Back to Home / #linode / 2004 / 05 / Prev Day | Next Day
#linode IRC Logs for 2004-05-27

---Logopened Thu May 27 00:00:13 2004
00:08--- ---> funkycow [] has joined #linode
00:08funkycow does anyone know how to set up incoming attachment size limits?
00:09funkycow someone needed to send a 10 meg file to 40 people on my system, and needless to say we dont have much space...
00:09@caker for what mta?
00:09funkycow sendmail
00:10funkycow i'm not finding anything really useful on google
00:10funkycow nor on's docs
00:10@caker O MaxMessageSize=1000000
00:10@caker bytes
00:10funkycow the only problem is that the .cf file gets regenerated from the .mc file, no?
00:11@caker define(`confMAX_MESSAGE_SIZE', `1048576')dnl
00:11@caker I think ...
00:12@caker how often do you regenreate the .cf file, though?
00:12funkycow not very
00:12funkycow thanks
00:12funkycow i have to give the guy a call tomorrow, tell him to get everyone to erase the attachment
00:45--- <<-- emcnabb [] has quit (Quit: emcnabb)
00:45--- ---> emcnabb [] has joined #linode
02:07--- ---> jeremy [~jeremy@] has joined #linode
02:07--- <<-- jeremy [~jeremy@] has quit (Quit: )
02:08--- ---> jmaccelari [~jmaccelar@] has joined #linode
02:15@caker hello jeremy
02:16--- ---> Guest43 [] has joined #linode
02:16--- <<-- adamg [] has quit (Read error: Connection reset by peer)
02:16--- User: *** Guest43 is now known as adamg
02:18--- ---> Guest43 [] has joined #linode
02:18--- <<-- adamg [] has quit (Read error: Connection reset by peer)
02:39--- ---> R0mMeL [] has joined #linode
02:40--- <<-- jmaccelari [~jmaccelar@] has quit (Quit: ChatZilla 0.9.52 [Mozilla rv:1.6b/20031208])
02:43--- <<-- probonic [~probonic@] has quit (Ping timeout: 480 seconds)
03:05--- <<-- Sh8d0w [] has quit (Quit: ...)
03:49--- <<-- Newsome [] has quit (Quit: Leaving)
04:04--- ---> adamg [] has joined #linode
04:04--- <<-- Guest43 [] has quit (Read error: Connection reset by peer)
05:28--- <<-- R0mMeL [] has quit (Read error: Connection reset by peer)
05:35--- ---> hetz [~d48f6d26@] has joined #linode
05:35hetz hi people
05:35hetz is anyone from linode here?
05:35hetz .
05:38hetz Does anyone knows if Linode has special policy for hosting open source projects?
05:46hetz :)
05:51adamg it is a bit early for caker
05:52hetz whats the time there?
05:52adamg im in the uk but where caker is it is about 4\5am
05:53hetz u work at linode?
05:53adamg no
05:53hetz ok
05:53adamg afk
06:00--- <<-- Ashen^zzz [] has quit (Quit: EOF)
06:08--- <<-- hetz [~d48f6d26@] has quit (Quit: CGI:IRC (Ping timeout))
06:40--- ---> sunny [] has joined #linode
06:42--- <<-- sunny [] has quit (Quit: )
06:42--- ---> sunny [] has joined #linode
08:26--- <<-- sunny [] has quit (Quit: sunny)
08:29--- ---> jax [~stbe@7-208-pool1.P-POOL.MARIST.EDU] has joined #linode
08:34--- ---> Redgore [] has joined #linode
09:13--- ---> Chang [] has joined #linode
09:17--- <<-- Chang [] has quit (Quit: )
09:36--- ---> sunny [] has joined #linode
09:47--- ---> guigouz [~guigouz@] has joined #linode
10:19--- ---> sjansen [] has joined #linode
10:35--- ---> Redgore_ [] has joined #linode
10:35--- <<-- Redgore [] has quit (Killed (NickServ (GHOST command used by Redgore_)))
10:35--- User: *** Redgore_ is now known as Redgore
10:37--- <<-- sjansen [] has quit (Quit: I'll show you cruel!)
10:41--- ---> fo0barf_ [] has joined #linode
10:43--- <<-- fo0barf [] has quit (Ping timeout: 480 seconds)
11:00--- <<-- emcnabb [] has quit (Quit: emcnabb)
11:02--- <<-- sunny [] has quit (Quit: sunny)
11:04--- ---> Newsome [] has joined #linode
11:15--- ---> square [] has joined #linode
11:15square caker, around?
11:21--- ---> eadz [] has joined #linode
11:22square sighup seem caker
11:22sighup square: sorry...
11:22square er
11:22square sighup seen caker
11:22sighup caker was last seen on #linode 9 hours, 7 minutes and 16 seconds ago, saying: hello jeremy [1085638541]
11:23square aw
11:23square sighup ping caker
11:23sighup caker: ping! ping! ping!
11:23sighup caker was last seen on #linode 9 hours, 7 minutes and 28 seconds ago, saying: hello jeremy [1085638541]
11:23square heh
11:23* guigouz lunch
11:26--- ---> emcnabb [] has joined #linode
11:26jax need something that anyone else can do square?
