cope | it is old |
19:09.56 | rhowe | These are network prefixes |
19:10.01 | rhowe | Not host addresses |
19:10.13 | Cope | mmkay |
19:10.22 | rhowe | "prefix" is a term used in many places to refer to the part of the CIDR spec before the / |
19:10.32 | rhowe | not quite sure what the bit after the / should be called |
19:11.19 | Cope | now our 4 / 29s - how do I know how big each network is again? |
19:11.25 | Cope | the answer is 30 isn't it |
19:11.35 | Cope | so... 1-31 |
19:11.45 | Cope | 32 = broadcast |
19:12.01 | rhowe | Our 4 29's are each 5 bits big |
19:12.06 | rhowe | er, 3 bits big |
19:12.24 | rhowe | you have (in total) 5 bits of the IPv4 address space to play with |
19:12.44 | rhowe | You've decided to use 2 of those bits to determine the network a host is on, and the remaining 3 to determine the host within that network |
19:12.45 | Cope | we've cut off 3 bits |
19:13.30 | rhowe | This way of doing it gets a bit messy when you have unequal-sized subnets in your supernet, but if you're at that stage you've probably already got a good grasp of CIDR |
19:13.38 | Cope | sure |
19:13.43 | rhowe | e.g. in your setup you could have one /28 and two /29's |
19:14.02 | rhowe | You could call your /29's sub-subnets, I guess |
19:14.16 | rhowe | and just repeat the diagram with the first 28 boxes shaded :) |
19:14.31 | *** join/#gllug \simon (n=simon@host163-7.pool8251.interbusiness.it) |
19:14.34 | \simon | hi all! |
19:14.54 | Cope | rhowe: still trying to remember (work out) how big our /29s are |
19:15.07 | rhowe | Cope: 3 bits |
19:15.16 | Cope | because 32-29 = 3 |
19:15.21 | rhowe | Yes |
19:15.33 | rhowe | Or because there are 3 boxes to the right of your 'subnet boundary' |
19:15.36 | Cope | 2^3 = 8 - 2 |
19:15.47 | rhowe | yep |
19:15.50 | rhowe | so 6 hosts |
19:16.03 | rhowe | x 4 networks = 24 hosts on 4 networks |
19:16.09 | Cope | hmm - now why did you ask at first the slightly misleading question - what's the *smallest* network you want |
19:16.13 | Cope | I said 8 |
19:16.22 | Cope | but that's the smallest |
19:16.31 | Cope | suppose I want a network with 11 machines |
19:16.33 | rhowe | So that we could summarise it as 4 networks of 8 hosts |
19:16.34 | Cope | now we're stuffed |
19:16.45 | rhowe | Just merge two adjacent networks |
19:16.59 | Cope | and have one broadcast? |
19:16.59 | rhowe | make sure they're properly aligned though |
19:17.09 | rhowe | You could merge 00 and 01, or 10 and 11, but not 01 and 10 |
19:17.14 | Cope | sure |
19:17.36 | rhowe | Then you have a /28 and two /29's |
19:18.15 | rhowe | e.g. 1.1.1.0/28, 1.1.1.16/29 and 1.1.1.24/29 |
19:18.22 | Cope | 82.163.12.32/27 |
19:18.29 | rhowe | or 1.1.1.0/29, 1.1.1.8/29 and 1.1.1.16/28 |
19:18.43 | Cope | so thats 82.163.12.33 - 62 |
19:19.21 | Cope | so I do an 82.163.12.33/28 |
19:19.38 | rhowe | OK, so you can have 82.163.12.32/28, 82.163.12.48/29 and 82.163.12.56/29, or 82.163.12.32/29, 82.163.12.40/29 and 82.163.12.48/28 |
19:20.10 | Cope | 82.163.12.48/29 |
19:20.31 | rhowe | no, /28 |
19:20.37 | rhowe | since you want a network with 11 hosts |
19:20.53 | Cope | and one more: 82.163.12.56/29 |
19:20.57 | rhowe | so you need a /28 and either one other /28 or two other /29's |
19:21.03 | Cope | yeah that's what I said |
19:21.07 | rhowe | Oh yeah |
19:21.18 | rhowe | except your first one should be .32/28 |
19:21.24 | Cope | yeah: |
19:21.27 | Cope | so I do an 82.163.12.33/28 |
19:21.33 | Cope | oh 32? |
19:21.46 | rhowe | Yeah |
19:21.50 | Cope | 82.163.12.32/27 |
19:22.02 | Cope | useable: 82.163.12.33 - 62 |
19:22.05 | Cope | surely? |
19:22.06 | rhowe | and also .32/28 |
19:22.08 | rhowe | nope |
19:22.21 | rhowe | networks can overlap precisely |
19:22.44 | Cope | hmm |
19:22.44 | rhowe | .33 wouldn't be a valid network address for anything other than 82.163.12.33/32, which isn't really a network :) |
19:22.45 | Cope | ok |
