00:01.40 | trousers_ | What's it about? |
00:03.30 | catalyst | Vista |
00:03.34 | catalyst | presumably ;p |
00:03.43 | catalyst | yeah |
00:04.32 | catalyst | I installed vista because I get a free key from the uni. and the theme is nice but it's far too annoying to use for any length of time :( it keeps losing my DVD drive for a start |
00:04.32 | wethrin | Damn |
00:04.33 | wethrin | I'm too slow |
00:05.28 | catalyst | the thing which gets me about vista is those "personally signed copies" by bill gates XD |
00:11.19 | trousers_ | I have my Dad's (MSDN Subscriber) copy in my bedroom |
00:11.23 | trousers_ | Trying to Xen-ify :P |
00:12.00 | Leeds | sabinef72: pong? |
00:12.55 | wethrin | trousers_: Yay |
00:13.57 | trousers_ | |. | |
00:13.57 | wethrin | Hmm. Slayer's not up to his usual trolling form |
00:14.18 | trousers_ | wethrin: By the way, Linux club at my school is up and running :) |
00:14.27 | trousers_ | 3rd presentation tomorrow --- Xen. |
00:14.44 | wethrin | yay! |
00:40.02 | *** join/#gllug gregj (n=gj@81-186-226-63.citynet.pl) |
01:29.57 | gw280 | woo |
01:30.24 | wethrin | oh. It's just you. |
01:30.53 | wethrin | nn! |
01:34.52 | *** part/#gllug mcgrof (n=mcgrof@ruslug.rutgers.edu) |
03:30.01 | *** join/#gllug Puss1nB00ts (i=PussInBo@gateway/tor/x-18535f24a133f9fd) |
03:42.57 | *** join/#gllug Leeds (n=richardc@awork006188.netvigator.com) |
07:05.59 | morsing | 'morning |
07:45.38 | morsing | Nothing wrong with an older woman :) |
07:47.05 | Leeds | oi - wrong channel! |
07:48.00 | morsing | Hmm... You're right |
07:48.04 | morsing | How did that happen? |
07:48.27 | Leeds | I don't know... be more careful... this channel is logged and public :-) |
07:48.32 | morsing | :) |
07:48.39 | morsing | Mmm... young girls :) |
07:48.46 | morsing | There... I reversed it now |
07:48.53 | morsing | Neutralised |
07:48.57 | Leeds | eh, either way :-) |
08:27.17 | morsing | That's weird... |
08:27.24 | Leeds | who is? |
08:27.53 | morsing | Enom |
08:31.15 | Leeds | gym time |
08:31.46 | morsing | Your NameServer request was successful. |
08:31.48 | morsing | Blimey |
08:33.00 | morsing | Who'd have thought... After only three weeks |
08:39.07 | morsing | ~seen wethrin |
08:39.32 | ibot | wethrin is currently on #gllug. Has said a total of 48 messages. Is idling for 7h 8m 39s, last said: 'nn!'. |
09:01.37 | *** join/#gllug SlayerXP (n=martin@gw-colt.fotango.com) |
09:12.00 | morsing | SlayerXP! |
09:12.10 | morsing | Someone here know about mortgages? |
09:14.13 | SlayerXP | yes |
09:17.01 | morsing | SlayerXP: My fixed rate ends in October, looking at this |
09:17.02 | morsing | http://www.personal.barclays.co.uk/BRC1/jsp/brccontrol?site=pfs&task=articleFWwealth&value=9633&target=_self |
09:17.13 | morsing | What will my interest rate be after that? |
09:18.01 | SlayerXP | no idea |
09:18.12 | morsing | SlayerXP: You suck |
09:18.18 | SlayerXP | i've opted for an offset mortgage |
09:49.09 | sabinef72 | hello ! |
09:54.54 | morsing | sabinef72! |
09:58.53 | sabinef72 | hi morsing ! |
10:05.24 | *** join/#gllug huw-l (n=scmhl2@valiant.cs.cf.ac.uk) |
10:06.52 | bilarh_ | i'm so depressingly tired today |
10:07.03 | bilarh_ | anyone know of a good efnet server that is likely to let me play if i'm with zen? |
10:07.15 | bilarh_ | not demon, because that's down at the moment by the looks of it |
10:17.30 | bilarh_ | i need a good tool for managing my task list |
10:17.37 | bilarh_ | preferrably something that I can add notes to |
10:17.46 | bilarh_ | like little yellow stickie tabs or post its |
10:17.52 | bilarh_ | but not IRL, so to speak :) |
10:20.46 | *** join/#gllug Leeds (n=richardc@210.245.153.200) |
10:22.18 | morsing | Leeds! |
10:22.24 | Leeds | hi |
10:23.39 | sabinef72 | thank you Leeds ! I got it yesterday morning |
10:23.57 | Leeds | sabinef72: ah-ha! :-) |
10:24.14 | sabinef72 | it was a nice surprise ! |
10:24.39 | Leeds | I guess you liked it then... I like giving nice surprises :-) |
10:32.51 | Leeds | Japanese bakery time! |
10:33.00 | *** join/#gllug tripitaka (n=tripitak@83.217.101.115) |
10:38.28 | morsing | tripitaka! |
10:40.15 | tripitaka | hello |
10:40.51 | *** join/#gllug goibhniu (n=cillian@grail1.oxfam.org.uk) |
10:41.25 | morsing | goibhniu! |
10:42.45 | *** join/#gllug [Clive] (n=ch@chills.demon.co.uk) |
10:44.01 | goibhniu | morsing! |
10:49.35 | Leeds | and in my continuing quest to go slightly local... I just bought a bottle of Hello Kitty branded green tea :-0 |
10:50.35 | tripitaka | sure it's not catnip? |
10:51.39 | goibhniu | Hello Kitty can't still be in fashion, can it? I thought Goodbye Kitty was old already |
10:51.54 | Leeds | nope, pretty sure it's green tea... that's about the only English on the label (there's a local label stuck on with more English) and she's holding a green tea leaf |
10:52.01 | Leeds | goibhniu: huge... massive... very very popula |
10:52.36 | goibhniu | Leeds: have you got five hundred flashing burgers hanging off your phone yet? |
10:53.11 | Leeds | no, I do not |
10:54.45 | morsing | ~seen Kitty |
10:55.18 | ibot | kitty is currently on #gllug, last said: 'you have more letters to guess ?'. |
10:55.18 | Leeds | I also don't wear it hanging around my neck |
10:55.54 | Leeds | ibot Kitty |
10:55.59 | ibot | somebody said kitty was OMGKITTYCATMEWMEWMEWMEW |
10:57.33 | morsing | Nice |
11:00.52 | wethrin | hello |
11:06.08 | *** join/#gllug robo_ro (n=robo_ro@82-70-146-30.dsl.in-addr.zen.co.uk) |
11:06.18 | robo_ro | lo |
11:06.38 | Leeds | robo_ro! |
11:06.43 | robo_ro | :-) |
