Back
[01:08:43] <ferdna> Casper, defragmentation is not requiered under linux...
[01:09:03] <Casper> it is
[01:09:12] <ferdna> no it is not.
[01:09:20] <Casper> show proof
[01:09:32] <ferdna> ok one second.
[01:09:46] <Casper> oh you can't? too bad, because I can give you proof that it's needed
[01:09:50] <Casper> like the
[01:09:56] <Casper> like the 27% fragmentation I had
[01:10:16] <Casper> and why shake got invented
[01:10:25] <ferdna> easy
[01:10:28] <Casper> (which do a bad job, but better than not use it)
[01:10:28] <ferdna> A true defragmentation tool does not exist for ext3
[01:10:30] <ferdna> read:
[01:10:35] <ferdna> http://en.wikipedia.org/wiki/Ext3
[01:10:38] <Casper> that is not a proof
[01:10:39] <ferdna> defrag section...
[01:11:01] <Casper> not said on the page: the kernel wasn't made to handle defragmentation
[01:11:13] <Casper> a complete rewrite of the ext3 code would have been needed
[01:11:22] <Casper> or do an offline defragmenter tool
[01:11:35] <Casper> which defeat the purpose
[01:11:56] <ferdna> no wrong...
[01:12:11] <Casper> one program started to be made
[01:12:21] <ferdna> linux FS are designed to not be defragged.... and if you do you will burn your discs....
[01:12:36] <Casper> but due to the idiot of linus tosvald that refused to include the patches
[01:12:48] <Casper> all the projects got forced to instinction
[01:13:05] <mrfrenzy> if the patches were needed they would be included, but they are not.'
[01:13:14] <ferdna> how can you defy the wikipedia mother of all learning resources...
[01:13:29] <Casper> due to popular pressure, one has been made for ext4, and will fix the need for the admin to take the server offline, copy all, mkfs, and copy back
[01:13:33] <Casper> mrfrenzy: they are now included
[01:13:40] <ferdna> +1 for ferdna... thank you mrfrenzy.
[01:13:55] <ferdna> casper 0;
[01:14:11] <Casper> mrfrenzy: linus always said that HE didn't needed defrag, so refused to let any defrag patch in
[01:14:30] <Casper> even when lots and lots of admin said that they needed to manually do it
[01:15:02] <Casper> it take only 1 person to deny such thing: linus, which he is the kernel maintainer
[01:15:02] <ferdna> Casper, also defrag in general will be a thing of the past... with SSD drives...
[01:15:18] <Casper> ferdna: yes and no, SSD drives ain't popular
[01:15:29] <ferdna> not yet... but hey will
[01:15:29] <Casper> and will probably never be for storage
[01:15:31] <Casper> and NEVER for server
[01:15:45] <ferdna> thats the future
[01:15:57] <Casper> not if they have such low write cycles
[01:16:07] <ferdna> how come usb portable drives are popular... same tech as in ssd...
[01:16:10] <Casper> 10k writes... don't even consider running your temp there
[01:16:15] <Casper> no
[01:16:20] <Casper> not the same purpose
[01:16:26] <Casper> you don't constantly write on usb
[01:16:42] <ferdna> cuz its a different technology
[01:16:45] <ferdna> but the same purpose
[01:16:52] <Casper> and yet, lots of people have reliability issues after a few years due to write cycles
[01:16:59] <Casper> different purpose
[01:17:08] <Casper> SSD is for constant use
[01:17:18] <Casper> while usb flash is for data transportation
[01:17:50] <Casper> until they make abordable unlimited write cycle SSD, mechanical drive will be the norm
[01:17:51] <Casper> beside
[01:17:59] <Casper> go to any computer store
[01:18:08] <Landon> I didn't realize mechanical drives were unlimited write cycle
[01:18:16] <mrfrenzy> mechanical drive will be the norm until ssd are cheaper
[01:18:25] <mrfrenzy> consumers don't care about reliability
[01:18:33] <mrfrenzy> and yes, mechanical drives are not unlimited
[01:18:42] <ferdna> mrfrenzy, consumers worry about reliability....
[01:18:44] <mrfrenzy> in the future ssd will have even better reliability
[01:18:48] <Casper> try to find any computer that come with SSD, you might, if you are very lucky, find one model only. But notice that it will most probably also have a mechanical drive too
[01:19:01] <Landon> now that my mind is wandering, what's the difference between mechanical drive cache and SSD?