11:27square maybe
11:27square how long you been with linode?
11:28Efudd 8 months here
11:28Efudd lala
11:28square ok
11:28square its just
11:28square according to 'my account'
11:28square i'm due on 1st of june
11:28square now in my timezone thats fine, cos i get paid on 1st june
11:29square but if caker is couple hours behind
11:29square its gonna try come out on 31st of may
11:29square which wont work, cos i dont get paid till day afer
11:29eadz couple of hours ahead you mean
11:29jax hmm, he might be able to make an exception for you for a couple of months
11:30inkblot surely your credit card bill doesn't come *that* quickly
11:30square yeh eadz
11:30square its not a credit card bill, its a visa debit card
11:30square which takes the money from my current balance
11:31square which untill 1st june wont have enough
11:31inkblot pretty tight finances there, cowboy
11:32square <-- student
11:32square and i have a baby :p
11:33inkblot i stand by my previous assessment
11:33square :p
11:33square thats not the point tho, i'll just wait for caker
11:34adamg caker doesnt set the billing off until the 1st june his time therefore wouldnt the money already be in your account
11:36--- ---> Sh8d0w [] has joined #linode
11:37square depends when 1st june is for him
11:37square how many hours ahead he is
11:37square caker TIME reply]: Thu May 27 11:42:37 2004
11:37jax EST, where are you?
11:38square hes 5 hours behind
11:38square thats perfect
11:38eadz man i wish they had debit cards here. we only have credit cards ( new zealand ). well we have debit cards, but thats like eftpos and not a visa or master card debit card meaning i can't use it online
11:38square heh
11:39square that means it'll be 5am before he even attempts anything
11:39square and it'll be in by then
11:39square usualy goes in about 3.30-4am
11:40eadz banks suck becasue if you have 2 payments, one going in, and one going out, the'll process the outgoing payment FIRST. so bouncing it, then process the incoming payment.
11:40square yeh
11:40square also in this country
11:40square say caker applys for payment on 1st june
11:41square it'll hit my account, and earmark the money
11:41square then it goes into my banks 'holding account' for a couple of days
11:41eadz but what most people don't know about credit cards is you can take out 10% over your limit, as long as you are under the limit. ie you limit is $1000 and you have used $999 ( $1 spare ) you can take out $100
11:41square so they can earn interest on it
11:41square then send payment to cakers' payment company
11:42eadz <-- knows all about tight finances ;)
11:42square for example:
11:42square (uk banks are pathetic):
11:43square April 22, 2004 Invoice #8839 $39.84
11:43square thats when i paid it
11:43square now on my bank statement:
11:43inkblot i try to keep a >$1000 balance in my checking account at all times
11:43square 26th APR Shore Network Tech 615-293-3169 US
11:44square \xA323.07
11:44* Efudd puts linode on an unlimited charge account :P
11:44square 4 days it took them to send payment to caker
11:44square or rather his payment processing comapnyt
11:45adamg it is standard for it to take 3/4 days for payment to be moved on all CC payments
11:45square yeh
11:45square but come on
11:45square its all computerised
11:45square it shouldnt take that long
11:46adamg that is not the point
11:46square it is
11:46adamg not all computerised things are automattic
11:46adamg there are reasons for the delays
11:46square i'm saying the fact that its computerised means it should be auto
11:47adamg it is automattic just not instant
11:47* jax is away: lunch
11:47square yeh
11:47square but
11:47square its the same for cheques too
11:47square they take like one day in holland
11:47square it takes up to 5 here
11:48square thats pathetic
11:48adamg not really
11:48square why not?
11:48square if holland can do it, why cant we?
11:48square holland are not the only ones either
11:48adamg different banking systems
11:48inkblot because holland is the size of new jersey
11:48inkblot i'm sure new jersey could do it
11:49square there was an article in the gaudian the other week about this
11:49square lemme see if i can find it online
11:49inkblot oh, but "there" means "new zealand" i guess
11:51adamg in the uk all checques from each branch are sent to the cheqeue processing center, there they go through a number of steps, the first is to put them through micr, the signiture is then checked, the issuers bank then has to check if the balance is available if it is then a fund transfer is put in, which takes about 3 days
11:51--- ---> cow [] has joined #linode
11:51--- <<-- funkycow [] has quit (Remote host closed the connection)
12:00--- ---> funkycow [] has joined #linode
12:00--- <<-- cow [] has quit (Remote host closed the connection)
12:03cout what's an MiB?
12:03Redgore man in black ? :P
12:05cout I have 64 men in black in my linode!
12:06Redgore lol not in the glossary ?