11:06.44 | Leeds | how are you, my friend? |
11:07.01 | Leeds | no, wait... that should be... who are you? my friend? |
11:07.10 | robo_ro | hmmm good... struggling with a monster sql statement... |
11:07.29 | robo_ro | er... quo vadis?!!! |
11:08.15 | robo_ro | leeds: you any good with sql stuff? |
11:08.23 | morsing | robo_ro! |
11:08.40 | Leeds | not hugely, no... |
11:08.43 | morsing | Is a graceful restart enough to read in a new ssl section/virtual host? |
11:08.52 | Leeds | morsing: should be |
11:08.55 | morsing | Cool |
11:10.27 | robo_ro | morsing! |
11:11.14 | Leeds | robo_ro: nobody can help you if you don't tell us what the problem is |
11:15.23 | *** join/#gllug rowanp01 (n=robo_ro@82-70-146-30.dsl.in-addr.zen.co.uk) |
11:15.37 | rowanp01 | shite our connection is jumpy today |
11:15.44 | tripitaka | no wait, I want to guess |
11:18.09 | robo_ro | hi sorry... my connection is going mad today |
11:18.39 | tripitaka | can anyone recommend a good approach for converting an odd timestamp (eg. 2007-1-30,5:24:31.0,-5:0) to epoch? Currently going in circles with perl Date::Manip |
11:18.57 | robo_ro | don't know what made it up in terms of my problem... but query needs to be something like: sum of stock.price grouped by month(order.purchasedate) and subgrouped by customer.country |
11:21.38 | goibhniu | Leeds: is Borat watching over you now? |
11:23.52 | Leeds | nope... still folded up, not put up |
11:31.33 | *** join/#gllug rowanp01 (n=robo_ro@82-70-146-30.dsl.in-addr.zen.co.uk) |
11:55.34 | bilarh_ | anyone dealt with lvm and kickstarting? |
11:56.00 | bilarh_ | and is it possible to have part of a drive under lvm management, or is it "all or nothing"? |
12:00.16 | Leeds | yes, it's possible |
12:00.54 | murb | bilarh_: when i don't have hardware raid i typically create 2 md devices |
12:01.00 | murb | one for root and the other for lvm. |
12:01.11 | bilarh_ | yeah but my raid1 is 280 gigs :) |
12:01.18 | bilarh_ | times three |
12:01.23 | murb | so? |
12:01.26 | bilarh_ | and i don't want 280gigs for system :) |
12:01.33 | bilarh_ | i want almost all of it for logging |
12:01.37 | bilarh_ | anyway, meeting,brb |
12:17.54 | wethrin | blarg. Random IM people-- |
12:18.06 | wethrin | "Good news. I got the offer" Who are you, and why do I give a shit? |
12:26.09 | morsing | wethrin! |
12:26.58 | morsing | Has anyone seen this? |
12:26.59 | morsing | http://www.imdb.com/title/tt0478049/ |
12:41.24 | keithlard | hmm, how do i grep for something AND NOT something else? |
12:41.34 | keithlard | eg 'error&!fatal' |
12:42.35 | bilarh_ | grep error | grep -v fatal |
12:42.45 | keithlard | in one regexp? |
12:42.50 | bilarh_ | dunno :) |
12:43.03 | wethrin | Don't think you can |
12:43.25 | bilarh_ | i'm trying to think of how to actually specify it even logically, but i can't think of it, given that you either match on exclusion or not |
12:43.38 | keithlard | ooh, i like a challenge |
12:44.06 | wethrin | keef! |
12:44.16 | morsing | Who was it who recommended ecards.msn to me last week? |
12:45.02 | morsing | Anyone here familiar with ecard.msn? |
12:45.09 | morsing | The Hallmark site? |
12:45.37 | goibhniu | :D |
12:46.28 | morsing | goibhniu: Do you? |
12:46.35 | goibhniu | morsing: weren't you going to buy a paper card instead? |
12:46.52 | goibhniu | morsing: do I look like someone who sends ecards? |
12:46.57 | morsing | goibhniu: I don't know if you'd normally do that. |
12:47.13 | goibhniu | morsing: I barely know my own birthday |
12:47.21 | morsing | goibhniu: Friends anniversary? Should I send a papercard? Isn't it only family who does that? |
12:48.00 | goibhniu | morsing: you're norwegian ... aren't you supposed to make a toast and sing a song for them? |
12:48.10 | morsing | I'm not norwegian though... |
12:48.36 | morsing | Leeds? |
12:49.44 | goibhniu | morsing: I did see some cute penguin ecards ... with lots of hearts on them ... any use? |
12:50.13 | goibhniu | morsing: how about a stirring powerpoint presentation with music? |
12:50.19 | spungles | Ecards are for sailors and gays. |
12:50.33 | spungles | if you're too cheap or lazy to send a real card, dont bother |
12:50.44 | goibhniu | spungles: we have no sex here |
12:51.00 | morsing | spungles: Thanks for the advice |
12:51.29 | morsing | ping Leeds |
12:55.27 | morsing | LEEEEEEDS! |
12:55.31 | morsing | ibot wake Leeds |
12:55.42 | ibot | ACTION throws a barrel-full of ice water on Leeds and shouts "GOOD MORNING!!!!" |
12:57.58 | goibhniu | morsing: if it's too late to send a card you could always make it personal ... take one of their wedding photographs and open it in paint ... and add rotting teeth, glasses and moustaches |
13:01.18 | morsing | goibhniu: It's not too late |
13:01.25 | morsing | You people all suck! |
13:03.49 | *** join/#gllug robo_ro (n=robo_ro@82-70-146-30.dsl.in-addr.zen.co.uk) |
13:04.59 | robo_ro | lo |
13:05.40 | robo_ro | bt engineer came to install another line in my office and since then my connection has been up and down like a bride's nightie |
13:05.45 | robo_ro | :-( |
13:08.28 | morsing | robo_ro! |
13:08.45 | robo_ro | lo morsing |
13:22.02 | morsing | robo_ro: Do you suck too? |
13:34.01 | robo_ro | like a henry vacuum cleaner |
13:48.11 | *** join/#gllug dick_turpin (n=peterc@host217-34-163-21.in-addr.btopenworld.com) |
13:48.17 | morsing | dick_turpin! |
13:48.45 | dick_turpin | morsing: Know anything about xbox360? |
13:49.12 | dick_turpin | Or anyone? |
13:53.44 | goibhniu | dick_turpin: iirc it has something to do with rugby |