[01:19:08] <mrfrenzy> 99% of consumers only worry about one thing - price
[01:19:09] <ferdna> Casper, all asus mini laptops have ssd drives..
[01:19:11] <Casper> ferdna: customers don't really care, they want cheap and fast
[01:19:20] <Casper> ferdna: not all
[01:19:28] <Casper> many have mechanical drives too
[01:19:34] <Casper> I've seen some with 1.8" hd
[01:19:43] <Casper> mrfrenzy: price and numbers
[01:20:48] <Casper> "oh this one have an amd x6 9764 and this one have an i5 2500 ... the amd must be almost 4 times faster!
[01:20:49] <Casper> "
[01:21:28] <Casper> "this one have 12G ram! and this one only 8. My internet will be faster with the 12!"
[01:21:53] <Casper> "this one have a green hard disk! it's better for the environnement"
[01:22:10] <Casper> but anyway
[01:22:24] <Casper> fragmentation does occur on ext2/3/4
[01:22:49] <Casper> linux do a pretty damn good job at AVOIDING fragmentation
[01:23:02] <Casper> it do NOT ensure that nothing will be fragmented
[01:23:26] <Casper> and contrary to what lots of idiots say: linux do not defragment in the background
[01:23:55] <Casper> however due to the frequent rewrite of the files, if it know the filesize it can find a space where to fit it
[01:24:08] <Casper> downloading files make it impossible to know, so place it in an hole
[01:24:18] <Casper> torrents are pretty damn good to fragment
[01:24:40] <Casper> so is running some applications that write many files at once
[01:24:51] <Casper> the bigger the file, the more it will fragment
[01:48:23] <ferdna> Casper, you are wrong... defrag is only for windows fags.... period.
[01:48:43] <Casper> and mac
[01:48:45] <Casper> and linux
[01:49:49] <mrfrenzy> any half decent torrent client will preallocate, so 0 fragmentation
[01:50:51] <nevyn> sparse tho right.
[01:51:50] <Casper> yes, but not enabled by default
[01:51:53] <Casper> beside
[01:52:03] <Casper> it do not fix the FTP problems
[01:52:07] <Casper> and even database issues
[01:55:56] <ferdna> Casper, the os handles the file management... so dont worry about what programs do what...
[01:55:59] <ferdna> defrag is not needed
[01:56:15] <Casper> wrong
[01:56:34] <dehuman> to overcome the speed limitations of the disks it is
[01:56:50] <dehuman> if its not a normal disc then it doesnt matter
[01:56:56] <dehuman> if it moves then it matters
[01:57:34] <Casper> tell that to the admins that need to periodically move out, format and move back the data to increase the server disk performance almost 10 folds
[01:57:45] <dehuman> the enhancements in other filesystems are more for failure, what to do when you suddenly turn it off, or otherwise searching and indexing metadata
[01:57:56] <dehuman> with some nod to efficency and placement of files
[01:57:58] <dehuman> but thats not its jb
[01:58:02] <Casper> the fuller the hd, the worse it get
[01:58:37] <dehuman> so yah if you got spindles you can defrag and see a benefit
[01:58:44] <dehuman> regardless of filesystem
[01:58:58] <dehuman> with use case of lots of dynamically allocated files
[01:59:08] <dehuman> ala torrents, etc
[01:59:38] <Casper> torrents, decompressing/compressing several files at once...
[01:59:44] <Casper> or simply download a few files at once
[01:59:54] <dehuman> now with ssd and flash it doesn't matter
[01:59:58] <dehuman> cause read time is same
[02:00:03] <Casper> to sum up, when you write several files at once, you get fragmentation
[02:00:08] <dehuman> there is no seek time
[02:00:14] <Casper> dehuman: actually, this is not fully true
[02:00:33] <Casper> due to the read ahead feature of the disk, controller and os, fragmentation does matter on ssd
[02:00:39] <Casper> however it's way less important
[02:00:46] <Casper> and almost negligible
[02:01:20] <Casper> SSD and controller both do not know about the filesystem and will read more sectors than requested
[02:01:44] <Casper> the os however might do file read ahead (but it's unclear how the os actually do it)
[02:01:55] <Casper> I read that on linux it read more blocks on the disk
[02:02:05] <Casper> and that's all what I could find of reliable info
[02:02:26] <Casper> then I found some info saying it's disk blocks, not files block
[02:02:43] <Casper> so if there is fragmentation, you still lose time since it read useless data
[02:04:19] <ferdna> Casper,
http://geekblog.oneandoneis2.org/index.php/2006/08/17/why_doesn_t_linux_need_defragmenting
[02:04:31] <ferdna> they explain it very well in that document
[02:04:34] <ferdna> read it..
