Back to Home / #linode / 2005 / 11 / Prev Day | Next Day
#linode IRC Logs for 2005-11-10

---Logopened Thu Nov 10 00:00:11 2005
00:00|-|ksmith99 [~ksmith99@] has quit [Read error: Connection reset by peer]
01:26<fake>how long does ticket response take? it's not urgent but just wondering.
01:29<@caker>fake: ask, and you shall receive
01:34<fake>ok, prepare for dumb question.
01:34<fake>oh wait
01:34<fake>you just answered it
01:35<fake>thanks ;p
01:35<fake>how's linode business?
01:35<fake>brb, rebooting.
01:36|-|fake [] has quit [Quit: leaving]
02:03<[|^__^|]>that's odd
02:04[~][|^__^|] waits for SupaZubon to ping out
02:04<[|^__^|]>I wonder what hit it
02:04[~][|^__^|] hopes that this is just network and not host troubles
02:06[~][|^__^|] can get to
02:06<[|^__^|]>but no further
02:06<[|^__^|]>not even to the host
02:07<[|^__^|]>caker: ssh: connect to host port 22: Connection timed out
02:07<[|^__^|]>caker: are you seeing that?
02:08|-|anderiv [] has quit [Ping timeout: 480 seconds]
02:08|-|SupaZubon [] has quit [Ping timeout: 480 seconds]
02:08<cmantito>apparantly :P
02:08<[|^__^|]>there we go
02:08<cmantito>not working for me either.
02:08<cmantito>even though it's not my host =P
02:10<[|^__^|]>okay, ticket filed
02:11<@caker>[|^__^|]: already on it
02:11<[|^__^|]>caker: many thanks.
02:21|-|srw [] has joined #linode
02:23<srw>Hello? Is there a problem with host37?
02:23<[|^__^|]>srw: yeah, caker is on it
02:24<srw>Great. Thanks. I guess I should get some sleep then. :-)
02:24<[|^__^|]>good idea
02:25|-|srw [] has left #linode []
02:28<@caker>Is there a kernel arg to prevent raid autostart?
02:31|-|fake [] has joined #linode
02:31<fake>cool, free ram ;p
02:33|-|anderiv [] has joined #linode
02:57<linbot>New news from forums: Host37 Dead Drive in System and Network Status <>
03:08<@caker>1 mil hours MTBF my ass
03:09|-|SupaZubon [] has joined #linode
03:12<fo0bar>caker: you know how they come up with those numbers, right?
03:13<@caker>fo0bar: yes
03:14<fo0bar>encode: heh. they start up something like 1000 drives, and if one fails after 100 hours, that means it's 1000 * 100 MTBF :)
03:15<fo0bar>caker: also, you were looking for me a few days ago?
03:15<@caker>fo0bar: yeah, but I don't remember why now
03:15<@caker>maybe to ask you something about Finnix+Linode
03:15<fo0bar>it works
03:15<fo0bar>under xen too
03:16<@caker>(xen2, btw?)
03:16<@caker>not that it matters to your distro
03:16<fo0bar>the only difference I had to make was starting up tty1 instead of tty0 for the shell
03:16<fo0bar>the rest is just how to parse /proc/cpu to make stuff look pretty
03:16<@caker>Did you make /dev/ubd.. nodes?
03:16<@caker>or do the ide=3 trickery?
03:18<fo0bar>yes. I did /dev/udb{a..h}, /dev/udb{a..h}{0..15}, and /dev/ubd/{0..7}, just to cover all grounds
03:18<fo0bar> /dev/ubd{0..7} to
03:19<fo0bar>heh, it's just 10k more in device nodes ;)
03:19<fo0bar>I suppose if I were really cool, I would get udev working
03:20<@caker>I've probably not given it the time it deserves. I've been soured by devfs
03:21<@caker>so I rebooted host37 into a 2.6.14 kernel I've had prepared...
03:21<@caker>they fixed cfq-ts in this version, after a long stint in mm for the past two minors (I think)
03:21<fo0bar>yeah. "devfs is the future! switch everything to devfs! oops, we don't like devfs, bye bye"
03:21<@caker>I'm impressed
03:22<@caker>it gave disk time to my "find /" process while all the nodes were a-booting. I actually got output
03:22<@caker>and output frequently, at that.
03:22<fo0bar>caker: what does "ide=3" entail on UML? I'm having trouble googling for uml and ide :)
03:22<@caker>That's been my usual test, and no other kernels come close (at least in that scenario, raging read/writes)
03:23<@caker>fo0bar: I think that's the throw. it turns ubd's major to 3 .. it might be "ubd=3"
03:24<@caker>but, major 3 has 32 minors, and ubd's only have 16, so I think stuff gets wacked if it expects partitions
03:24<@caker>which, I've honestly never seen a UML disk image that was partitioned
03:24[~]fo0bar tests
03:24<@caker>./linux --help
03:25<fo0bar>wow, I didn't know you could pass --help
03:25[~]fo0bar looks
03:25<@caker>also: --showconfig
03:27|-|rafx [] has joined #linode
03:28<fo0bar>I don't know how much you've played with cow, but it's working fine for finnix
03:28<fo0bar>TMPCOW=`mktemp -p .`
03:28<fo0bar>rm -f "${TMPCOW}"
03:28<fo0bar>./2.6.14-um-bs1 mem=64M fake_ide fakehd con=null con0=fd:0,fd:1 devfs=nomount root=/dev/ubda ubda=${TMPCOW},finnix.img ubdb=finnix_swap.img eth0=tuntap,finnix,fe:fd:42:dc:01:f7 uml sshd
03:28<fo0bar>rm -f "${TMPCOW}"
03:29<fo0bar>I suppose you could reuse the cow file, but where's the fun in that? :)
03:29<@caker>that reminds me what I wanted to ask you
03:30<@caker>Wouldn't it be possible to run Finnix read-only (in Linode's environment, for a "rescue/tools" image)
03:31<@caker>I was thinking about a shared copy on each host, that ran read-only and wouldn't require deployment/disk space that way (supported by some neato wiz-bang interface in the LPM)
03:32<fo0bar>... not without a custom kernel or essentially rewriting finnix completely. finnix normally works by booting up a kernel with unionfs support, which merges the read-only media with a ramdisk
03:32<fo0bar>I can't see a way around it without either using uml cow, or some horrible hacks
03:32<@caker>What's the min. size for what's copied into the ramdisk?