13:54.12 | bilarh_ | anyone know of any god ISA's out there? |
13:54.20 | bilarh_ | HSBC offer 4.75% which sounds a bit rubbish, really |
13:54.33 | bilarh_ | oh, unless i pay 19.99 a month to be a "premier customer" |
13:54.38 | dick_turpin | goibhniu: ? |
13:54.41 | bilarh_ | in that case i get 5.15 :) |
13:54.48 | bilarh_ | so that doesn't sound like a very good deal to me :) |
13:56.05 | morsing | dick_turpin: Of course not! |
13:56.54 | dick_turpin | morsing: Neither do I, all I need to know is can you use an external USB HDD instead of the default HDD |
13:57.45 | *** join/#gllug catalyst (n=catalyst@allegro/user/angelchild) |
14:01.17 | morsing | catalyst! |
14:01.26 | morsing | dick_turpin: Yes |
14:04.17 | catalyst | morsing! |
14:04.31 | dick_turpin | morsing: OK many thanks I'll tell 'She who must be obeyed' that it was you who gave the professional advice, I'm too scared to take responsability :-) |
14:06.06 | morsing | dick_turpin: Well done |
14:07.27 | wethrin | goibhniu: LIES! |
14:08.04 | goibhniu | on "free software day" no less! |
14:08.36 | goibhniu | in fairness you did look a little drunk and confused |
14:08.46 | wethrin | It runs Linux |
14:08.52 | wethrin | It's my video-playing device |
14:09.05 | wethrin | was it you who had a spare remote control, or do I disremember? |
14:09.37 | wethrin | Yay |
14:09.57 | morsing | By magic obviously |
14:10.19 | wethrin | yup |
14:10.29 | dick_turpin | morsing: You 'B' I'm just sorting out a CentOS VM to get nagios going |
14:10.37 | mumperer | arternoon |
14:10.39 | morsing | B? |
14:10.52 | morsing | mumperer! |
14:11.00 | dick_turpin | morsing: Bugg*&R |
14:12.02 | mumperer | Hi there - everything sorted out peeps, or is there still room for discussion/humour/ribaldry etc? |
14:12.35 | wethrin | Depends where |
14:12.51 | mumperer | Ah - elbow room, eh? |
14:14.04 | wethrin | indeed |
14:15.18 | morsing | Why do I get this? |
14:15.18 | morsing | DNS CRITICAL - '/usr/bin/nslookup -sil' output parsing exited with no address |
14:16.03 | morsing | Anyonoe? |
14:16.05 | morsing | Hali? |
14:17.12 | dick_turpin | morsing: I've not got that far this will be my second attempt, I have some help stuff at home however I know many a good sysadmin has been crippled by nagios |
14:18.12 | hali | morsing: does the plugin work from command line? |
14:19.22 | goibhniu | wethrin: HA! I didn't have a spare control ... but I do have a microsoft mouse ... just in case anyone thinks I'm a hippy or something |
14:20.01 | wethrin | I've got a Microsoft mouse here too! |
14:20.19 | wethrin | morsing: Why don't you try running it from the command line, and see what it gives? |
14:21.16 | morsing | wethrin: It gives the same (That's how Nagios works) |
14:21.41 | wethrin | Yeah, I know |
14:21.57 | wethrin | But doing the stuff by hand might give some insight into where or why it goes wrong |
14:24.50 | *** join/#gllug clive-h (n=ch@chills.demon.co.uk) |
14:25.20 | morsing | clive-h! |
14:25.38 | clive-h | Hi morsing |
14:35.17 | tripitaka | morsing: are you passign an argument to it? |
14:35.42 | tripitaka | with no argument (ie. host to look up) nslookup goes into interactive mode |
14:42.22 | tripitaka | what gets me with nagios is that no perl I write will work without bizarre errors. I have to wrap my perl scripts in one line shell scripts, which it doesn't seem to mind. |
14:42.23 | morsing | tripitaka: 127.0.0.1 |
14:42.32 | morsing | tripitaka: It terminates instantly |
14:43.05 | clive-h | I did do perl scripts standalon using nagios |
14:43.17 | clive-h | why do you have to wrap them? |
14:43.50 | tripitaka | the perl scripts run fine from command line, but through nagios they throw interpreter errors |
14:44.07 | clive-h | errk |
14:44.29 | bilarh_ | env not set correctly? |
14:44.46 | clive-h | sounds rather like it |
14:45.10 | bilarh_ | tripitaka: what assumptions have you made while writing the module? |
14:45.24 | tripitaka | morsing: are you using check_dns or writing your own? |
14:45.51 | tripitaka | bilarh_: none that I'm aware of, all very simple stuff, with use: strict |
14:46.19 | tripitaka | I've just done one to check for clock skew - let me see what the errors were before wrapping it.. |
14:48.18 | morsing | check_dns |
14:53.09 | tripitaka | hmm. ok, maybe it was just bad code. nagios reports on perl errors which /usr/bin/perl itself misses |
14:54.00 | tripitaka | eg. I had tested for an empy variable with 'if ($string == "")'. pn the command line it works fine, but via nagios, I get '**ePN /usr/lib/nagios/plugins/check_time.pl: "Argument "" isn't numeric in numeric eq (==) at (eval 7)' |
14:54.18 | tripitaka | changing it to 'if ($string eq "")' makes it work in nagios |
14:54.36 | hali | yeah, i get those weird **ePN errors in nagios quite often |
14:54.44 | tripitaka | s/empy/empty |
14:54.51 | hali | could be related or oracles internel perl intepreter trying to do something |
14:57.38 | antiphase | eq is a string comparison operator, == is a numeric comparison operator |
14:58.26 | tripitaka | yep. I was wrong, but nagios's embedded perl died on the error, while perl with strict sailed past it |
14:58.39 | antiphase | If you 'use warnings; use strict;' these things should never bother you |
14:58.56 | antiphase | If you don't you should start programming VB.NET or something |
14:59.48 | tripitaka | I'll bear it in mind |
15:06.20 | bilarh_ | what FS would you chose for a 1400Gb filesystem with smallish files on it? |
15:06.34 | bilarh_ | it has to be part of redhat es4 with an unmodified kernel, so no reiserfs |