[02:05:42] <Casper> let's see the wrong resoning there
[02:06:55] <Casper> ferdna: I can see many wrong assumption there
[02:07:02] <Casper> first, that the hd is empty
[02:07:14] <Casper> second, that there is a block big enought
[02:07:20] <Casper> third, that the file will not grow
[02:07:37] <Casper> forth, that you do not constantly write and delete
[02:08:14] * ferdna kicks Casper in the butt.
[02:08:17] <Casper> fifth, trying to prove a point based on rumors and misinformations
[02:08:25] <Landon> what, did you actually read it?
[02:08:38] <Landon> I barely skimmed the first bit and there's an example of a file growing
[02:08:46] <Casper> skimmed throught it, and found exactly what I expected to found
[02:08:53] <ferdna> lol
[02:08:59] <Landon> HOWEVER
[02:09:00] <Casper> file growing by almost nothing
[02:09:02] <Landon> Fragmentation thus only becomes an issue on ths latter type of system when a disk is so full that there just aren't any gaps a large file can be put into without splitting it up. So long as the disk is less than about 80% full, this is unlikely to happen.
[02:09:22] <Casper> unlikelly to happend WITH SMALL FILES
[02:09:30] <Casper> and this is where things get dirty
[02:09:52] <Casper> then you have this assumption: big files don't need to be read fast
[02:10:13] <Casper> based on that assumption, they do not need to be defragmented at all...
[02:10:15] <Landon> what's your application that needs the big files to be read fast? just curious
[02:10:29] <Casper> ... I had some file that just couln't be played in mplayer
[02:10:32] <Landon> only time I need it to be fast is copying back and forth over my lan
[02:10:45] <Casper> the fragmentation was just that bad
[02:11:15] <Casper> (hd near full, and torrents... with a stupidly small chunk size)
[02:11:32] <Landon> why not preallocate space if you know the torrents are going to be an issue?
[02:12:03] <mrfrenzy> indeed, preallocation solves the problem
[02:12:14] <Landon> sure it may not be default, but that's what planning ahead solves
[02:12:48] <Casper> Landon: big torrent take forever to preallocate, beside, at that time, there was a massive problem with cpu usage for the kernel
[02:13:00] <Casper> basically, if your hd was 100% used, you couln't move the mouse
[02:13:16] <Landon> so, that's another problem solved by planning ahead
[02:13:18] <Casper> now they fixed that issue
[02:13:34] <ferdna> Casper has a 20TB server with an LSI card...
[02:13:47] <ferdna> so space is not an issue for him
[02:14:14] <Landon> ooh LSI, I knew I forgot to apply somewhere
[02:14:23] <Casper> ferdna: 15, not 20
[02:14:43] <Casper> but my previous raid was full
[02:14:47] <ferdna> lol... still too much...
[02:14:48] <Casper> and so was the one before
[02:15:10] <Casper> still, the fuller the hd, the more it will fragment
[02:15:46] <Landon> ack, I can't trust my data to raids anymore, I've had too many software raids fall off the map
[02:16:01] * Casper wonders if there is a ntfs defrag for linux..
[02:16:16] <mrfrenzy> whether they fall off the map doesn't depend on software or hardware raid
[02:16:18] <Casper> Landon: I don't have software raid issues, but hardware... ew...
[02:16:21] <mrfrenzy> rather using too little redundancy
[02:16:33] <Casper> mrfrenzy: or simultanious failure
[02:16:42] <Casper> like using those power Y cables....
[02:16:49] <mrfrenzy> IE too little redundancy
[02:16:54] <ferdna> guys that last article i provided the link for.. its awesome...