03:33<fo0bar>but I'll think about it
03:33<fo0bar>at the end of bootup, about 6MB, give or take
03:33<@caker>That'll work under Linode kernels (they support ramdisks)
03:33<@caker>that's nothing
03:34<@caker>just like it's booting from a CD
03:34<fo0bar>but that's with unionfs, which only copies files as they're needed
03:34<@caker>aye ... which requires kernel patches?
03:35<fo0bar>yeah. like I said, I'll think about alternatives though
03:35<@caker>well, could always have a special rescue kernel to go along with the rescue image
03:35<rafx>odd...i wonder how my linode was powered off 15 minutes ago. didn't see shutdown request in panel, and didn't see shutdowm command in history after rebooting
03:35<@caker>rafx: check Lish's "logview" command?
03:36<rafx>holy crap, up late tonite eh? :)
03:36<@caker>server emergency a bit ago got me up :/
03:36<rafx>ouch, hope its under control
03:37<@caker>thanks :)
03:38<rafx>heh, i just did logview and was reminded about how i accidentally tried to cat a binary file. screen's all garbled
03:39<@caker>rafx: which username?
03:41<fo0bar>caker: I think it can be done with an initrd and some careful pre-setup, but my vote for a rescue image is still a temporary COW image that is just deleted after each run
03:42<fo0bar>the initrd would essentially have to set up the environment the way that livecds used to do it, before the enlightened days of unionfs
03:43<fo0bar>(IE, create ramdisk, mount image, make symlinks, copy /etc and /var and others to ramdisk, pivot_root)
03:49|-|rafx [] has quit [Quit: Leaving]
03:53|-|andrewjw [] has joined #linode
04:12|-|dddd44 [dhb55@] has joined #linode
04:14|-|dddd44 [dhb55@] has quit [Read error: Connection reset by peer]
04:21|-|dddd44 [dhb55@] has joined #linode
04:29|-|dddd44 [dhb55@] has quit [Read error: Connection reset by peer]
04:31|-|dddd44 [dhb55@] has joined #linode
04:40|-|dddd44 [dhb55@] has quit [Read error: Connection reset by peer]
04:41|-|internat [] has joined #linode
04:43|-|zsek [~zsek@] has joined #linode
04:45|-|zsek [~zsek@] has left #linode []
05:36|-|spunky [] has joined #linode
05:37|-|spunky [] has quit [Quit: ]
05:39|-|internat [] has left #linode [Leaving]
06:10|-|abock [] has joined #linode
06:10|-|abock [] has quit [Quit: ]
07:12|-|thoth39 [] has joined #linode
07:24|-|linville [] has joined #linode
07:44|-|javafueled [] has joined #linode
08:09<linbot>New news from forums: Slackware 10.2 in Feature Request/Bug Report <>
08:16|-|ksmith99 [] has joined #linode
08:33|-|alnr [] has left #linode []
08:33|-|alnr [] has joined #linode
08:43<ksmith99>does anyone know how to get a hold of caker??
08:43<ksmith99>long story, but I can't login to my account
08:44<bendy24>ping him
08:55<ksmith99>bendy24, thanks... pinged last nite and this morning
09:07|-|thoth39 [] has quit [Quit: Leaving]
09:08|-|Redgore [~Redgore@] has joined #linode
09:13|-|emcnabb [] has quit [Quit: IRC: Where men are men, women are men, and little girls are FBI agents]
09:19|-|ksmith99 [] has quit [Quit: ksmith99]
09:25|-|spr [] has joined #linode
09:33|-|D[a]rkbeholder [] has joined #linode
09:33|-|darkbeholder [] has quit [Read error: Connection reset by peer]
09:38|-|emcnabb [] has joined #linode
09:52|-|dddd44 [dhb55@] has joined #linode
10:01|-|dddd44 [dhb55@] has quit [Ping timeout: 480 seconds]
10:19|-|spr [] has quit [Quit: Spoon!]
10:26|-|tvakah [] has quit [Quit: leaving]
10:38|-|Dreamer3 [] has joined #linode
10:38<Dreamer3>is host 22 ok?
10:39<Dreamer3>i can't get a login to host itself and my box is SLOW
10:40<linbot>New news from forums: Slackware 10.2 in Feature Request/Bug Report <> || Host37 Dead Drive in System and Network Status <> || LASIK surgery in /dev/random <> || Apache or BlueDragon config problem in Linux, Apache, Mysql and PHP (LAMP) Forum <> || Kernel
10:40<linbot>New news from wiki: User:Javafueled <> || User:Javafueled <> || User:Javafueled <> || Multiple IPs <>
10:41<Dreamer3>mikegrb: it's been kinda slow lately in gen... though it was just memory load on my box, so i was going to close some things and it's dragging
10:42<javafueled>the bot is slow on that announcement.
10:43<@mikegrb>Dreamer3: swap thrasher, should be getting much better now
10:44<risto>phew... what an episode trying to use GoDaddy's VPS...
10:44<risto>lesson learned: don't try it again
10:44<risto>they have apparently fast systems and things work well... then just suspend you
10:44<risto>because I had an IRC server...
10:45<Dreamer3>mikegrb: hmm
10:45<Dreamer3>mikegrb: is it mine?
10:45<Dreamer3>mikegrb: i only have 1000 tokens
10:45<javafueled>risto: yeah, i looked at that EULA at GD and bailed.
10:45<Dreamer3>is that right?