12:06inkblot "mebibytes"
12:08inkblot "mega-" is a standard prefix meaning that the unit is expressed in lots of one million
12:09inkblot "mebi-" indicates lots of 1048576
12:09inkblot which is 2620
12:09inkblot er, 2^20
12:09inkblot (bad shift)
12:09inkblot see also "kibi-", "gibi-", "tebi-", etc
12:18--- <<-- emcnabb [] has quit (Ping timeout: 480 seconds)
---Logopened Thu May 27 12:19:57 2004
12:19--- ---> mikegrb_ [] has joined #linode
12:19-!- Irssi: #linode: Total of 44 nicks 1 ops, 0 halfops, 0 voices, 43 normal
12:19--- ---> mikegrb_ [] has joined #linode
---Logclosed Thu May 27 12:20:02 2004
12:20--- <<-- mikegrb_ [] has quit (Quit: )
12:31--- ---> emcnabb [] has joined #linode
12:37--- ---> Redgore_ [] has joined #linode
12:37--- <<-- Redgore [] has quit (Read error: Connection reset by peer)
12:38--- User: *** Redgore_ is now known as Redgore
12:42square lmao
12:53SupaDongzu no, there's no prefix "lmao-"
12:53square heh
13:03--- <<-- guigouz [~guigouz@] has quit (Quit: There's no place like
13:15--- <<-- eadz [] has quit (Quit: [BX] Mr. Peanut uses BitchX. Shouldn't you?)
13:41--- ---> fo0barf [] has joined #linode
13:47--- <<-- fo0barf_ [] has quit (Ping timeout: 480 seconds)
13:50* jax is back (gone 02:02:56)
13:50--- <<-- emcnabb [] has quit (Ping timeout: 480 seconds)
14:04--- ---> sunny [] has joined #linode
14:20--- ---> emcnabb [] has joined #linode
14:38--- ---> Redgore_ [] has joined #linode
14:38--- <<-- Redgore [] has quit (Read error: Connection reset by peer)
14:39--- User: *** Redgore_ is now known as Redgore
14:49--- ---> pdx6 [] has joined #linode
14:50pdx6 caker: I'm thinking of getting a linode64 in addition to my linode192. The website says 5/31, but do you know if there will be one open sooner?
14:50pdx6 I'd like to bring up the host over the 3day weekend
15:07square is it possible to block refering ip's with .htaccess?
15:12--- <<-- emcnabb [] has quit (Ping timeout: 480 seconds)
15:13@caker pdx6: I have a new host online, but I'll be spending the next few days testing some performance patches
15:13square caker :)
15:13@caker pdx6: people generally cancel towards the end of the month, so chances are something will open up
15:14@caker hello square
15:14square you read my msg's up there ^^ ?
15:14@caker billing q?
15:14square yeh
15:14@caker what TZ are you in?
15:15square BST atm
15:15square GMT
15:15@caker I don't know what that is :)
15:15@caker ok
15:15square aka GMT
15:15@caker I'm .. -5 GMT?
15:15square yep
15:15@caker and billing doesn't fire off until 6:00 AM my time on the 1st
15:15square 11 am my time
15:15square perfect :)
15:15@caker if it doesn't go through right then, we try again every day for the next four days, so you should be good
15:16square didnt want to be late on my first payment :P
15:16@caker I don't consider late unless it's the 5th or so :)
15:16square ah k
15:17pdx6 alright cool. Someone on the forums posted that he was trying to sell his contract so I might try to get that in the mean time
15:17@caker ok
15:19Griswald yeah, and if it is late, caker is usually a pretty reasonable guy if you talk to 'em XD
15:35--- ---> emcnabb [] has joined #linode
15:44@caker Efudd: got a minute?
15:57Efudd yah?
15:57Efudd busy drooling, about it.
15:57Efudd just add 2 20" LCDs to my desk :-)
15:58Efudd 2 20" 1600x1200 LCDs, 1 19" 1280x760 LCD, 1 15.4" 1920x1200 LCD (laptop) and 1 10.6" 1280x760 LCD (laptop) on my desk. :)
15:59@caker don't want to waste your time, but are you clued in to the difference between 3550-24/48 vs 3524/3548s?
15:59Efudd oh
15:59Efudd the 3524 3548's are EOLd
15:59Efudd the 3550-xx's are the new products
16:00@caker ok, but they're still L2, right?
16:00Efudd L2? Yes. L2/3 in reality.
16:00Efudd altho i only use for L2 still.
16:00@caker is that the difference between the two lines? the 3550's have some L3 stuff?
16:00Efudd the 3524/48's do too, I think
16:01Efudd the 3550 was a more modular platform from *ciscos* standpoint
16:01@caker ok .. big price jump vs the EOL'ed models
16:01Efudd Yup......
16:01Efudd there are probably other differences
16:01Efudd like new IOS platform
16:01Efudd updated CPUs
16:01Efudd larger memory capabilities
16:01Efudd I got my 3550-48's cheap
16:01square caker, still around?
16:01Efudd 2500each? perhaps less
16:02@caker hrm .. they're still cheaper than that on eBay, but that's a great price for new ones
16:02@caker really good
16:02--- <<-- Newsome [] has quit (Quit: Leaving)
16:02Efudd Might have been a little more
16:02Efudd It was ~8k perhaps, for 3? it's been a while
16:02Efudd heck, i have one in my office ><
16:02Efudd has 4 devices on it ><
16:02square caker, any word on fedora core 2? :/
16:02@caker geesh ..