[02:17:00] <ferdna> it actually explains in detail
[02:17:04] <mrfrenzy> if you have enough redundancy simultaneous failur is no problem
[02:17:08] <mrfrenzy> I only use raid6 nowadays
[02:18:33] <Casper> ferdna: it do not explain anything, it over simplify everything, which is completly wrong in this case
[02:19:38] <mrfrenzy> this is almost like watching a discussion between two priests of different religions
[02:21:20] <Casper> that booster really look good
[02:22:08] <Landon> mrfrenzy: it wasn't redundancy problems, it was that some configuration files for the software raid disappeared
[02:22:22] <Landon> and as I understood it at the time, that meant the data is locked away never to be seen again without black magic
[02:23:06] <Casper> talking about it, I wonder if I saved the config somewhere...
[02:23:30] <Landon> right now I just have a delayed raid1 ;)
[02:24:03] <ferdna> guys i am going to bed... can you guys save the log for me so i can read it in the morning...
[02:24:05] <mrfrenzy> well that understanding is wrong
[02:24:05] <Casper> I once had a 4 disks raid 1 :D
[02:24:11] <ferdna> good night...
[02:24:13] <mrfrenzy> you don't need any more than the disks themselves
[02:25:59] <Casper> it was when I had my root fs on a raid
[02:26:10] <Casper> so I had partitionned my 4 disks the same way
[02:26:28] <Casper> since it was useless to have a bigger partition for some disks
[02:26:35] <Casper> I just did the same layout on all disks
[02:26:48] <Casper> so that 4 raid1 was really like 100MB, the /boot
[02:27:13] <Casper> bed time
[02:27:13] <Casper> nite
[02:27:18] <mrfrenzy> n8
[02:27:25] <mrfrenzy> (morning here)
[02:55:17] <ome> Casper: Do you reckon Struct and struct container would be a better option then class ? (though they are not that diff. IMHO).
[15:52:31] <chupas> GERRRR is there a place to download avr studio without having to enter all my info AGAIN
[15:57:58] <kline> chupas: what version do you want?
[15:58:05] <kline> or did you go ahead?
[15:58:05] <chupas> whatever latest is
[15:58:16] <kline> one sec, let me see if it still have the installer
[15:58:51] <chupas> there must be a reason why they make people register every time. I cant imagine they havent had 100000 complaints about it so far
[15:59:01] <kline> marketing $$$
[15:59:12] <chupas> how is it makeing money?
[15:59:21] <chupas> maybe I dont understand marketing
[16:00:00] <kline> they probably stick it in there "people we can potentially mail about stuff they can buy", or maybe even sell it on. dont know, check the ToS and priv. policies. AS5 installer = 616MB. how do you want it sent?
[16:00:11] <Martyn> I just found a major bug in the SAM3U .. interrupts don't fire correctly
[16:00:13] <Martyn> damnit
[16:00:24] <OndraSter> duh
[16:00:25] <OndraSter> sounds serious
[16:00:28] <Martyn> *shakes fist at Atmel*
[16:00:30] <OndraSter> what do you mean correctly?
[16:00:45] <Martyn> I mean it's missing interrupts, timers are firing at the wrong time
[16:00:50] <chupas> Eh, Ive never gotten anything in the mail from atmel ever. I first registered 4 years ago
[16:01:01] <kline> nor me. *shrugs*
[16:01:01] <OndraSter> are you sure you haven't set up something wrong, Martyn ?
[16:01:06] <Martyn> which means moving my code from AVR to SAM3U isn't going to work
[16:01:14] <OndraSter> I am pretty sure that they would notice that :
[16:01:15] <OndraSter> :)
[16:01:19] <Martyn> Yep. Just had an Atmel FAE spend three hours with me verifying the bug.
[16:01:25] <OndraSter> oh ok
[16:01:33] <chupas> Ive enteres my information correctly once, and 20 times just bogus info. I guess its their problem
[16:01:39] <Martyn> She even took a board back to Atmel's HQ.. it's a pretty serious issue
[16:01:49] <kline> chupas: 616MB. how do you want it sent? or what?
[16:01:52] <Martyn> maybe one of the reasons why Arduino is taking so long with the Due
[16:02:10] <chupas> kline do you just have the file name? I can probably find a mirror with that
[16:02:24] <kline> as5installer-5.0.1163-full.exe
[16:02:52] <chupas> Cool, thanks!