10:45<risto>java, well, that is not documented
10:46<risto>they say not to allow "anonymous services"
10:46<javafueled>it was a lot of "reading between the lines."
10:46<@mikegrb>Dreamer3: what username?
10:46<risto>and I had to suck information 3 days to have them explain what happened
10:46<risto>I first thought it was my outbound email volume
10:46<risto>btw, they block outbound 25... you have to use their relays
10:46<@mikegrb>oh, yes, it is you
10:47[~]Dreamer3 frowns.
10:47<@mikegrb>or rather, were
10:47<risto>then one time they started blocking inbound 6667
10:47<@mikegrb>your rate doesn't seem so bad now
10:47<risto>no worries, mine runs 6660-6670 :-)
10:47<risto>then they just suspended me...
10:47<Dreamer3>mikegrb: memory pressure is too high though, i'll have to change somet higns
10:48<javafueled>GD, cheeky bastards.
10:48<risto>luckily I was able to get them let me run the system for 30 minutes so I could do final rsync...
10:48<risto>at that time all my stuff had long been back in linode
10:49<risto>GD might be good if you don't do anything weird
10:49<risto>if you just want root, apache and mysql
10:49<javafueled>IRC is weird? :)
10:49<@mikegrb>that leaves bendy24 out
10:49<@mikegrb>javafueled: most people ban it
10:49<risto>they're cheap... they gave me 30% discount for annual plan, too
10:49<risto>they use virtuozzo
10:50<javafueled>mikegrb: true. never figured out why though. bandwidth?
10:50<risto>but then... you get what you buy... faceless support (no mikegrb or caker), no community, no consoles (you're on your own if you mess your iptables)
10:50<javafueled>too risque? abuse of what? oh, robots and zombies.
10:50<@mikegrb>just about every dos targetting a linode that I know of was targeting a linode running an irc server
10:50<@mikegrb>that too
10:51<Dreamer3>mikegrb: do you have link to caker's mysql config stuff?
10:51<@mikegrb>somewhere in there
10:52<@mikegrb>make sure you add "skip-innodb"
10:52<Dreamer3>mikegrb: yeah, i nkow that :)
10:53<risto>well, now I'm back in linode... and bigger plan so running out of io_tokens shouldn't be a problem anymore :-)
11:00<risto>btw, host53.... is that in TP in Dallas?
11:01<Dreamer3>mikegrb: any way to get io stats from inside the box?
11:01<risto>cat /proc/io_status
11:04<risto>I have a little script that looks for tokens and if they go too low, they can run a script and hopefully stop the drain
11:05<risto>and when they're back to normal, they can run another to start things up again
11:05<risto>looks for the system load, too
11:05<Dreamer3>anyone else run Ruby on Rails on a linode?
11:06<tjfontaine>interesting script
11:07<risto>tj, any feedback is welcome
11:07<Dreamer3>anyone know how to reduce memory usage for spamd?
11:07<kvandivo>/etc/init.d/spamd stop
11:07<risto>dreamer, I haven't found any way...
11:08<risto>how much yours take? mine is 20-30M
11:08<risto>kvand, or rm /usre/share/spamassassin/*.cf
11:08<Dreamer3>risto: two processes 24mb, 17mb
11:08<Dreamer3>risto: and a smaller thirs
11:08<Dreamer3>risto: and a smaller third
11:08<risto>dreamer, sounds normal to me
11:08<Dreamer3>i have the config set to run two
11:08<risto>oh, give -m 1
11:09<tjfontaine>pool your money together and get a dedicated spamd server and resell to folks who'd like the service
11:09<risto>that way it won't spawn more than one child
11:09<tjfontaine>I'd pay $5/mnth for a dedicated av/spam service
11:09<risto>dreamer, or install greylist and run it before spamd
11:09<tjfontaine>even if it were just clamd+spamd
11:10<risto>thay way you'll reduce the load on spamd significantly
11:10<Dreamer3>risto: i have greylist, can't remember where in the process it runs
11:10<risto>what you mean where in process? I'm using sendmail and run them as milters
11:11<risto>tj, that might work :-)
11:11<tjfontaine>if I'm not mistaken on how grelisting works, it would have to be the first thing that happens regardless, just SERVICE TEMP UNAVAILABLE
11:12<risto>yes, greylist should be first
11:12<@mikegrb>if you are using greylisting, concider trying just that
11:12<@mikegrb>it's all I use
11:12<risto>no point putting them in other order
11:12<@mikegrb>I get one or two spam messages a day now
11:12<@mikegrb>and normally those are from debian mailing lists
11:13<tjfontaine>I'm not too thrilled with the prospect of greylisting
11:13<risto>does greylisting work well for viruses, too?
11:13<risto>clamav isn't that heavy as spamd
11:13<npmr>tjfontaine, i wasn't either until i gave it a try
11:13<risto>at least memorywise
11:14<risto>same here... I tried and now I wouldn't go without greylisting
11:15[~]mikegrb two
11:15<risto>greylisting probably blocks 90% of spam, then spamd rest of the 90%... it's pretty clean after that
11:15<@mikegrb>especially concidering it is all I use now
11:15<risto>numbers are just swags :-)
11:16<@mikegrb>can't imagine going back to the memory consumption of spamd
11:16[~]npmr uses bogofilter
11:17<risto>ok, if someone runs spamd as servvice... how do you prevent bayesian database from being contaminated?
11:17<risto>just turn bayesian off would work but bayesian is a good thing to use
11:18|-|ksmith99 [] has joined #linode
11:18<risto>hmmm, I think it could be done per-user bayesian database
11:18[~]risto is just thinking out loud
11:19<tjfontaine>and aloud as well
11:22<risto>why couldn't linode provide such a service?