16:02* square hides
16:02Efudd i've not had time to steup my lab
16:03Efudd so i stole this switch for my office for a while
16:03@caker square: no, I haven't had time, but I'd guess it's a very easy yum upgrade
16:03square umm, maybe..
16:03Efudd don't upgrade
16:03* Efudd hasn't in a long time ><
16:03* caker 'sploits Efudd
16:03square any reason why fedora core1 doesnt come with any compile tools or tcl?
16:03square all the other distros do
16:04Efudd :)
16:04adamg debian doesnt
16:04@caker square: no reason
16:04adamg on linode anyway
16:04square i havent tryed debian yet
16:05@caker Efudd: any preference (if you were paying out of your pocket) on going for the 3550's or just the 3548's?
16:06Efudd i'd probably get the 3550s, still. but i'm odd that way
16:06Efudd i'd just wait for a deal
16:06@caker mmm .. my pro/1000 mt quad's shipped
16:06@caker Efudd: ok, thanky
16:07Efudd ayup.
16:08@caker $3k new
16:09Efudd eh
16:09Efudd i found better pricing through resllers than ebay fwiw
16:09@caker linkage?
16:14Efudd dunno. been a while
16:14@caker ok
16:14Efudd i'd have to find po's or something
16:17fo0bar caker: careful, that EMI software is "pirated"
16:17@caker fo0bar: I noticed... wasn't bidding on that, just using it for pricing
16:17fo0bar cisco looks the other way for upgrading SMI to EMI, as long as it's for "educational/lap" use
16:18adamg emi smi?
16:18@caker enterprise software image, vs standard
16:18fo0bar but I have about a half dozen 3550s, most 24 except one 48, and a mixture of SMI and EMI... and they all work well
16:19@caker Only thing EMI does differently that I know of is that it's clusterable
16:19@caker what is the difference, actually?
16:19fo0bar SMI doesn't have ospf/bgp, whatnot
16:19adamg just wondering that
16:19@caker bgp on a switch?
16:20fo0bar SMI really just has basic routing options
16:20fo0bar yeah
16:20fo0bar the 3550 is a full layer 3 switch
16:20fo0bar essentially a big router with a lot of ports :)
16:20fo0bar but the SMI software is severely crippled routing-wise
16:21@caker sounds like what I was looking for .. another layer of DoS protection/ACLs/rate-limiting
16:22fo0bar the rate limiting is not that keen... I think you can only do 5 policies :(
16:22fo0bar but it processes ACLs like there's no tomorrow
16:22adamg i guess that can be done at router level
16:23@caker fo0bar: any way around the issue of switchs unicast'ing unknown packets to every freaking port, other than hard-coding macs?
16:23@caker I suppose in layer 3 land you could do some trickery, but not with only 5 policies ..
16:24fo0bar sorry, I don't know...
16:24--- <<-- sunny [] has quit (Ping timeout: 480 seconds)
16:25* fo0bar is not really one with the policy map
16:25@caker fo0bar: ok :)
16:26Efudd eh
16:26Efudd i need to do some policy map stuff.
16:26Efudd my voip traffic here is lacking
16:26Efudd caker, oh, yeah, the trick is to buy non emi :)
16:27Efudd then just er.. add it
16:27Efudd caker, all of the catalysts do layer3 + now
16:27square eth1 Link encap:UNSPEC HWaddr 40-3E-BE-DE-00-00-00-00-00-00-00-00-00-00-00-00
16:27square heh, wtf
16:27* caker Efudd I need to find a good cisco warez site
16:28@caker mine is lacking :-p
16:28Efudd so does mine.
16:28Efudd the trick is this....
16:28Efudd unless thye've fixed it....
16:28@caker seriously though, I was suprised how hard it was to find even legit updated images on cisco's site
16:28Efudd buy a smartnet contract on an 800 series enduser dsl router
16:28Efudd :)
16:28Efudd attach that to your CCO login
16:28Efudd and bewm, full access to images
16:29Efudd my CCO access expired in 96ish?
16:29Efudd it was valid until about 6-9months ago ><
16:34--- <<-- square [] has quit (Quit: )
16:37* caker orders two more servers
16:38@caker time for lunch ...
16:38Efudd agree.
16:40--- <<-- Redgore [] has quit (Killed (NickServ (GHOST command used by Redgore_)))
16:40--- ---> Redgore_ [] has joined #linode
16:41--- User: *** Redgore_ is now known as Redgore
16:41fo0bar Efudd: but stealing software from cisco is WRONG!
16:41fo0bar :)
16:42--- ---> sunny [] has joined #linode
16:43--- ---> Newsome [] has joined #linode
16:49--- <<-- jax [~stbe@7-208-pool1.P-POOL.MARIST.EDU] has quit (Quit: reading for class)
16:51Efudd righ!
16:51Efudd t
17:19@caker woot
17:19@caker x2
17:22Efudd not a STRKMEWNIT!
17:23@caker the DS72 is worth $300 alone .. gotta love ebay
17:23@caker hopefully, it won't be an ancient epromm
17:23Efudd wtf is that?
17:24fo0bar whoah
17:24fo0bar I didn't know baytech made a DS9
17:24@caker those DS30's will remain unused .. I think they hook into an RPC unit ..