[16:03:13] <kline> may be a newer version, but thats the one i have
[16:03:26] <Tom_itx> 5.1 is out
[16:03:31] <chupas> Oh yea
[16:03:34] <chupas> 5.1 at least
[16:03:46] <kline> well, back to the sign up :P
[16:03:47] <chupas> ugh, guess ill "register"
[16:04:07] <Tom_itx> dean gave us a way around it but i didn't mark it
[16:04:21] <chupas> betaware used to be what I used
[16:04:26] <chupas> but they stoped updateing after 4.18
[16:04:28] <chupas> or someting
[16:05:37] <chupas> http://www.avrfreaks.net/index.php?name=PNphpBB2&file=viewtopic&theme=print&p=938456
[16:05:45] <Tom_itx> i have 5.1 exe
[16:06:16] <kline> i think its time to nuke my entire toolchain and reinstall fresh
[16:06:22] <chupas> wheres his greasemonkey script!
[16:06:31] <Tom_itx> in the logs
[16:07:14] <chupas> huh/
[16:08:05] <Tom_itx> http://pastebin.com/8RGCFU5s
[16:09:59] <chupas> AHh TY
[16:10:47] <chupas> speak of the devil
[16:16:58] <chupas> haha that script works lovley!
[16:17:07] <chupas> haazaaaaa
[17:57:21] <OndraSter> I've been always wondering, when they throw some electric device into water, how come that it kills people in the water?
[17:57:32] <OndraSter> shouldn't the loop close through the shortest path? :P
[17:57:37] <OndraSter> kills = hurts*
[17:57:40] <OndraSter> and in movies it is
[17:57:55] <OndraSter> and circuit breakers/fuses work really fast
[18:01:05] <Casper> there is no such thing as shortest path
[18:01:24] <Casper> it's a matter of proportional resistance
[18:02:08] <Casper> it does make a big zone for the kill zone
[18:02:55] <Casper> do the test with a 12VAC transformer and ONE HAND... not both... you'll feel it...
[18:05:16] <Landon> OndraSter: I seem to remember a mythbusters on that, toaster in a tub or something similar
[18:06:11] <OndraSter> hmm
[18:06:20] <OndraSter> yes, I know that it is proportional resistance
[18:06:34] <OndraSter> but the wire has to melt with high currents when it is dropped into water
[18:06:39] <OndraSter> Landon, don't you happen to know the ep number? :)
[18:07:10] <Casper> water isn't very conductive
[18:07:18] <Casper> so the wire won't melt
[18:07:19] <Casper> but
[18:07:24] <Tom_itx> distilled water isn't
[18:07:31] <Casper> there is some exposed contact that are
[18:07:32] <Tom_itx> high mineral water certainly is
[18:07:41] <Casper> Tom_itx: even distilled water is slightly conductive
[18:08:03] <Casper> due to the natural presence of H+ and O- ions (I think I got the right polarity, not sure)
[18:09:23] <OndraSter> that means that even with 230VAC
[18:09:30] <Landon> OndraSter: nope, I just have a vague memory of the beloved ballistics gel and some current monitors embedded
[18:09:41] <OndraSter> I am not sure that if you kick into pool some radio
[18:09:47] <OndraSter> it will kill you
[18:10:16] <Landon> http://mythbustersresults.com/episode19
[18:10:22] <OndraSter> thanks :)
[18:15:47] <vectory> "If killer quicksand is busted, please explain to me how bear grills from man vs wild was just in some up to his neck and could barely move. he only escaped due to his knowledge from survival experts.
[18:15:52] <vectory> roflma
[18:15:54] <vectory> o
[18:17:31] <Landon> vectory: oh god, you're reading a comments section on the internet?
[18:17:53] <vectory> i was looking for context
[18:20:16] <Casper> there is also several kind of quicksand
[18:20:28] <Casper> some are filled with air, and you WILL drown in it
[18:21:00] <Casper> but man vs wild... what a joke
[18:21:11] <Casper> all scripted
[18:22:23] <Landon> Casper: have you seen the behind the scenes videos for man vs wild?
[18:22:53] * Landon would only call it scripted in the loosest sense
[18:23:15] <Landon> perhaps not the behind the scenes, but the camera crew videos*
[18:23:23] <vectory> Landon: _all_ scripted, even the behind the scenes material
[18:23:41] <vectory> w/e
[18:23:42] <Landon> that's the first I've heard of it
[18:26:06] <Landon> well, further research on the internet says it's so, guess I'd better take it at face value