11:22|-|jcox [] has joined #linode
11:22<risto>or thinking of it even more.... linode should a way of their customers creating their services and then having their customers having a way to sign up
11:23<risto>let's say someone wants to have spamd service... they create it and linode adds it to "extra services", $1/month
11:23<risto>and then credits the creator of the service for 50% of revenues
11:24<risto>that way spamd service provider gets his own linode for free
11:24<jcox>Just poking around here... curious about level of satisfaction with linnode, uptime, etc... anybody care to comment?
11:24<tjfontaine>if you check the forums progress was made on doing so, adamg is working with linode on the service
11:24<risto>no cash would be paid out but credits could be used to buy linode services
11:24<tjfontaine>jcox: love it
11:24<tjfontaine>risto: there is already a mysql beta test going on
11:24|-|Newsome [] has joined #linode
11:24<risto>tj, exactly... I know
11:25|-|optikal [~optikal@] has quit [Read error: Connection reset by peer]
11:25<jcox>great to hear... does it feel like your're scraping along at the minimum cpu rate most of the time, or does it usually feel a lot faster ?
11:25<tjfontaine>risto: for the status of the service you'd have to ask the @'s
11:26|-|optikal [~optikal@] has joined #linode
11:26<risto>see, the problem isn't spinning up linode & spamd
11:26<risto>that's trivial... but how do I charge those $1/month
11:26<jcox>just knew about usermode linux from kernel debug... 1st time i saw there is a virtual linux host based on it... seems like a great idea if the quality of serivce is there, bandwidth, & whatnot....
11:27<risto>jcox, yes it is great service
11:27<jcox>was asked to setup a cheap & cheerful website for a non-profit.. but.. don't want to get stuck maintaining hardware ;)
11:27<risto>jcox, you just need to know your stuff
11:27<jcox>yet.. they wanted all kinds of features that implied me writing custom code
11:27<jcox>(I'm a software engineer)
11:28<risto>well, php & mysql is what I use... it's a quick way to create custom apps
11:28<risto>of course, pros use java
11:28<jcox>.. or perl.. or ruby or... <flamesuit on>
11:28<risto>or so called "pros"
11:29<jcox>I've ended up using just about everything over time.
11:29<risto>yep, all work and you pick your favourite... no flames needed
11:29<risto>but yeah, linode works great :-) I'm just back from GoDaddy's VPS
11:29<jcox>so... if there's trouble in paradise.. .how difficultl are the folks at linnode to contact?
11:30<risto>they worked great, too, but they were predators
11:30<jcox>yeah .. a friend of mine wrote up a comparison.. made me check out linnode
11:30<risto>jcox, it's easy... just open the ticket
11:30<jcox>how so?
11:30<jcox>how were they predators, that is.
11:30<risto>jcox, well... faceless support, no community, and dictatorial admins
11:31<risto>I was suspended because I supposedly violated their policies
11:31<jcox>"pfftt! to them"
11:31<risto>unwritten if I might add
11:31<jcox>ok yeah.. that's just out.
11:31<risto>I ran IRC server and some bots on it
11:31<jcox>I want this to be *** my ** box
11:31<risto>they just shut it down last Sunday... and it took me 3 days to even find out why
11:32<risto>jcox, yep, GD lets you first believe that, too
11:32<risto>I was linode customer before GD for 2 or so years, but the problem was I was running out of io_tokens in my 128 plan
11:32<risto>I didn't want to spend money first to upgrade and GD looked good at first sight
11:32<jcox>so at linnode, it's your machien, your policies... and as far as they are concerned, your'e jsut a RAM/CPU/disk/bandwidth load.
11:32<risto>and they are powerful systems using virtuozzo
11:33<risto>but they are big faceless business
11:33<risto>they don't care about you, just your money
11:33<tux_rulez>If there a program that will continually check a logfile then if there is a change will spit the change to the screen?
11:34<risto>linode comes with community, understanding admins (mikegrb/caker), consoles, open policies
11:34<risto>tux, md5sum :-)
11:34<risto>or do this:
11:34<risto>install rcs, and check the file in "ci -u file"
11:35<jcox>usually i have just set up complete boxes ... but trying to get away from that... really I just want CPU/disk/ram in exchange for money.... not that the non-profit is a porn site or anything (it teaches squash to inner city kids ) .... but on the principle of the thing, i realy want a clear seperation between service provision and admin policies.
11:35<risto>then say rcsdiff file
11:35<jcox>ok thsi is great to hear.
11:35<jcox>how long have you been with linnode now?
11:35<@mikegrb>tux_rulez: I think what you want is "tail -f file"
11:36<risto>jcox, linode is for you
11:36<risto>jcox, I signed up long time ago... 2002?
11:36<tux_rulez>mikegrb: Oh, my i didn't know tail would do that
11:36<risto>then I was gone for a month just until this Sunday night
11:36<jcox>how much downtime since you jioned?
11:37<risto>jcox, there has been some. One long due to datacenter outage, some due to failing disks, etc
11:37<jcox>(I'm looking at apolo, spry, emax, godaddy, verio, linnode, redwood, ev1servers )
11:37<risto>but in all cases admins did their best to communicate status and whatever was needed
11:38<jcox>how long were you out, and how long ago was it?
11:38<risto>and they were totally unavoidable
11:38<risto>oh, I don't remember.... TP power outage was rather long... 12h?
11:38<jcox>oh ok
11:38<risto>you can probably find the hiostory in the forum
11:38<@mikegrb>well, they would have been avoidable had the datacenter followed directions
11:39<@mikegrb>rather then do what they were asked they plugged all the servers from three seperate circuits into one
11:39<risto>mikegrb, probably... but my point being there was not much you could have done
11:39<risto>and screwups like that can happen to anyone
11:39<jcox>ok that kinda sux, but not the end of the world... if the overall service has been fine then i thnk that's prob good enough
11:39<risto>jcox, read the forums:
11:40|-|optikal [~optikal@] has quit [Quit: Leaving]
11:40<jcox>i will.. thanks
11:40<Dreamer3>anyone else running Ruby on Rails?
