IRC Logs for #crux Friday, 2007-05-25

*** pitillo has joined #crux00:07
*** jaeger has quit IRC00:23
*** j^2 has quit IRC00:51
*** j^2 has joined #crux00:55
pitillogood morning01:02
vektoriGood morning spooky moonbeam.01:03
*** namenlos has joined #crux01:04
pitillo:)01:04
*** sepen has quit IRC01:21
*** mike_k has joined #crux02:14
*** sepen has joined #crux02:20
sepenmorning all02:20
mike_kmorning02:21
rxinipuL: roger that02:38
*** RedShift has quit IRC03:39
morlenxusDoes someone has some experience with async and sync mounts and know if there is some speedup if i would use async mounts?03:48
morlenxusI also would like to know how long max it will take until the data is written.03:49
vektoriasync is the default.03:52
vektoriThat's what you already have.03:52
vektoriHopefully, because sync is damn slow.03:53
prologicanyone know of any good radius web interfaces ?03:58
morlenxusvektori: No i have sync because of data security.04:02
morlenxusI don't know how long it will take until the data is written to disk.04:03
vektoriFor database operation, sync mounts are painfully slow.04:05
morlenxusoh04:05
morlenxusYes but if there is some data write is cached, how long does it take max to write it really?04:06
morlenxusI mean i can't accept holding data in memory for about 5 minutes without writing them down.04:07
treachafaik, nobody knows how long stuff will be held if you use async.04:09
predatorfreaktreach: Can't you force the kernel to flush to disc?04:10
treachsure, but that's a bit tedious, isn't it? ;)04:11
predatorfreaktreach: Hey, in the cases where you want data security it's nice.04:11
predatorfreakAlthough to be honest.04:11
predatorfreakThe best solution is just have a UPS.04:11
treachalso, that sort of negates the purpose of async writes.04:11
predatorfreaktreach: I'm not saying do it every write ;)04:12
predatorfreakHeck, I don't even think it's really smart myself.04:12
predatorfreak(might even be down-right stupid with things like XFS)04:13
morlenxusSure ups is one part of data security. But i can't hold data in memory without written to disk for a long time, for example the server crashes and a hugh bunch of data is lost.04:13
predatorfreakmorlenxus: I suppose you should just use sync then.04:14
predatorfreakFlushing it to disc every couple of minutes or whatever would probably just be idiotic.04:14
predatorfreakmorlenxus: Really though, I just trust that my system doesn't suck.04:17
predatorfreakAfter all, I've set everything up.04:17
predatorfreakSo if it sucks, I've done something horribly wrong :)04:17
morlenxusWell what if there is a hardware error?04:18
morlenxusYou can't control everything.04:18
morlenxusSo the data should be written in time to disk.04:18
predatorfreakmorlenxus: If my hardware is going fuckity.04:18
predatorfreakThen in all likeliness.04:18
predatorfreakIt won't matter on PC-class hardware.04:18
treachI seem to remember there is a limit of 256 pending write operations until a flush is forced.04:18
treachor something like that.04:18
morlenxusah04:18
predatorfreakSince it's so crap at dealing with things like that.04:18
treachBut I'm not 100% sure, and it might only apply to nfs..04:19
predatorfreakI've heard that supposedly when you loose power on a standard PC voltages random fuck around all over the board.04:19
predatorfreakfor like 1-2 seconds.04:19
morlenxusoO04:19
predatorfreakmorlenxus: Plus, sudden failure is impossible to be fully safe from.04:20
predatorfreakIt could die during writing for example.04:20
predatorfreakand leave you with half-written data.04:21