15:06.40 | bilarh_ | ext3 any good for this? |
15:06.42 | clive-h | smallish? |
15:07.12 | huw-l | define smallish 1MB? |
15:07.58 | wethrin | Reiserfs? |
15:08.08 | clive-h | I don't do reiserfs |
15:08.10 | wethrin | Oh, buh |
15:08.12 | wethrin | Didn't see that |
15:08.14 | *** join/#gllug Jasperw (n=jasperw@62.49.196.35) |
15:08.39 | huw-l | ext3 will be fine on 1.4TB, XFS might be worth a look but if you have lots of files less than 1MB performance will suffer on XFS |
15:08.58 | huw-l | reiser and JFS are the kings of filesystems with lots of small files |
15:09.57 | SlayerXP | if you accept that reise3rfs will randomly trash your entire dataset |
15:10.14 | wethrin | I've tended to treat XFS that it's good for large files |
15:10.23 | huw-l | it is |
15:10.31 | huw-l | and v.v bad for small files |
15:10.37 | wethrin | Yeah |
15:10.42 | wethrin | reiser's good for lots of small files |
15:10.44 | huw-l | where small is 10kb |
15:10.52 | morsing | Jasperw! |
15:11.01 | wethrin | Such as a newsfeed, where if it all gets trashed it doesn't matter too much |
15:13.56 | *** join/#gllug digits_ (n=henrik@82-68-33-30.dsl.in-addr.zen.co.uk) |
15:14.00 | digits_ | wtf, good timing |
15:14.12 | digits_ | yeah, it will be traffic logs of a couple of hundred megs each at a guess |
15:15.28 | antiphase | read/write performance varies significantly between file systems too |
15:15.30 | morsing | digits_! |
15:15.45 | digits_ | antiphase: yeah i know |
15:16.00 | digits_ | was there something said after clive-h asked "smallish?" :) |
15:16.08 | antiphase | XFS pisses on ext3 write performance, I find it swings the other way for reads |
15:16.09 | digits_ | that is of relevance to the discussion, that is |
15:16.26 | wethrin | 15:07 <huw-l> define smallish 1MB? |
15:16.27 | cpufreak | http://badger.ing.me.uk/digits.txt |
15:16.30 | cpufreak | ^^ |
15:16.56 | cpufreak | really you can't make a choice on file systems without benchmarking for your work load |
15:16.59 | digits_ | oh great, my nick changed too :D |
15:17.13 | digits_ | cpufreak: ah, what do you think then about this particular one |
15:17.27 | cpufreak | it depends how your traffic logs are to be written |
15:17.31 | digits_ | cpufreak: it's a network sniffer :) |
15:17.42 | digits_ | just a wireshark-log, afaik |
15:17.51 | digits_ | atleast that's what my spec says |
15:18.41 | cpufreak | oh right |
15:18.43 | cpufreak | its you henrik |
15:18.48 | bilarh | yes |
15:18.55 | cpufreak | in which case if there's going to be decent throughput, you need to worry about the disk subsystem much before you worry about filesystem |
15:18.59 | bilarh | i got conveniently disconnected just as i asked a question :) |
15:19.13 | cpufreak | in this case I think its been on raid10 in that v40z |
15:19.23 | bilarh | aha, so raid5 is probably not a good idea then :) |
15:19.33 | bilarh | damn, how much data are they shuffling through that pipe? :) |
15:19.40 | antiphase | raid5-- |
15:19.45 | wethrin | Raid5 ain't fast |
15:19.48 | bilarh | i like raid5 |
15:19.56 | bilarh | it's usually fast enough if you need lots of space |
15:20.05 | bilarh | here, lots of space == longer logs |
15:20.06 | antiphase | The only people that like RAID5 are the ones that are too poor to buy disks |
15:20.27 | wethrin | Like me! |
15:20.54 | wethrin | Or who want a little bit of redundancy, but have a decent backup strategy *ahem* |
15:20.58 | antiphase | It's not even particularly (i.e. not at all) resilient wrt multiple disk failures. KILL THE RAID5 |
15:22.07 | wethrin | Sure, not multiple failures |
15:22.12 | wethrin | hence the need for backups |
15:22.24 | wethrin | RAID10 isn't resilient if > half of your disks die |
15:23.06 | antiphase | o_O |
15:24.52 | wethrin | So it's all a balance of redundancy against capacity |
15:26.20 | antiphase | What about throughput? |
15:27.00 | spungles | I can't remember the differences |
15:27.04 | spungles | but with raid 5 |
15:27.23 | bilarh | well, you could lose a whole raid10 with just two failed disks also |
15:27.24 | spungles | you chases of loosing your volume are something like 1*n |
15:27.30 | spungles | where n is number of disks |
15:27.32 | bilarh | although a raid-5 is more failure-prone, i suppose |
15:27.33 | spungles | but with raid 1+0 |
15:27.37 | spungles | it's 1*n/2 |
15:28.02 | antiphase | It's in that order. RAID5 also performs worse the more disks you have in an array |
15:28.03 | spungles | or something |
15:28.16 | spungles | certainly for write, yes |
15:28.18 | antiphase | relatively worse, rather than absolutely I mean |
15:28.25 | spungles | you waste spindles for read two |
15:28.36 | spungles | since you need to fetch from several disks for a read |
15:29.19 | wethrin | I've got an 8-disk RAID5 array at home on a fileserver. I've not had problems with performance, but then there's not much activity going on - the rare write, and reads tend to only be one file at a time |
15:29.19 | spungles | given the cost of disk, why bother :) |
15:29.19 | spungles | .2 |
15:29.21 | wethrin | And the cost of a controller? |
15:29.31 | bilarh | spungles |
15:29.34 | wethrin | And space in the case, and number of slots to put the controller in? |
15:29.34 | spungles | same as raid 5 |
15:29.35 | wethrin | etc. etc.? |
15:29.46 | wethrin | What, with RAID10? |
15:29.52 | antiphase | It's also got to the point where you can store so much on one spindle that throughput is the limiting factor for any serious storage array |
15:29.55 | spungles | www.baarf.org I think |