11:40<risto>I think the whole story is somewhere under System and Network Status
11:40<tux_rulez>mikegrb: Cool "tail -f file" is doing exactaly what I want, No I can see an up-to-date view of what is happing in the log.
11:41<risto>one big plus with Linode is the community (this irc & forums)... you will really stay up to date
11:41<kvandivo>what's even neater is that gnu tail can do tail -f * and watch all the files you list
11:41<risto>one time GD went down on me... no explanation or anything. Just some dang email 12h later.
11:42<jcox>This is probably an RTFM question.. but... what about disater recovery? If their hard drive crahes, are they backing up these virtual machines? if so .. how often?
11:42<risto>they have mirrored drives but you want your own backups
11:42<Dreamer3>jcox: i think backups are use responsibility, though the drives are mirrored
11:42<risto>I rsync my stuff out daily
11:43<jcox>i wuv rsync ;)
11:43<risto>ok... want to hear negative things about linode?
11:43<risto>you need to watch what you're running
11:43<jcox>in a sense, kinda too bad they don't offer offsite rsync to other boxes
11:44<jcox>that way i could just off load the entire thing
11:44<risto>they have good ways to io-limit your boxes to protect others
11:44<risto>well, that is good thing actually
11:44<risto>but you need to watch what you run and optimize it
11:45<kvandivo>basically, don't plan on actively having swap
11:45<risto>yep, buy a big enough plan
11:46<jcox>will prob have some jpg / gif ... prob mysql, maybe mod_perl or php.. apache, random scripts
11:46<jcox>is it hard to upgrade your plan as you go ?
11:47<kvandivo>(assuming there are linodes available at the level you want to go to)
11:47<jcox>yeah was actually kind of suprised that there were no nodes of a particular type available
11:47<kvandivo>they sell out quick
11:47<risto>actually that is a good question?
11:47<jcox>right now there are no el-chepos
11:47<risto>do you have to change the box if you upgrade?
11:48<kvandivo>it might happen behind the scenes, but you only notice if it you happen to change datacenters, at which point your ip will change
11:48<jcox>don't exactly know what this group will need yet... but i plan on setting up a minimal debian system ... might need to scale the disk space or CPU or bandwidth later.
11:49<kvandivo>mail server with spam checking?
11:49<risto>well, the pricing is such that you don't really want to do it that way
11:49<jcox>that's it, in a nutshell. Wanted to start them off small, grow it if necesary assuming it's not a huge hassle.
11:49<jcox>oh ?
11:49<risto>it's quickly cheaper to go to next plan level and not buy extra RAM/disk
11:49<jcox>have not worked it out yet.
11:49<jcox>can you check how you're doing in an automated way?
11:49<kvandivo>based on your specs a nicely tuned 80 or a 120 would do what you need, without mail/spam checking
11:49<jcox>if so , i could write a program to upgrade/downgrade myself! hahahahah
11:50<kvandivo>upgrading isn't quite _that_ easy
11:50<jcox> ;) hey.. got to ask, right? ;)
11:50<risto>buying extra RAM is easy
11:50<risto>buy & reboot basically
11:50<jcox>ah.. because of the limited availablity of physical nodes.
11:50<risto>but like I said, pricing isn't good for that... cheaper to go to next plan level quickly
11:51<risto>and then get extra disk, CPU, bandwidth, etc
11:51<jcox>do you have access to the bandwidth stats ?
11:51<risto>nice graphs, too
11:51<jcox>ah ok
11:52<risto>let me show you
11:52<jcox>actually as long as i can get the bandwidth number numerically, then i can monitor cpu load & disk myself.. and set up an automated email or something to the person who will ultimately maintina this ting...
11:54<jcox>so they can say "oops.. time to upgrade.. " -- I want to avoid having to babysit them... yet.. i want to make sure they don't get soaked with pay-as-you-go pricing... hence my desire to automate a reminder message, if possible.... I suppose this isn't critical, but it would be kinda nice.
11:55<kvandivo>i think you are over thinking how often upgrading will need to be done
11:55<jcox>risto/kvandivo: really appreciate the friendly help.
11:55<jcox>ah great
11:55<risto>there's similar graph for monthly
11:57<jcox>ok i think i'm sold .. seems like a great deal
11:57<jcox>thanks again
11:57<risto>yes it is... not the cheapest but great value
11:58<jcox>still ahve to check the others in my list, but so far this seems to be the sweet spot for what i need.
11:58[~]kvandivo thinks mikey and the Cake should set up some sort of referral program for these times when they don't type a single character and get a sale.
11:58<risto>what others you have on the list?
11:58<risto>kvandivo, yeah!
11:58<jcox>(I'm looking at apolo, spry, emax, godaddy, verio, linnode, redwood, ev1servers )
11:58<risto>scratch godaddy :-)
11:59<risto>been there, done that
11:59<kvandivo>get the t-shirt?
11:59<risto>get free RAM?
11:59<jcox>GoAwayDaddy ?
12:00<risto>yeah, although you don't have anything special so GD might work for you
12:00<risto>but you never know...
12:00<risto>they don't actually tell what you can and cannot do
12:00<jcox>that's a major turnoff
12:00<risto>they also claimed me using too much CPU/resources
12:01<risto>they use virtuozzo that has pretty good resource limiting, called beans
12:01<risto>but apparently they don't rely on that
12:01<risto>my load was averaging way below 1.0
12:01<@mikegrb>jcox: re overage/pay as you go... cost of bandwidth after the fact == same as before
12:02<jcox>I'm doing this as a sort of good-will thingy for the squash big brother folks... they're a nice bunch & I want to help them.... Yet, it's also kind of a forray into the world of doing this sort of thing for ppl who actualy want to pay me to do it.
12:02<risto>I think they overload their systems
12:02<@mikegrb>50 cents a gig
12:02|-|optikal [~optikal@] has joined #linode
12:02<@mikegrb>we don't do the 50 cents a gig per month to add bandwidth to your server and then $10/gb if you go over
12:03<risto>jcox, how many hits you're expecing?