predatorfreakmorlenxus: Basically, like you said, I can't control everything.04:22
predatorfreakThat's why I do what I can and whatever happens happens.04:23
treachpresumeably it also depends on what scheduler you're using..04:23
morlenxusDoes sync also means that data is read everytime from disk while async means that data is stored in mem and only read from disk on file change?04:23
vektorimorlenxus: The Right Way is to use transactions in the database. A well designed DBMS flushes the data to disk when you commit the transaction.04:24
treachno, async means "write when convinent" iirc.04:24
morlenxusSo async/sync does nothing have to do with the read operation?04:25
vektoriNo.04:25
morlenxusk04:25
vektoriAlso, just to make you feel more insecure: even if you use 'sync' mounts, the data is almost always stored in the hard drive's write-back cache, so you can't be sure if you're safe. :)04:26
* treach learns to spell *convenient*04:26
treachheh, you've got to disable that in bios. :)04:27
treachand then you'll have REALLY sucky performance. :D04:27
morlenxustreach: No, i use a raid with a battery backup.04:28
morlenxusSo if the power is gone, the data still can be written to disk.04:28
vektoritreach: I doubt that'll help.04:28
treachvektori: I was thinking of some older really broken ide controllers that caused serious corruption unless you took that step.04:28
vektoriYep. But hard drives which say "yeah yeah, I flushed the stuff to disk" although they don't really flus are quite common.04:30
morlenxustreach: You seem to know a bit of nfs right? Do you know if it makes sense to define the wsize and rsize when using nfs3 over tcp? I readed that nfs3 server and client set that using handshake.04:30
treachmorlenxus: I don't no nearly as much as I'd like to, and I've never used NFS over tcp. Sorry.04:31
treachs no/know04:31
morlenxushm04:33
*** mike_ has joined #crux04:34
treachmorlenxus: http://bioinformatics.org/pipermail/bioclusters/2005-August/002693.html04:35
treach6th paragraph04:35
morlenxusAh interesting.04:38
morlenxusMaybe i can find another source confirming that.04:38
morlenxusIt's hard to find detailed inside information for stuff like that.04:39
treachmmh, a lot of things you find is old. :/04:40
morlenxusYepp, this also.04:40
*** predatorfreak has quit IRC04:43
*** namenlos has quit IRC04:46
treachhttp://nfs.sourceforge.net/#section_b C3.. yikes. Didn't know that..04:48
*** mike_k has quit IRC04:52
morlenxustreach: Good link!05:02
morlenxusAnd it seems that rsize and wsize are really unused on nfs tcp.05:02
*** roliveira has joined #crux05:14
treach:)05:34
morlenxustreach: B6 says that NFS3 async is not needed and is also not a speedup.05:50
rxiasync isnt used anymore is it05:51
morlenxusHm can't say for shure.05:52
rxii never use it anyway05:52
morlenxushttp://nfs.sourceforge.net/#section_b <- b6 last part05:52
rxiyeah since noone would use v2 anymore05:55
rxii use sync05:55
morlenxusYes, but even on nfs3 it doesn't make a difference.05:56
rxii think im using v405:57
morlenxusReally?06:01
morlenxusnfs4 is imho not stable and needs special client side tools.06:01
*** RedShift has joined #crux06:02
rxidunno ive enabled it on the server/client in the kernel06:03
rxiwho knows i set it all up like 3years ago and havent needed to touch it since06:03
*** acrux has joined #crux06:10
*** thrice` has joined #crux06:54
*** cptn has joined #crux06:57
*** ChanServ sets mode: +o cptn06:57
cptnhe06:57
cptnerr hey06:57
rxihey06:59
*** Viper_ has joined #crux07:09
pitillohello cptn07:15
*** tilman_ has joined #crux07:33
*** tilman_ is now known as tilman07:33
*** ChanServ sets mode: +o tilman07:34
*** Viper_ has quit IRC07:39
tilmanmavrick61: ping? :D07:55