17:24@caker Efudd: console server chassis
17:25fo0bar caker: I really wish I could sell you some of the baytech equipment that is just lying in a box by my foot :)
17:25@caker fo0bar: damn ...
17:25* fo0bar is looking at.... at least 20 DS72s
17:25@caker yeah, I need a bunch of DS74s
17:26@caker fo0bar: those aren't the broken ones, are they? :)
17:26Efudd oic.
17:26fo0bar caker: well, the firmware that came with them was so broken it wasn't funny, but I updated them a long time ago
17:27mikegrb caker2000++
17:28fo0bar the ones that are in use are all being graphed by MRTG, and I wrote a web-based power management system that controls them via snmp
17:28@caker fo0bar: what are you graphing, exactly?
17:29fo0bar amperage
17:29@caker ahh
17:29@caker fo0bar: have you seen/used these DS30 modules?
17:29fo0bar no, all we have are DS3/DS72/DS74s
17:30fo0bar not sure what the DS30s do
17:31fo0bar if you do ever use MRTG with the DS72s, my patch is in for recent versions of MRTG are patched to make the unit ID display correctly on baytechs
17:31fo0bar search the page for baytech
17:32--- <--- Newsome [] has left #linode (Leaving)
17:38@caker fo0bar: cool
17:47fo0barf GOD DAMNIT
17:48fo0barf let me guess, a tornado just hit fremont
17:48fo0barf caker: ...
17:48--- <<-- Redgore [] has quit (Quit: )
17:49fo0barf so now it's back up
17:50fo0barf nm
17:50fo0bar ...
17:50@caker fo0barf: something just go weird?
17:50fo0bar yeah, about 90 seconds of dead time
17:50fo0bar of wich I recorded about 40 with mtr
17:51fo0bar 11. 0% 207 207 20 20 31 234
17:51fo0bar 12. 15% 178 207 21 21 109 2332
17:51fo0bar 13. 20% 166 207 22 22 30 1344
17:51@caker maybe they're moving switches around from the emergency yesterday
17:51@caker s/switchs/ports
17:53adamg did you get a response to your complaint email
17:54@caker yup
17:55@caker the emails served their purpose... she wasn't aware of the BGP problem on the 16th, and hadn't heard about the switch failure
17:55@caker (yet)
17:55@caker asking for a redundant switch uplink kind of confused her, she starting asking my about BGP and ASN numbers ...
17:56@caker s/my/me
17:56adamg she hadnt heard about the bgp problem which took out the entire dc
17:56@caker guess not ...
17:56@caker makes me wonder if the noc guys keep their fuck-ups to themselves
17:57adamg you would have thought others would have complained
17:57@caker she also shed a little more detail about the other two problems (UPS problem, and then a giant DoS attack they had on the 19th)
17:58adamg so she had heard about them then
17:58@caker yeah
17:59adamg I ca understand her not hearing about the switch since that would have been isolated to just a small group of people
17:59@caker she asked for more info about the BGP problem, but I didn't save any of the route-loops
17:59* caker waits for mikegrb to say something about fruit loops
18:00fo0bar seeing -Funroll-loops in gentoo always makes my day
18:00adamg but with the bgp it was dc wide so whe should have info on it
18:00fo0bar funroll loops!
18:00@caker after the looping, I remember traces ending at
18:01fo0bar caker: so you just have one uplink from each cabinet to HE?
18:01@caker fo0bar: cross connects back to my switch, with one uplink into HE from there
18:02@caker I was suprised she started mentioning BGP and routers and whatnot when I asked for another uplink port...
18:02fo0bar with that sort of setup, HSRP on HE's side would be SO easy
18:02fo0bar and would solve problems like that
18:02@caker ...just let me speak to someone with a clue, rather than make stuff up
18:02@caker fo0bar: yeah, that's what we have at ThePlanet
18:03adamg hsrp?
18:03@caker TP has an updated website .. they're now touting "carrier neutral" colo
18:03@caker hot-swap router protocol
18:03fo0bar or if you want to get fancy, do ospf
18:03fo0bar but BGP for somebody like you is like killing a fly with a bazooka
18:04* adamg goes to look
18:04fo0bar caker: "carrier neutral"? is that fancy talk for BYOB?
18:04@caker fo0bar: more or less ..
18:04fo0bar (bring your own bandwidth)
18:04fo0bar heh
18:04@caker TP does have an impressive number of providers, now ..
18:05adamg there total control thing looks intresting
18:05@caker the dedis?
18:05@caker expensive
18:05adamg didnt see the prices
18:06--- ---> pknetworks [] has joined #linode
18:06adamg how many drops do you have at tp
18:06fo0bar caker: we run a very similar setup to TP, except we run cisco 7606s instead of junipers
18:06fo0bar and of course, not as many providers ;)
18:06fo0bar 6513s in the core
18:07pknetworks Hello. :)
18:07@caker adamg: two, for HSRP
18:07adamg I like the web based camaras in the cages
18:07adamg thats intresting they can do cross-connects into equinix
18:08@caker I think EQ has a DC near them, or in the same building
18:08adamg the cage pricing does seem fairly good
18:14--- <<-- Sh8d0w [] has quit (Quit: ...)