12:03<jcox>That's the hard part -- no clue.
12:03<risto>bandwidth usually is the least worry
12:03<risto>worry about CPU and especially memoty
12:03<risto>memory rather
12:04<jcox>They have a fairly small community right now, but they may end up running other sorts of fund raising things in cooperation with other charities... that's what makes sizing this kind of difficult.
12:04<risto>two questions... will you need database backend? will you need inbound email?
12:04<Newsome>what sort of files will it be serving? static pages? big binaries? pages generated from a database?
12:04<jcox>yeah, will prob do mysql or something
12:05<jcox>might setup qmail and just forward email .. not sure yet.
12:05<risto>database backend can easily be CPU hog if you don't do it well
12:05<risto>jcox, well, if you need email then you have all the problems with spam filtering
12:06<risto>and some tools used will require a lot of memory, so account that when you buy the plan
12:06<jcox>at home i use tmda
12:06<jcox>love it
12:06<jcox>but that's too techy for them
12:06<kvandivo>completely removing incoming mail is probably the wisest thing i did with my linode
12:06<jcox>so i might just punt on that part.
12:06<risto>I'm referring to spamasssassin/spamd
12:06<risto>never heard of tmda
12:06|-|flatronf700B [~flatronf7@] has quit [Ping timeout: 480 seconds]
12:06<jcox>oh it works nicely with spammassisn
12:06<jcox>it's awesome
12:06<jcox>basically it is like this:
12:07<risto>spamassassin = memory hog
12:07<@mikegrb>ugh, tdma
12:07<@mikegrb>I get /thousands/ of tdma challenges a month for emails I never sent
12:07<kvandivo>it just uses the from: address? blech
12:08<risto>btw, running spamassassin would probably get GD guys freak out... so you're now crossing the geek limit ...
12:08<@mikegrb>so you don't get spam but in turn you spam someone else
12:08<kvandivo>no dout
12:08<kvandivo>doubt, even
12:08<jcox>whitelist / wildcardwhitelist -- get in.... blacklist / wildcardblacklist -- Bzzt. Then everythign else is "dark grey", but tagged by spamassin for optional removal. Now consider what's left as "light grey"... fot that, it issues a small challenge message, to which the sender must reply to unlock the inital email. If they dont' then wieth in a certain amount of time, the "light grey" mail in purgatory is sent to hell. That's it.
12:09<risto>oh no
12:09<jcox>For a while, I felt bad about issuing back a reply -onlychalleng because of the extra bandwidth.. but...
12:09<jcox>theni did a calculation..
12:09<jcox>at home, i get about 300 spams / per day
12:09<risto>that might work if you only do emails for yourself
12:09<risto>but you're thinking of doing it others, too
12:09<jcox>not suitable for here i think
12:09<jcox>so ..
12:10<@mikegrb>but what about people like me
12:10<kvandivo>there aren't people like you
12:10<@mikegrb>those several thousand tdma messages sure eat a lot of my bandwidth
12:10<@mikegrb>kvandivo: :<
12:10<jcox>if you add up the bandwidth of the 300 spam challenges / day, it' amounts to someting like 2 average jpgs.
12:10<kvandivo>mikegrb: you asked for it
12:10<risto>yes, or like in my case... I have some email lists with 1000+ people
12:10<jcox>because the challenges are tiny..
12:10<risto>I would not want to issue tdma's for email list traffic
12:11<jcox>well for a sophisticate user it has other options.. like you can have temp addresses, or keyword addressses
12:11<risto>like I said... maybe good for personal use but not for more than 1 person domain :-)
12:11<jcox>but again, that's just not appropraite for unwashed masses.
12:11<risto>and with low volume
12:11<risto>greylisting is good, too
12:12<jcox>will prob just tell them "use yahoo / google / whatever" for email
12:12<risto>it would probably do the same as your tdma alone
12:12<jcox>i tried so many things for myself.. was a loosing battle
12:12<kvandivo>jcox: i just reroute all incoming mail at the dns level to a better location.. works great
12:13<jcox>explain that
12:13<jcox>oh ..
12:13<risto>I have greylist->clamav->spamassassin with secondary MX in my home linux
12:13<risto>works great, and volume is thousands of emails every day
12:13<jcox>ah interesting
12:14<risto>but turnside is that those needs some CPU/memory
12:14<risto>160 plan might jus be enough but not if you have a lot of other things, too
12:14<kvandivo>kids these days.. think they have to have lots of memory to accomplish anything
12:16<jcox>I might just punt entirely on email for now.. but i like the MX approach as a way to offload the linnode
12:17<risto>you can outsource your email, too
12:18<jcox>Most of my geek life has been either admining my own box, or doing scientific apps .. . not really that experienced as an admin for a large number of users.. . though is someting worth learning
12:18<jcox>ic.. again using the MX trick ?
12:18<jcox>I'd hate to go over quota on spam! That would be genuinely irritating.
12:20<jcox>Or maybe i could just do form-based "contact us"... kind of cheezy, but possibly an acceptable low road. ;)
12:20<jcox>Have you ever farmed out email to another provider ?
12:21|-|jcox [] has quit [Remote host closed the connection]
12:21|-|ksmith99 [] has quit [Quit: ksmith99]
12:22|-|jcox [] has joined #linode
12:22<risto>nope... I have't. I have thought of it though.
12:22<jcox>ic.. well this is really helpful
12:23<jcox>thanks again
12:23<risto>yw :-) looking forward to see you back soon
12:23<jcox>thanks also kvandivo
12:23<risto>from linode account :-)
12:23<jcox>where r u folks?
12:23<jcox>boston here
12:23<risto>Hayward, CA
12:24<jcox>Used to be in Sunnnyvale... had to move back to boston for family reason... Now bracing for the long dark chill...
12:24|-|spr [] has joined #linode
12:25<kvandivo>is that analogous to the big dig?