tilmanrugek: drhouse in contrib? :>08:01
tilmanoh, err.08:01
tilmanrugek: n/m, i'm confused08:01
rugek...08:04
rugek;)08:04
tilmanECAFFEINE08:04
*** jdolan_ has joined #crux08:06
nipuLholy butt fucking monkey cannons!! is anyone else on the openbox ml?08:09
tilmanno, what happened?08:10
nipuLsomeone set off a mail bomb08:10
tilmantheir new wiki sucks08:10
cptn"Someone set up us the bomb"08:10
tilmanapparently they don't link to the ml archive anymore08:11
j^2so my gov'r here in texas just passed a bill to seal public inquiries about consealed handguns...08:12
j^2btw, hi!08:12
nipuLi finally got a header check in place for the offender, but only after several hundred emails hit my server and brought dovecot to it's knees08:12
nipuLs/server/el cheapo vps/08:13
tilmanfun08:13
*** percent20_work has joined #crux08:14
*** jaeger has joined #crux08:19
*** ChanServ sets mode: +o jaeger08:19
*** percent20_work has quit IRC08:25
*** efika|cruxppc has joined #crux08:36
*** roliveira has quit IRC08:41
*** sepen has quit IRC08:48
*** percent20_work has joined #crux09:04
tilmanhttp://icculus.org/cgi-bin/ezmlm/ezmlm-cgi?2409:05
tilmanthere it is09:05
*** sepen has joined #crux09:06
*** destruct_ has joined #Crux09:07
tilmangit sucks.09:17
tilmanwhy the hell does it install it's craptastic tcl gui if i don't even have tcl!?09:17
vektoribzr FTW!09:18
jjpkMercurial gets my vote.09:19
vektoriMercurial is also nice.09:20
* jaeger is still using svn09:20
vektoriSubversion is pretty nice for centralized VC.09:20
jjpkI would probably use it as well, but for my needs a decentralized VC works better.09:20
vektoriYeah.09:21
jaegerso far I haven't needed decentralized :) but yeah09:22
jjpkUsed it for around six months for managing my $HOME's and my crux ports :p09:22
*** destruct has quit IRC09:22
vektoriThe fact that SVN doesn't support renames too well sucks, though. :\09:23
tilmanfor quite some time i didn't notice how they were cheating about renames09:23
tilmanit's almost offending to claim that svn supports renames09:23
vektoriYeah.09:24
tilmanoffending to hear them claim*09:24
tilmanrehabdoll: i poked the xorg-server release manager about your problem btw. i think i'll wait until monday or tuesday, and add the patch to the port then09:25
tilmani'd hate to add it now just to remove it on monday because he made a release09:26
tilman:D09:26
tilman  - "git archive" does not insist you to give --format parameter09:27
tilman    anymore; it defaults to "tar".09:27
tilman\o/09:27
*** mike_ has quit IRC09:44
*** cptn has quit IRC09:58
*** sepen has quit IRC10:18
*** pitillo has quit IRC10:28
*** Roomster is now known as Romster10:59
*** sepen has joined #crux11:01
*** sepen has quit IRC11:06
*** jaeger has quit IRC12:41
rehabdolltilman: sweet12:48
aontilman: \o/12:48
aon(fwiw, i'm using cvs for my own crap)12:49
tilmangrep Maint /usr/ports/opt/cvs/Pkgfile12:49
tilmanthat's a good thing aon12:49
aonwhich one?12:49
aonthe one that i'm maintaining cvs or that i use the stuff i maintain?12:50
aons/the one //12:50
tilmani think if you stopped *using* it, your motivation to maintain it might decreasing12:51
aonyes12:51
aonalthough there's not very much to maintain12:51
tilman:)12:52
aoni think i'll migrate to opencvs if there will be p releases of that12:55
*** LithiumFX has joined #crux12:59
LithiumFXHowdy all12:59
thrice`hi13:00
LithiumFXAt the risk of getting accused of being a ricer, I'm wondering if anybody has any experience with Beryl under CRUX (Xorg 7.1)13:00
LithiumFXI'm curious to see what all the fuss is about13:00