18:16@caker > Regarding your uplink inquiry, are you looking to run BGP as well? If you
18:16@caker > have your own AS number there would be no additional cost to run BGP
18:16@caker > otherwise you can get an ASN thru us for $100/mo.
18:16@caker ... get an ASN through them??
18:17adamg for that sort of price
18:17adamg are they crazy
18:17adamg basically all they do is fill in the paper work for you
18:19--- <<-- pknetworks [] has quit (Ping timeout: 480 seconds)
18:20adamg I guess
18:22fo0bar adamg: arin charges for ASNs, but it's a onetime cost, I believe below $1000
18:22adamg I know
18:22adamg but if they are charging 100 a month and you have to keep it though them it is fecking expensive
18:23fo0bar yeah
18:23fo0bar that makes no sense
18:24adamg the reg fee is 500
18:24fo0bar ehh, I was only off by 100% :)
18:24adamg although they do also charge 100 a year as well as a maint fee
18:32--- ---> compuserve [~c89e159a@] has joined #linode
18:33@caker word of advise: stay away from 2.6.6 - it has major ide problems
18:34sunny works dandy here :)
18:34@caker my boxen no likey
18:34sunny what IDE u got ?
18:34@caker Intel Corp. 82801CA IDE U100 (rev 02)
18:35sunny err
18:35sunny Intel Corp. 82801CAM IDE U100 (rev 02)
18:35sunny near same
18:35@caker hah
18:35@caker I could barely get a cmd across while it was syncing the raid drives
18:36@caker rebooted to 2.6.4, works fine, rebooted to 2.6.7-rc1, works fine
18:36sunny ah, I have a single HD
18:36sunny no raid or anything
18:36sunny caker: are u on reiser ?
18:36@caker there are some known issues with 2.6.6 where it spins down drives on reboot, but never brings them back up
18:36@caker no, ext3
18:36sunny I've heard of IDE+rfs problems
18:36@caker it also fails to flush the cache on reboot (ugh)
18:37sunny err, I've had no problems so far :)
18:37@caker that's good :)
18:40@caker [root@host25 root]# uname -a
18:40@caker Linux 2.6.7-rc1-bigmem-sysemu #4 SMP Thu May 27 01:30:57 EDT 2004 i686 i686 i386 GNU/Linux
18:40@caker time to test sysemu-enabled UML :)
18:40--- <<-- smohekey [] has quit (Remote host closed the connection)
18:41* mikegrb tests caker
18:41mikegrb test sat
18:41sunny what about our quad opteron UML hosts?!
18:41sunny >:)
18:41@caker real men use 8 bits
18:43--- ---> csy [] has joined #linode
18:43@caker in case no one believed me yesterday about the switch failure: here's the old HE-provided mrtg graph for that switch:
18:44mikegrb caker: I still don't believe you
18:45mikegrb :P
18:47--- ---> smohekey [] has joined #linode
18:51fo0bar caker: I'm running 2.6.6-mm3 on my laptop for the last 3 weeks, and it seems to work fine
18:51@caker fo0bar: I think mm has a fix for it, not sure ..
18:52fo0bar frankly, the only reason I'm running mm is because I need to to get accelerated X working on the laptop
18:52adamg what is the difference between mm and the normal
18:52fo0bar and video and my DVD/CDRW are the only reasons I'm really running 2.6 :)
18:52@caker adamg: mm has a bunch of experimental/"in the queue" patches which get tested for a long time before they get merged
18:52fo0bar adamg: -mm has some not-fully-tested additions
18:53fo0bar yeah
18:53adamg k
18:53* adamg is feeling rather stupid today
18:53* guinea-pig *is* rather stupid today
18:54fo0bar before this laptop, I never owned a combo DVD reader and CDRW... and I was always a fan of ide-scsi, because it just kinda worked... but if I did ide-scsi on here, I couldn't set the DVD drive to DMA, which made watching movies impossible :)
18:54fo0bar so I had to switch to 2.6 so I could watch a DVD and burn a CD without rebooting
19:20Efudd sighup: bash 178890
19:20sighup Ack! That one's too long! :(
19:20Efudd sighup: bash 178890 -force
19:20sighup *** Now talking in #christian
19:20sighup -Word_of_God- Welcome Abstruse to #christian I am a Bible Bot. For more info type: /msg Word_of_God !info
19:20sighup <Abstruse> !kjv numbers 22:21
19:20sighup <Word_of_God>\xA0 Numbers 22:21 -- And Balaam rose up in the morning, and saddled his ass, and went with the princes of Moab. -\xA0 (KJV)
19:20sighup *** SageRider sets mode: +b *!*
19:20sighup *** Word_of_God was kicked from #christian by SageRider (Please dont Swear)
19:20sighup <Abstruse> I know I'm never going to be able to come back in this channel again after this, but damn was it worth it to see that...
19:21@caker heh
19:21Efudd hilarity
19:21@caker ass is a curse word?