15:29.58 | bilarh | spungles: i'd say that if you lose two disks, the chance that the second disk in a raid1 to die is 1/(n-1) |
15:30.23 | spungles | yeh |
15:30.36 | wethrin | Don't forget to take Sod's Law into account |
15:30.43 | spungles | yeh, but you need to lose 2 specific disks to lose your volume |
15:30.49 | spungles | rather than 2 disks in general |
15:30.54 | bilarh | yeah, true :) |
15:31.02 | bilarh | what about raid5 + hs |
15:31.06 | wethrin | So the chance that the next disk that dies is the one that takes down your whole array is proportional to the importance of the data on your array |
15:31.11 | spungles | and with raid 5, you shag the whole array on rebuild |
15:31.19 | spungles | as opposed to just the opposite disk |
15:31.20 | antiphase | Yes. The probability of the second disk losing you your array is dependent on your stripe width (i.e. the number of pairs of mirrors) |
15:31.51 | antiphase | The more you have, the lower the probability, but the lower the overall MTBF as well |
15:31.57 | bilarh | anti: well, you could also say that it's proportional to the number of disks in your array, as that's a lot easier :P |
15:32.12 | antiphase | :P |
15:32.30 | spungles | basically, raid 5 sucks |
15:32.40 | bilarh | unless you're poor :) |
15:32.44 | spungles | well, yes |
15:32.50 | spungles | but if you're poor, then most things suck |
15:33.13 | spungles | and thanks - i'll not be trusting my data to poor people |
15:33.33 | wethrin | I presume this is for a company, rather than for an individual? |
15:33.49 | bilarh | s/sum/balance/ |
15:33.57 | eyore | what's the easiest to use server health graphing package for things like diskIO, load, CPU, etc, |
15:34.07 | cpufreak | use orca |
15:34.09 | spungles | wethrin: well, I personally store no data on raid 5 |
15:34.10 | wethrin | If so, the number of disks the company is willing to buy is proprotional to how much they care about their data :) |
15:34.19 | spungles | and I wont let the SAN guy at the office give me raid 5 either |
15:34.20 | antiphase | Individuals just need to balance the fact they can't possibly back up n TB of data against the size of their wallet and whether they can be arsed to lose everything |
15:34.21 | eyore | cpufreak: does that work with remote servers? |
15:34.25 | cpufreak | eyore: yes |
15:34.29 | wethrin | I store 1.3TB of data on RAID5 |
15:34.36 | spungles | we tried using it for backups but it ended up shagging the whole array because it was too slow |
15:34.37 | eyore | ta |
15:34.41 | spungles | filled the write cache |
15:35.19 | wethrin | But it's mine own, rather than anything critical |
15:35.19 | bilarh | i used to run a software raid-5 array at a previous place |
15:35.19 | bilarh | it was 1.4tb :) |
15:35.19 | bilarh | and very, very slow :) |
15:35.19 | bilarh | avg write speed of 300kb/s :P |
15:35.19 | bilarh | disksuite that was |
15:35.21 | spungles | note, we have a fairly large emc |
15:35.23 | wethrin | Ow! |
15:35.23 | hali | i have an old ide raid card in a box somewhere, free as in beer :) |
15:35.26 | bilarh | however, hardware raid5 is a different ball game |
15:35.28 | spungles | and a single oracle backup would take down the array |
15:35.30 | hali | promise, 4 channel |
15:35.39 | wethrin | Mine's faster than that. |
15:35.51 | wethrin | But given I rarely write to it, slow write speeds don't bother me |
15:35.54 | bilarh | this was on a v240 |
15:36.06 | bilarh | and a normal storedge 3000 |
15:36.16 | wethrin | that's something wrong there |
15:36.42 | bilarh | mind you, with raid10 it flew :) |
15:36.47 | bilarh | i blame disksuite |
15:36.54 | hali | disksuite is the biatch here yes |
15:37.15 | hali | raid5 in svm has always been dog slow |
15:37.16 | cpufreak | software raid5 on a sparc powered box is never clever. |
15:37.27 | bilarh | but it is BIG :D |
15:37.33 | hali | raidz on the other hand... speedy! |
15:37.35 | wethrin | isn't raid5 just some magic XOR function? |
15:37.37 | cpufreak | as its not what the processor is optimised for |
15:37.37 | bilarh | at the same place I also had a 4Tb raid-0 stripe :P |
15:37.39 | wethrin | Mmm....raidz |
15:38.10 | bilarh | sparc's seem to suck at everything though |
15:38.26 | cpufreak | ! |
15:38.43 | spungles | ooer, new ipods |
15:38.45 | bilarh | seriously, sun on intel is faster than sun on solaris :) |
15:38.51 | wethrin | Err |
15:38.54 | cpufreak | bilarh: in answer to your original question - ask ISSecurity what they require performance wise, and how much space they expect each days logs to take, and how long they want to keep stuff for |
15:39.00 | cpufreak | and draw a conclusion up from there |
15:39.01 | wethrin | DYM Solaris/Intel, rather than Solaris/Sparc? |
15:39.18 | bilarh | wethrin: yes, atleast it feels that way :) |
15:39.20 | cpufreak | bilarh: depends what you're trying to do |
15:39.34 | bilarh | possibly |
15:39.42 | cpufreak | for I/O generation, and massive parallel processing, sparc scales very elegantly |
15:39.48 | bilarh | i might be a bit biased, but i've never seen a sun perform well for the money |
15:39.56 | wethrin | Have you tried a 32-CPU Sparc? :) |
15:39.57 | cpufreak | for HPC, x86/x86_64 is better suited |
15:40.12 | bilarh | wethrin: i used to have a 48 way dual core 6900 :P |
15:40.13 | wethrin | cpufreak: By HPC, you mean non-shared-memory? |
15:40.19 | wethrin | bah :-P |
15:40.35 | cpufreak | bilarh: our E6900s here perform nicely |
15:40.45 | bilarh | cpufreak: hrm, what do they actually do? |