12:25<risto>cool, I work right next to Sunnyvale (MV)
12:25<jcox>That's the big wet tunnel.. um.. wait.. no...
12:25<jcox> ;)
12:26|-|andrewjw [] has quit [Quit: using sirc version 2.211+KSIRC/1.3.12]
12:26<jcox>Big dig == enormous boondoggle. Leaking like mad still. Wish they put the same $$ into the subway system.
12:27<jcox>Boston subway is mostly good, except it closes too early, not enough trains... but .. excellent to not need a car in the city... it's like a baby-nyc.
12:27<kvandivo>i hear driving there is hideous
12:28<jcox>Not bad if you don't mind living in a modern-day Ben Hur movie. Exciting !
12:29<jcox>yeah ppl are aggressive drivers and the roads are poorly marked , confusing 1-ways .. so even if someone isn't an aggressive driver, they often end up doing stupid things.
12:30<jcox>However, it's a small enough city that we're in a reduced-anonymity zone. The antidote is to smile, or make non-angry eye contact. Then ppl tend to shrug or laugh... so it's not really mean-spirited, just on the intense side.
12:31<jcox>Driving is kind of a risk-embracing competitive game in Boston, but not hostile per-se.
12:32<jcox>With usermode linux... how does swap work?
12:33<jcox>If I have 240 M ram... and i create a swap partition... what does that actually mean in terms of performance? What's a resonable strategy for swap space in this context?
12:36<risto>you want to minimize use of swap
12:36<risto>you want to minimize swapping in and out
12:36<risto>you can use swap for some idle processes
12:36<risto>but don't let them come in and out to memory
12:38<jcox>ic.. any other artifacts of running in a virtual machine besides swap slowness?
12:38<risto>swap is not really slow
12:39<risto>but systems are io-bound and you don't want to do too much of it
12:40<jcox>so the real thing that pushes you into a plan tends to be RAM...
12:40<risto>yes, when you run out of the RAM
12:40<jcox>so much for tomcat...
12:40<risto>with active processes
12:40<jcox> lol ;)
12:41<risto>things like bash shells you just idle are fine going to swap
12:42|-|Dreamer3 [] has left #linode []
12:43<jcox>k.. I'm off.. nice chatting
12:43|-|jcox [] has quit [Remote host closed the connection]
12:44<risto>so do we get our referral bonus when he joins? ;-)
12:58<Newsome>unfortunately no
13:05<linbot>New news from wiki: User:Javafueled <> || User:Javafueled <> || User:Javafueled <> || Multiple IPs <>
13:07<javafueled>WTFudge? the bot okay? this news is 5 days old.
13:07<linbot>New news from forums: Slackware 10.2 in Feature Request/Bug Report <> || Host37 Dead Drive in System and Network Status <> || LASIK surgery in /dev/random <> || Apache or BlueDragon config problem in Linux, Apache, Mysql and PHP (LAMP) Forum <> || Kernel
13:19<tizen>hey mike
13:24|-|tux_rulez [] has quit [Ping timeout: 480 seconds]
13:50|-|cmantito [] has quit [Quit: Xchat uprade.]
13:53|-|cmantito [] has joined #linode
13:58|-|dddd44 [dhb55@] has joined #linode
14:00|-|womble [] has quit [Quit: What's behind the round window...]
14:17|-|JasonF [] has quit [Ping timeout: 480 seconds]
14:23|-|cmantito [] has quit [Quit: restarting xchat to enforce some settings changes]
14:23|-|cmantito [] has joined #linode
14:26|-|GN [] has joined #linode
14:28<GN>anything exciting going on in here?
14:31<linbot>New news from forums: Host51 Kernel Panic in System and Network Status <>
14:32<bendy24>kvandivo: it seems you are the only one to respond to that
14:33[~]kvandivo looks around.
14:35|-|abock [] has joined #linode
14:36<abock>host51 seems to be running extra, extra slow... at first my web site was timing out, so I checked my members page, and my linode had "turned off" during the night, and I had to boot the thing
14:36<tjfontaine>15:31 < linbot> New news from forums: Host51 Kernel Panic in System and Network Status <>
14:37<bendy24>thats like area51
14:37[~]taupehat hugs host15
14:38|-|spr [] has quit [Quit: Spoon!]
14:39<kvandivo>bendy24: area51 doesn't exist
14:39<bendy24>kvandivo: thats what "they" want you to believe
14:40<kvandivo>The Frisbees of Dreamland, however, are just plain cool
14:40<kvandivo>bonus points for anyone that got that reference
14:42[~]jams_ wishes for the new season
14:53|-|Redgore [~Redgore@] has quit [Quit: A geek without purpose - | SMDC-Network IRC -]
14:53|-|Redgore [~Redgore@] has joined #linode
15:17|-|Newsome [] has left #linode [Linux: Now with special employee pricing!]
15:22|-|JasonF [] has joined #linode
15:36|-|GN [] has quit [Quit: back to work :(]
15:45|-|javafueled [] has quit [Quit: javafueled]
16:02|-|spr [] has joined #linode
16:04|-|linville [] has quit [Quit: Leaving]
16:48|-|womble [] has joined #linode
16:49<thelsdj>anyone know if i cancel linode account does it stay active till end of my current billing period?
16:49<kvandivo>i think it depends on how you do it
16:51<thelsdj>ah hrm, guess i'll just ask caker
16:52<thelsdj>i like linode but was a bit expensive and just got my boss to give me free space on a xen so no real reason to keep payin (and i need to save cash heh)
16:52<kvandivo>cash is overrated
16:53<thelsdj>christmas season coming up gotta pay for airfare for my sister and me to visit family and probably moving into a new place so security deposit and crap
16:55<kvandivo>well, just speaking for myself.. i can safely say that there'll be an empty place in my heart and that linode just won't be the same once you've gone
16:55[~]Redgore passes kvandivo some tissues
16:55[~]kvandivo sniffs sadly.