19:22Efudd apparently. :)
19:27Dave ass = arse
19:34mikegrb heh
19:34mikegrb that is a good one
19:35--- ---> Sh8d0w [] has joined #linode
20:18* caker wins an 3524-XL-EN auction
20:19@caker 20% improvement with a kernel compile with the sysemu patches ...
20:19@caker pretty freaking good
20:19@caker 5m15.182s with SYSEMU, 6m23.646s without
20:23@caker ok, second round was only 12s faster
20:24@caker compiling is mostly user-space, anyhow... need something that makes mega syscalls
20:24* caker looks for mysql bench or test or something
20:33--- ---> Newsome [] has joined #linode
20:49Sh8d0w how do people create hardlinks inside an initrd?
20:49@caker from/to the initrd itself?
20:50Sh8d0w hardlinks of directories inside the initrd
20:50@caker I don't think you can hardlink a directory to begin with ...
20:50@caker also, why do you need to use hardlinks?
20:51Sh8d0w that was the structure of the original initrd, i just want to recreate it
20:51@caker I would try mkdir'ing the directories, and then hardlink'ing each file ..
20:52Sh8d0w they're all empty dir's
21:00--- ---> krag [] has joined #linode
21:01@caker howdy
21:06SupaDongzu boy howdy
21:06SupaDongzu you want some PIE, boy?
21:06SupaDongzu I say, d'you want some PIE, boy?
21:06@caker Power = Current * Volts
21:06SupaDongzu Well then get you some PIE, boy!
21:07SupaDongzu Resistance = Volts / Current
21:07SupaDongzu May 27 17:35:11 zork chfn[21516]: changed user `octal' information
21:07SupaDongzu mcm
21:10@caker mcm?
21:10--- <<-- fo0barf [] has quit (Ping timeout: 480 seconds)
21:10SupaDongzu multi-channel madness
21:12* caker wonders why "Hit me with your best shot" just popped into his brain
21:13SupaDongzu perhaps you should be checked for tetanus
21:13SupaDongzu I hear that in addition to lockjaw, you might hear irritating pop songs
21:13@caker interesting
21:14@caker Common first signs of tetanus are a headache and muscular stiffness in the jaw (lockjaw) followed by stiffness of the neck, difficulty in swallowing, rigidity of abdominal muscles, spasms, sweating and fever, and hearing annoying pop songs.
21:14@caker FACT: Tetanus is caused by a toxin produced by a type of bacteria found worldwide in soil, dust and manure
21:15SupaDongzu the same manure they use to manufacture top-40 hits
21:15SupaDongzu thus the connection
21:15@caker hehe
21:16--- ---> fo0barf [] has joined #linode
21:25--- <<-- krag [] has quit (Quit: Leaving)
21:39--- <<-- csy [] has quit (Ping timeout: 480 seconds)
23:02--- <<-- Griswald [] has quit (Quit: Pie is good.)
23:03--- ---> jft [~42be5727@] has joined #linode
23:03jft hello
23:05tjfontaine hi
23:05--- <<-- jft [~42be5727@] has quit (Quit: )
23:05--- ---> jft [~42be5727@] has joined #linode
23:05jft hello
23:05--- <<-- jft [~42be5727@] has quit (Quit: )
23:06--- ---> jft [~42be5727@] has joined #linode
23:09--- <<-- jft [~42be5727@] has quit (Quit: )
23:21--- ---> jft [~42be5727@] has joined #linode
23:21@caker hello
23:21jft hi
23:21jft IE seems to work better than Mozilla
23:21jft caker: I really love linode
23:22@caker jft: glad you like it :)
23:22jft do you ever sleep? :-)
23:22tjfontaine I love linode and its hoster!}
23:22tjfontaine ;-)
23:23jft caker: do you remember if the 2.4.26 kernels have transparent proxy support enabled?
23:24@caker jft: what .config option is that?
23:25jft um, let me hunt it down.
23:28@caker CONFIG_NETFILTER=y
23:29@caker those are the only two that mention transparent proxies
23:29@caker I think you've got what you need already ...
23:29fo0bar I find it ironic that today I replaced a $15k cisco vpn concentrator with linux pptp/mppe on a spare box :)
23:30fo0bar s/ironic/amusing/
23:30fo0bar IT'S LIKE RAAAAAAAIN....
23:30jft yea, i was just seeing those on
23:31jft caker : about how many servers do you run in each DC?
23:31@caker jft: eight at ThePlanet, and .. 16 at HE
23:32jft wow, i thought it would have been more, seeing as how quickly they sell out
23:32fo0bar damn, so logic dictates you're up to host24?
23:33* fo0bar remembers when he got in on shiny new host16
23:33@caker is that right? 25 is coming online soon
23:33jft which vendor do you use for the servers?
23:34@caker jft: SuperMicro
23:34@caker jft: get the boxes from mwave, RAM from direct
23:34fo0bar when Linode is a big corporation with thousands of hosts and hordes of customers, I can rock out with the "back in MY day, I was on host16, and it was buried in snow, uphill, both ways!"
23:34@caker muhahahaa
23:34@caker hopefully I'll still have hair by then...