15:40.48 | bilarh | run oracle? |
15:40.52 | cpufreak | uh huh |
15:41.00 | bilarh | how big's the db out of interest? |
15:41.00 | cpufreak | They're UltraSparc IV+ though |
15:41.04 | cpufreak | not IVs |
15:41.13 | cpufreak | lots of improvements in the IV+ |
15:41.14 | hali | the "+" adds a lot of ummpf |
15:41.31 | hali | 1.6 and 1.9's are pretty fast |
15:41.43 | eyore | Whatever happened to cell |
15:41.44 | hali | we've got a domain in a e25k in one of our customer sites |
15:41.50 | hali | it's pretty fast |
15:41.55 | bilarh | the thing is as well, the 6900's we had used to randomly stop and sit around at the OK prompt :) |
15:41.56 | cpufreak | latency on e25ks is too high |
15:42.11 | eyore | what kind of latency? |
15:42.19 | hali | eyore: between the cpu boards through the fireplane |
15:42.31 | hali | it's acceptable.. but not great |
15:42.39 | hali | but it's no hypertransport |
15:42.49 | bilarh | bring back the mainframe |
15:42.54 | eyore | well quite |
15:42.55 | cpufreak | its not good for realtime OLTP |
15:43.08 | bilarh | i want a os390 box running linux in a vm |
15:43.10 | spungles | 10G clusters on opterons |
15:43.12 | spungles | thanks bye |
15:43.12 | bilarh | just because :) |
15:43.26 | cpufreak | spungles: 10G doesn't work withour workload. |
15:43.27 | hali | oracle clusters do limp a bit if the database is too big |
15:43.37 | hali | if we are talking rac |
15:43.51 | hali | it's better to have a uniformed memory access system for that |
15:44.15 | hali | like a (i hope) sun x4800 :) |
15:44.32 | hali | the rumourmill talks about a board based opteron box from sun |
15:44.45 | cpufreak | interesting |
15:44.49 | hali | each board can take two of the cpu boards the x4600 can take |
15:45.09 | hali | the distance between the first and last board will be quite long though.... |
15:45.15 | hali | i would guess they cross mount it |
15:45.21 | hali | like a e4500 |
15:45.26 | hali | with both front and back boards |
15:45.46 | eyore | Do you reckon gradually sun will move from sparc to x86_64? |
15:46.00 | wethrin | No, I think they'll have a mix |
15:46.01 | hali | not yet |
15:46.09 | eyore | not yet, for sure! |
15:46.10 | bilarh | it must cost them a bomb to maintain the sparc processor set |
15:46.18 | hali | looking at rock and other new sparc lines it looks quite promising |
15:46.20 | bilarh | there's not enough sparc's in comparison to intels about |
15:46.30 | cpufreak | I think their partnership with Intel will be interesting, especially if they can get help from intel in bringing products to market on time |
15:46.34 | eyore | bilarh: my thoughts |
15:46.35 | bilarh | so it must cost a lot more r&d in comparison for a sparc vs an intel |
15:46.36 | cpufreak | as thats really intels strongestpoint |
15:46.53 | cpufreak | bilarh: hence why they opensourced the spec |
15:47.06 | cpufreak | and partnered with fujistu |
15:47.13 | hali | sun doesn't maintaine the line themselves, they do it with fuji.. |
15:47.14 | hali | yes |
15:47.18 | cpufreak | its curious what they're doing though |
15:47.42 | bilarh | but still, there's not that many sparcs knocking about, is there? |
15:47.53 | cpufreak | as the Advanced product line (APL) which is due this year, is appearing to be just a stop gap until Rock comes along |
15:48.01 | hali | we have a 95% sparc deployment base with our customers |
15:48.03 | bilarh | i mean if you compare total sales of sparcs vs total sales of x86, there must be like a factor 10 (at least) more x86's sold |
15:48.12 | hali | we have one customer running wintel, all others are sparc |
15:48.22 | cpufreak | although sun are now trying to backtrack on there, and claim that APL is still teh daddy |
15:48.43 | hali | just the niagra line sold machine worth more than 100 million dollors though, so even a new sparc platform can get some real attention |
15:48.45 | eyore | bilarh: sure, but isn't th real question how quickly intel will move towards high end features. |
15:49.03 | bilarh | eyore: high end features like what? |
15:49.07 | cpufreak | bilarh: IA64 has hurt intel badly |
15:49.15 | bilarh | how about opteron though? :) |
15:49.20 | hali | the 7000-series xeons do support 32-socket setups on paper, would be intersting to see if anyone will build anything with more than 4 sockets |
15:49.41 | cpufreak | if only the details of the sun/intel partnership were more transparent |
15:49.47 | bilarh | i suppose the problem is that you cant have 192 or whatever intel/amd processors on one mb |
15:49.56 | bilarh | or in one computer, i mean |
15:49.57 | eyore | bilarh: can't you hotswap out nearly every part of a Sparc box with some kind of automatic failure detection and don't they scale much better. |
15:50.20 | bilarh | eyore: hrm, but why does it scale better? is it because of the processor or because of the bits around it? |
15:50.24 | bilarh | eyore: and indeed, does it scale better? |
15:50.45 | eyore | Because intels core market didn't need 64way processors |
15:50.49 | cpufreak | well with opteron scaling beyond 8 physical sockets is a fucking nightmare |
15:51.05 | cpufreak | as you only have 3 HT paths from each socket |
15:51.15 | eyore | that does sound a mess |
15:51.19 | bilarh | hmm ok then :) |
15:51.30 | cpufreak | so unless you want to risk high latency (ie more than 3 hops) you're screwed |
15:51.40 | eyore | but gradually i guess intel/AMD will move towards what is currently the high end Sparc market |
15:51.47 | bilarh | ok, so that's high end stuff |
15:51.49 | bilarh | what about mid-range? |