17:05<thelsdj>yea, i am one of the 4 paying clients
17:09[~]tjfontaine wonders who the other 2 are
17:11<kvandivo>i hear there's a guy in utah...
17:11<kvandivo>don't know who the 4th would be, though
17:17|-|abock [] has quit [Ping timeout: 480 seconds]
17:17|-|Jeremy [] has quit [Write error: connection closed]
17:17|-|J[SS] [] has quit [Remote host closed the connection]
17:18|-|J[SS] [ph33r@2001:4830:2065::1] has joined #linode
17:18|-|Jeremy [~ph33r@2001:4830:2064::9] has joined #linode
18:06<linbot>I'm Jumpin' Jack Flash, It's a Gas Gas Gas!
18:16|-|jcox [] has joined #linode
18:16<@caker>jcox: hello
18:19<jcox> howdy
18:19<kvandivo>the man.. the man.. the caker, caker man
18:20<jcox> run run as fast as you can... you can't catch him he's the caker man...
18:20<jcox> ;)
18:26[~]caker suspends freeloaders
18:31|-|ronpoz_ [] has joined #linode
18:38<JasonF>nothing like pulling a plug to get someone's attentino
18:49|-|emcnabb [] has quit [Quit: IRC: Where men are men, women are men, and little girls are FBI agents]
18:51|-|shakr [] has quit [Remote host closed the connection]
18:55|-|guinea-pig [] has joined #linode
18:56[~]guinea-pig wonders what to do with mikegrb
18:57|-|LokiLV [] has quit [Remote host closed the connection]
18:57|-|LokiLV [] has joined #linode
19:03<thelsdj>problems with HE?
19:04|-|kvandivo [] has quit [Remote host closed the connection]
19:04|-|shakr [] has joined #linode
19:06<guinea-pig>nah, i just noticed i had a /dcc from mikegrb that's been going on for who knows how long...
19:07<guinea-pig>i hate it when people send me "null"
19:07<guinea-pig>well, that answers that question... oct 1.
19:09|-|kvandivo [] has joined #linode
19:10<thelsdj>i think maybe some routes are bad while others are fine
19:28<@caker>guinea-pig: get it sorted?
19:29<thelsdj>caker: if i cancel account does it still stay active till end of billing cycle?
19:29<@caker>thelsdj: nope :/
19:29<thelsdj>ah k, then i'll just wait till i know i'm done with it
19:30<tizen>hey caker
19:30<tizen>did you guys block my internet yesterday or was that a coincidence? account abci ?
19:31<@caker>thelsdj: I didn't and didn't hear anything from Mike about it
19:31<tizen>Ah ok
19:31<tizen>My CC was expired
19:31<tizen>and i was waiting for the new CC in the mail
19:31<tizen>but suddenly yesterday I had no internet to my node
19:31<tizen>but console was fine
19:32<tizen>so i used another CC temporarily
19:32<tizen>and rebooted
19:32<tizen>and it came back up fine
19:32<tizen>I was wondering if it was something you did to remind customers the week was up
19:32<@caker>It wasn't until about an hour or two that I suspended accounts, and that's very obvious when I do that (node shutsdown, only thing you can do when you log into the LPM is update CC and submit payment)
19:33<tizen>Oh, well maybe its on my side then
19:33<tizen>odd coincidence though
19:33<tizen>after 2 months without me logging in and running perfectly
19:33<@caker>I've heard grumblings about TP changing router MAC addresses
19:34<@caker>which required either to wait until your MAC table timed it out, until you pinged the gateway, or until you restarted networking (or rebooted)
19:34<@caker>but, that happeened earlier this week, not yesterday
19:34[~]caker shrugs
19:36<tizen>ok, fair enough
19:37<guinea-pig>caker: sorted? yeah, i think mike was just messing with me... a month ago.
19:37<tizen>thanks though
19:37<guinea-pig>why else would he be sending me a null file for 43 days? :P
20:41<fo0bar>caker: still around?
20:54|-|LokiLV [] has quit [Quit: We All Should Wonder Quis custodiet ipsos custodes!]
21:04|-|ronpoz_ [] has left #linode []
21:23|-|edlang [] has joined #linode
21:25<edlang>mikegrb: you there?
21:26<fo0bar>edlang: you still believe in mikegrb? hahaha, what's next, the tooth fairy?
21:27<edlang>Yes, well.
21:28<edlang>I'm trying to get a hold of the crackmonkey, but 1) my linode account is suspended and 2) I don't have a password on zork, so I can't log in.
21:28<edlang>and 3) I can't remember what fqdn he gave the old zork, if indeed it's still up
21:29<fo0bar>and isn't crackmonkey on this channel?
21:30<edlang>I thought gaspar was the new one
21:30<fo0bar>[|^__^|] == crackmonkey
21:31<fo0bar>no, frotz is the new linode one
21:31<edlang>I want the one without the OPIE stuff
21:31<edlang>ah hmm.
21:31<fo0bar>but I think gaspar is off unless requested
21:31<edlang>well, that's the one my key is on. I couldn't put it on the new one because I never had a password.
21:32<edlang>I'd always used RSA auth
21:32|-|dddd44 [dhb55@] has quit [Read error: Connection reset by peer]
21:32<fo0bar>hah, that sucks
21:32<guinea-pig>heh, same here
21:34<edlang>So... yeah.
21:39|-|edlang [] has left #linode []
22:31|-|jcox [] has quit [Remote host closed the connection]
22:59|-|VS_ChanLog [] has left #linode [Rotating Logs]
22:59|-|VS_ChanLog [] has joined #linode
23:09<[|^__^|]>I fucking copied over edlang's goddamn key
23:34|-|Newsome [] has joined #linode
23:54|-|womble [] has quit [Quit: What's behind the round window...]
23:56|-|spr [] has quit [Quit: Spoon!]
---Logclosed Fri Nov 11 00:00:11 2005