23:35* caker has LinodeID's #1 and #2 :-P
23:36fo0bar where can you see your ID?
23:36@caker I don't think you can anywhere .. you're #817
23:37fo0bar I kinda doubt you have ~1k active customers, so I assume you're not recycling the IDs
23:37@caker autoincrement
23:37fo0bar yeah
23:39@caker ordered up two more servers, two DS9's, a 3524-XL-EN, and a 1000B-t GBIC today
23:39@caker now I just have to pay for it all :)
23:40@caker Financing some of this stuff is kind of attractive.. somehow it's easier to swallow paying a few hundred/mo for a bit, rather than laying out a few grand for cisco gear
23:40jft what is a ds9 and a 3524-xl-en ?
23:40@caker console units, and cisco switch
23:40jft ah
23:40@caker i kind of jumpped the gun on the 3524, but no biggie
23:41fo0bar our snmp polling system (used for calculating bandwidth to bill) takes a poll for every device every 5 minutes. the poll database has a primary key, and is currently up to.... 122,048,767
23:41jft I am surprised you need over 100 MB
23:41fo0bar of course poll #1 was my workstation :)
23:41fo0bar same idea
23:41@caker jft: that would be for uplinks/cross-connects
23:41@caker fo0bar: nice :
23:41@caker :), even
23:41jft ah, your gear isnt located in the same rack?
23:42@caker cross-connects for redundancy .. something like this: for a new build out (still not certain who will receive my business)
23:42fo0bar caker: heh, I wrote this to use postgres, which admittedly I didn't know much about back then... particularly, I didn't know you had to vacuum periodically
23:42@caker fo0bar: is that like optimize in mysql?
23:43@caker fo0bar: or does that reset autoincrements or some dealio?
23:43jft linux router? interesting.
23:43fo0bar so after about 8 months, I check the script and notice that it's taking about 18 minutes to run (on a crontab that runs every 5 minutes)
23:43@caker jft: that's not decided for certain, either
23:43@caker hahaha
23:43jft fb:heheh
23:43@caker sounds like my old job .. I was made head of a performance team for a CFM/SQL-Server application
23:44@caker lots of stories of crap like that
23:44jft cold fusion?
23:44@caker no indexes on tables with millions of rows, etc
23:44fo0bar caker: yeah, a little like optimize, but postgres also keeps the old row hidden (but in the database) if you do an update or a delete
23:44@caker really crappy unions/joins, etc..
23:44--- <<-- Newsome [] has quit (Quit: Leaving)
23:44fo0bar and of course my script was updating a customer record row for each customer every 5 minutes
23:44fo0bar vacuuming just cleans those out
23:45@caker I had one-minute samples per Linode for bandwidth placed into the database ..
23:45@caker that didn't last long :)
23:45fo0bar among other things
23:45jft caker: what do you use for your intervals, now?
23:45@caker cleans out updates?
23:46@caker jft: it does a sum/update on one record per linode per month
23:46jft ah
23:46fo0bar anyways, I never noticed this before because when the 5 minute cron job ran, it processed each customer and device in the same order it did last time... so there would be 3-4 cron jobs running in parallel, but amazingly it all worked out cause each device was still being polled approx every 5 minutes!
23:46@caker oh man
23:47fo0bar anyways, I did a vacuum, and an 18 minute cron job turned into a 40 second cron job :)
23:47@caker so in postgresql, when you do an update to a row, it .. fragments the data?
23:47jft did you put vacuum in cron, also?
23:48@caker duh, missed your response ..
23:49fo0bar caker: not really, it just makes the database HUGE. so a 400-row device database becomes millions of rows, because the interface counter fields were being updated for every row every 5 minutes
23:49fo0bar selects are still fast, but write operations slow down linearly over time
23:49@caker fo0bar: gotcha
23:49fo0bar jft: no, I just vacuum it every month or, or when I remember :)
23:49fo0bar I really should cron it though
23:49fo0bar err or so
23:53jft caker: does the host linode OS run a journaling fs?
23:54--- <--- UML_ChanLog [~stats@] has left #linode (Rotating Logs)
23:54--- ---> UML_ChanLog [~stats@] has joined #linode
23:54jft if so, do you still need to run one inside a uml instance?
23:54fo0bar UML_ChanLog really needs to install ntp or something
23:55@caker jft: yes, the hosts are ext3. I've questioned if ext2 with large inode support would have been better, but if it works ...
23:56jft do the uml instances need to run a journaling fs, too?
23:56@caker jft: yes, they do by default (ext3 also)
23:56jft how come?
23:57@caker well, there's not much for the host to journal, as far as meta data goes, not much changes
23:57@caker and a journal inside the UMLs helps recovery time, in case of a crash/whatever
23:58@caker there's also a disconnect between when the UML thinks data is commited to disk, and when the host actually flushes its buffers
23:58jft ah, ok that is what I wasnt sure about
23:58@caker you can get around that by running UML is "sync" mode, but it's awfully slow
23:58@caker s/is/in/
23:59* caker waits for AIO and O_DIRECT support for UML on 2.6 hosts
---Logclosed Fri May 28 00:00:09 2004