15:51.57 | cpufreak | eyore: intel have been trying to do that |
15:52.00 | eyore | depends how you define midrange |
15:52.02 | hali | 8 socket is midrange enough these days, where we have the sun x4600 |
15:52.07 | cpufreak | which is why they developed IA64... about 10 years ago |
15:52.18 | cpufreak | and its sucked, and hurt them very badly, as its pissed millions (billions away) |
15:52.23 | hali | and hp used to have the dl700's |
15:52.41 | eyore | cpufreak: sure, but their mainstream processors will gradually move closer to sparc capabilities |
15:52.52 | eyore | maybe they won't make another itanic. |
15:53.19 | cpufreak | this is why the sun/intel partnership is somewhat interesting |
15:54.02 | cpufreak | as I suspect there's some kind of deal behind the scenese where sun might help intel engineer the IA-64 so it scales better, and intel might help sun improve their processes, and development techniques so they get products to market quicker |
15:54.09 | cpufreak | mostly so they can both steal marketshare from power. |
15:54.40 | eyore | i would guess sun is thinking of gradually moving away from sparc over the course of several years. but that is a guess |
15:54.54 | eyore | i can see that |
16:05.44 | *** join/#gllug robo_ro (n=robo_ro@82-70-146-30.dsl.in-addr.zen.co.uk) |
16:06.05 | *** join/#gllug Cope (n=nelsonst@88.212.149.148) |
16:06.11 | Cope | lo |
16:06.24 | morsing | Cope! |
16:06.32 | Cope | hello |
16:07.34 | *** join/#gllug Cope (n=nelsonst@88.212.149.148) |
16:08.01 | morsing | Cope! |
16:08.05 | Cope | hehe |
16:08.07 | Cope | hello again |
16:18.53 | wethrin | Cope! |
16:26.18 | Cope | hello wethrin |
16:26.37 | wethrin | We should meet up sometime! |
16:26.58 | Cope | yes indeed!! |
16:27.09 | Cope | that would ro><or |
16:27.21 | wethrin | You ever around Londonistan? |
16:27.32 | Cope | I may be :) |
16:27.50 | wethrin | aha! |
16:28.04 | Cope | are you not in oxbridge-reject land any more? |
16:29.01 | wethrin | I am, still |
16:29.12 | wethrin | But Londonistan is easier to get to than out in the sticks |
16:51.25 | keithlard | ooh |
16:52.08 | wethrin | keef! In your quest to buy things from local shops for local people, how do you deal with the situation that these places are often not open outside standard office hours? |
16:53.10 | keithlard | well i am quite lucky because we don't work wednesdays, so i can go to some shops then! |
16:53.43 | keithlard | and there is saturdays of course |
16:54.03 | keithlard | and there is the little organic butcher in kentish town that i can call in on the way to work, and put stuff in the fridge to take home later |
16:54.28 | keithlard | and the organic butchers in mill hill does free local delivery! |
16:54.52 | keithlard | and there are some places where you can order online and have it delivered, or they send you a basket of meat and veg every week or something |
16:55.31 | keithlard | i am having to think further ahead though as i can't just pop in tescos on the way home after work to get something for dinner :) |
16:55.36 | goibhniu | abel & cole ftw! |
16:56.53 | wethrin | Mm.....I suppose if you go in to work after 9am |
16:57.17 | keithlard | no question, it's less convenient if you avoid supermarkets... that is precisely what they were invented for |
16:58.14 | wethrin | well, yes, and why they gained in popularity |
16:58.23 | keithlard | damn their varied, well-stocked, high-quality, convenient, late-opening, all-in-one shoppability! |
16:59.11 | wethrin | Quite. I suppose it's also easier in London than out here in the sticks |
16:59.34 | keithlard | still, the more you use your local shops, the more of them there will be :> |
17:00.19 | keithlard | i am quite pleased as i asked in my little shop if they could stock my favourite beer, and they did! so that is reely nice. unfortunately, it's caught on in popularity, and now they keep running out as everyone else buys it :O |
17:00.30 | wethrin | There aren't many, sadly. There's a fruitier, which closes before I get home |
17:00.40 | wethrin | Time to switch favourite beers! |
17:02.48 | keithlard | never! |
17:03.09 | keithlard | is it fruitier or fruiterer? |
17:03.13 | keithlard | fruiterer sounds weird |
17:03.38 | wethrin | Fruitier |
17:03.44 | keithlard | that sounds french |
17:03.52 | wethrin | Can't have that! |
17:03.56 | wethrin | Fruit&veg shop |
17:04.03 | keithlard | do they deliver at all |
17:04.06 | wethrin | Come to Belgian, and drink lots of Leffe, then! |
17:04.27 | wethrin | Not sure. I suppose I could buy stuff in the morning from there, and nip back home, then go to department |
17:05.09 | keithlard | what about a meatier? |
17:05.37 | wethrin | Butcher in the market, but that involves quite a bit of time-out |
17:05.50 | wethrin | So that's closed when I come home, too |
17:05.54 | keithlard | baah |
17:05.58 | wethrin | indeed |
17:06.21 | wethrin | I wonder if the local shops would get more custom if they were open, say, 3pm to 8pm |
17:07.35 | Leeds | oh, to live in a later-night city |
17:08.15 | wethrin | shush, you :-P |
17:13.06 | keithlard | yeah i'm sure they would |
17:25.24 | wethrin | Mrumph. Need to print stuff off about whisky. |
17:34.04 | Cope | that's sillu |
17:34.16 | Cope | i think you should invest time drinking it |
17:35.01 | wethrin | I've just found out I need to lead a tasting |
17:35.06 | wethrin | By just, I mean, a few hours ago |
17:36.33 | Cope | oh |
17:37.08 | wethrin | Friend who was supposed to be doing it had to stand in on rugby duty |