19:23.18 | Cope | i'm confused again now - i was just about to say I understood |
19:23.29 | Cope | I've been given a notional /27 by $isp |
19:23.37 | Cope | 82.163.12.32/27 |
19:23.46 | Cope | .32 is the network address |
19:23.52 | Cope | .63 is the broadcast address |
19:24.02 | Cope | so I only have 33-62 to carve up |
19:24.36 | Cope | ahh... are you saying that because I've "converted" that /27 into a /28 and 2 x /29 I can start on the network address? |
19:24.59 | rhowe | yes |
19:25.09 | rhowe | The /27 isn't a /27 to you |
19:25.11 | rhowe | Only to the ISP |
19:25.16 | Cope | yeah - exactly |
19:25.24 | rhowe | You then take that /27 and slice it into 3 pieces |
19:25.29 | Cope | they have one routing table entry |
19:25.29 | rhowe | on even bit boundaries |
19:25.39 | rhowe | Yes |
19:25.47 | Cope | what I do with it is my business |
19:25.56 | rhowe | When you subnet your /27, you don't need to tell your ISP you've done it. All they need to do is make sure anything in that address range gets sent to your router |
19:26.18 | rhowe | You can even be *really* sneaky and use the network and broadcast addresses as host addresses |
19:26.20 | Cope | yep |
19:26.42 | rhowe | By configuring your network as 82.163.12.0/25 |
19:26.55 | Cope | only took 1 hr |
19:27.07 | rhowe | Since your ISP will route 82.163.12.32 and 82.163.12.63 to you |
19:27.16 | Cope | yep |
19:27.38 | rhowe | If you configure a /27, your router will drop packets destined for those two hosts (if it's a good router - it shouldn't allow people outside your network to contact an address it knows is a broadcast address) |
19:28.14 | rhowe | But, if you configure a /25, which has a network address of .0 and a broadcast address of .127, then .32 and .63 are just regular host addresses to your router |
19:28.17 | rhowe | and it'll deliver packets to them |
19:29.17 | rhowe | The problem then is that hosts on your LAN think that 82.163.12.1 (for example) is a local address, and won't be sending packets via the router for that address. They'll just ARP on the local ethernet segment, not get a response and give host unreachable |
19:29.53 | rhowe | You could be even sneakier by getting your router to proxyarp for those addresses which aren't yours, I guess... and in that way you'd have full connectivity and 2 free IP addresses! |
19:30.12 | rhowe | But we're getting complicated then :) |
19:31.17 | Cope | hehe |
19:31.21 | Cope | right - dinner time |
19:33.57 | \simon | i'm using VirtualDocumentRoot in the apache configuration. is there a way to setup a default DocumentRoot for those hostnames that dont match a directory? |
20:37.25 | gregj | any funny, british, podcasts you guys can recomend ? |
21:15.34 | gregj | or radio shows I can put on me iPod |
21:19.05 | rhowe | gregj: Not that I can think of.. you could see what the BBC offers, but it's unlikely to be in an iPod-friendly format |
21:21.05 | gregj | mp3 ? :> |
21:22.14 | rhowe | More like realaudio or WMA |
21:23.38 | gregj | :( |
21:23.46 | gregj | W word |
21:24.24 | gregj | bloody bbc |
21:48.21 | gregj | what's the plural for cheque ? |
21:48.30 | gregj | checks ? ;) |
21:48.39 | gregj | chiks |
21:48.40 | gregj | :D |
21:48.42 | gregj | chicks |
21:49.03 | gregj | or just cheques |
21:49.09 | gregj | (ispell doesn't like it) |
21:53.11 | rhowe | cheques in British English, checks in American |
21:53.23 | rhowe | Americans say check, we say cheque |
22:19.52 | rhowe | ugh, /var just exploded on my mail server |
22:20.04 | gregj | yep, of coz I am speaking about British English, I call American English by name explictly |
23:30.51 | George | hey, it's gregj! |
23:37.22 | gregj | :P |
23:38.03 | George | ewwwwwww |
23:38.27 | George | sick man |
23:38.42 | gregj | for you anything, boy |
23:43.44 | *** join/#gllug AngelChild (n=Catalyst@jamesmorse.plus.com) |
23:49.03 | George | fuck off |