17:38.46 | bilarh | ow ow ow, my forex bet is going down the pan |
17:39.27 | antiphase | That'll teach you |
17:40.56 | wethrin | z00! |
17:41.51 | *** join/#gllug Cope_ (n=nelsonst@88.212.149.148) |
17:41.57 | Cope_ | woo z00dax |
17:42.15 | Leeds | woo Cope_ |
17:42.20 | Cope_ | hello Leeds |
17:49.17 | z00dax | Cope: howse it going then |
17:49.32 | Cope | oh ok, thanks! |
18:00.40 | *** join/#gllug Cope_ (n=nelsonst@88.212.149.148) |
18:29.52 | *** part/#gllug goibhniu (n=cillian@grail1.oxfam.org.uk) |
18:40.00 | *** join/#gllug Puss1nB00ts (i=PussInBo@gateway/tor/x-a9f4cd961b9fc2f2) |
18:47.32 | *** join/#gllug ldnblk (n=Just@82-35-34-220.cable.ubr01.camd.blueyonder.co.uk) |
18:50.08 | ldnblk | hello there leeds |
19:55.03 | morsing | ping wethrin |
19:56.14 | z00dax | ping morsing |
19:58.16 | morsing | Pong |
20:00.21 | z00dax | ping morsing |
20:00.51 | morsing | Anyone here good at setting commas? |
20:01.12 | z00dax | ,,,,,,,,,,,,,, |
20:01.18 | z00dax | there you go, you can use mine |
20:03.05 | boudiccas | Anyone here, good, at setting commas |
20:04.16 | z00dax | all hail the New York Giants! |
20:24.10 | morsing | So no-one sane or useful in here tonight then? |
20:31.52 | boudiccas | define sane, or useful? |
20:31.56 | boudiccas | beer! |
21:05.49 | *** join/#gllug catalyst (n=catalyst@allegro/user/angelchild) |
21:17.06 | *** join/#gllug goibhniu (n=cillian@87-194-36-120.bethere.co.uk) |
21:27.38 | gw280 | lalalalalalala |
21:27.43 | gw280 | differential equations rock my world |
21:28.44 | clive-h | I'm saneish |
21:58.16 | morsing | gw280! |
22:00.02 | gw280 | morsing: piss off |
22:00.12 | gw280 | morsing: or I'll differentiate you twice and then shove you into a differential equation |
22:03.45 | *** join/#gllug sabinef72 (i=sabinef7@barcelone.ipv6.popipo.fr) |
22:04.37 | morsing | sabinef72! |
22:11.17 | sabinef72 | stupid connexion |
22:11.44 | morsing | Mmm... |
22:12.01 | *** join/#gllug benoit (n=benoit@santander.ipv6.popipo.fr) |
22:12.01 | morsing | benoit! |
22:12.13 | benoit | hi morsing |
22:12.19 | benoit | had some problem with my server ... |
22:12.25 | benoit | DNS was down and reboot did not work :-( |
22:12.29 | benoit | but should be ok now |
22:42.31 | trousers_ | Yay, my Xen talk at school went down fairly well :) |
22:42.44 | trousers_ | And beamer-LaTeX looks quite nice |
22:42.44 | clive-h | yay |
22:43.08 | trousers_ | AND... I did a live 1st ever test of XP in the talk, and it installed perfectly 1st time :) |
22:43.36 | clive-h | xp under xen/ |
22:43.41 | clive-h | ? |
22:43.44 | trousers_ | Yarr |
22:44.00 | clive-h | Cool |
22:44.09 | clive-h | what box? |
22:44.31 | trousers_ | My shuttle server - X2 4200+, 1GB DDR2 RAM, 250GB HDD |
22:44.42 | trousers_ | (SK22G2) |
22:44.46 | clive-h | Ooh cool |
22:45.04 | clive-h | I want to run XP under xen here |
22:45.18 | clive-h | similar spec |
22:45.31 | trousers_ | The Xen official WIndows as guest guide worked a te |
22:45.39 | trousers_ | s/te/treat/ |
22:45.43 | clive-h | except 3800+ and 160Gb |
22:45.53 | trousers_ | SVM support? |
22:46.00 | clive-h | yahh |
22:46.20 | trousers_ | Funky |
22:46.26 | trousers_ | What's dom0? |
22:46.28 | murb | XP under Xen isn't very amazing. |
22:46.45 | murb | it is better if you get the non-free pv drivers for windows. |
22:46.55 | clive-h | it works? |
22:47.00 | trousers_ | Non-free pv? |
22:47.01 | murb | but vbox is faster on the same hardware. |
22:47.26 | trousers_ | Also --- I felt that WinXP was the best and easiest example of an unmodified x86 guest kernel |
22:47.27 | clive-h | hmm vbox? |
22:47.29 | trousers_ | vbox? |
22:47.42 | murb | trousers_: paravirtulised, otherwise it uses the qemu device model to simulate ide, vga and ethernet devices. |
22:47.46 | murb | http://www.virtualbox.org/ |
22:48.03 | clive-h | ta murb |
22:48.22 | murb | it doesn't yet support 64b host/guest |
22:48.27 | murb | but it is nearly there. |
22:48.29 | clive-h | pass |
22:48.42 | trousers_ | I need my own classroom at school, then I could dump old hardware in there, and make a giant Gentoo cluster for the club :D |
22:48.49 | clive-h | I'm still running 32bit anyway |
22:48.59 | murb | trousers_: old kit costs a lot to run. |
22:49.12 | murb | as does new kit. |
22:49.12 | trousers_ | murb: Note --- SCHOOL classroom :) |
22:49.31 | murb | trousers_: and the school gets free energy? |
22:49.39 | trousers_ | No, but I don't pay for it |
22:49.51 | trousers_ | So for me it's free |
22:50.01 | clive-h | i'm tempted to buy another box and run it 64bit |
22:50.11 | murb | tansfaafl |
22:50.33 | trousers_ | tansfaafl? |
22:50.41 | clive-h | though pentium D perhaps |
22:51.00 | clive-h | there's no such thing as a free lunch |
22:51.07 | trousers_ | Oh, but there is |
22:51.45 | trousers_ | Wait, how does that work? Surely it's tnstaafl |
22:52.09 | clive-h | I'm english |
22:52.26 | clive-h | you got what I'd say |
22:52.34 | trousers_ | Hmm |
22:53.06 | trousers_ | ibot seen mozrat |
22:53.29 | ibot | mozrat <i=mozrat@outbound.silenceisdefeat.org> was last seen on IRC in channel #gllug, 9d 22h 8m 51s ago, saying: 'I did send one out back in December'. |
22:53.42 | trousers_ | :( No mozrat |
23:09.20 | *** join/#gllug keithlard_ (n=keith@i-83-67-2-128.freedom2surf.net) |
23:18.03 | trousers_ | ibot Will there ever be another GLLUG meeting? |
23:19.32 | murb | never |
23:20.29 | keithlard_ | aw, why |
23:20.38 | murb | because i'm going to bed. |
23:25.36 | morsing | keithlard_! |
23:26.19 | keithlard_ | ello |
23:26.30 | keithlard_ | i like gllug! |