thrice`a couple.  jaeger has builds in his repo for it13:00
LithiumFXIt's in my repo too13:00
LithiumFXI can build it fine13:00
thrice`I think the gentoo wiki has a pretty good tutorial13:01
LithiumFXI'm trying to use it with KWM, but beryl-manager says no screens are found13:01
thrice`I've never tried, as I don't run a desktop environment13:01
LithiumFXI'm confused about XGL, is that something I need as well as Nvidia drivers?13:01
tilmanxgl is a opengl based x server13:02
tilmanit is one of two ways to run beryl13:02
tilmanthe other is aiglx13:02
thrice`which is included in xorg13:02
thrice`but I think nvidia provides their own in that respect13:02
LithiumFXSo, if I've got Nvidia drivers, and a stock Xorg 7.1 build13:05
LithiumFXWill I need aiglx or xgl?13:05
tilmanno clue13:05
tilmannvidia's binary crap probably doesn't support all the extensions that you need for aiglx13:05
LithiumFXI don't intend to switch to beryl, I'll probably give up trying getting it to work13:07
LithiumFXI like my Openbox + Xfce4 Panel setup13:07
LithiumFXThanks for your help13:08
*** LithiumFX has quit IRC13:08
*** jaeger has joined #crux13:12
*** ChanServ sets mode: +o jaeger13:12
tilmanaon: does bsd make have the -j switch?13:16
aonyes13:16
tilmanwhat does it do?13:16
aonthe same13:16
tilmangnu date doesn't13:17
aonerm13:17
tilmancan you quote from the bsd manpage?13:17
aonget yo' shit together13:17
aonwhich one, make or date?13:17
tilmanwtf13:17
tilmansorry13:17
*** lasso has joined #crux13:17
tilmani mean date13:17
treachheh, for a moment I thought that was the quoute. :P13:17
aon     -j      Parse the provided date and time and display the result without13:17
aon             changing the clock.13:17
jaegerhehe13:18
aonpoplar[~]$ date -j 200705251234.5613:18
aonFri May 25 12:34:56 EEST 200713:18
aonpoplar[~]$ date13:18
aonFri May 25 21:18:26 EEST 200713:18
tilmanseems gnu date behaves the same without -j13:19
jaegerit does13:19
tilmandoes bsd date *set* the clock if you omit -j and give a date?13:19
jaegerbased on that, I'd guess so, but don't have one handy to test13:20
*** Romster has quit IRC13:24
aontilman: if i drop -j from there, yes13:27
aonbut that's what gnu does, too13:27
tilmanmmh13:27
aonif you give it 052512342007.5513:27
tilmandamn bsd crap13:27
aon(of course they had to fuck up the format...)13:27
aonwhat are you trying to achieve?13:27
aonyou can do date +%H:%M or anything like that on bsd and it doesn't set it13:27
tilmanfixing unit tests in rubinius13:28
aon-j just emulates setting the date without actually doing it13:28
tilmanthey are written against bsd date13:28
aonwhat's the actual command?13:28
tilmandate -j -f %s 0 "+%Y"13:30
tilmaneg13:30
tilman-j -f %s ...13:30
aonokay, so13:31
aonopenbsd date doesn't have -f13:31
tilmanit's probably os x crap13:31
aonah, i get it13:31
aonthat probably returns '1970'13:32
tilmanyeah13:32
aonthis seems to need digging into coreutils sources13:35
* tilman casts level 3 'no thanks'13:36
tilmanaon: thanks though13:36
aoni'm doing it already13:36
aonit seems really grotesque13:36
tilmanheh13:36
aonbeech[~]$ date -d "@0" +%Y13:42
aon197013:42
aonit was in tar info pages13:42
aon:-D13:42
aonperhaps others too, though13:42
tilmanmmh13:43
aonwell, that does what you want to do13:43
tilmanyeah, thanks13:44
aon'-j -f %s 0' == 'simulate setting date in format 'seconds since epoch', the value of which is 0'13:44
aonprobably13:44
tilmanyes13:44
aondamn, i missed the beginning of smith&jones13:44
aon->13:44
*** mike_k has joined #crux14:12
*** Romeo- has joined #crux15:33
Romeo-hello15:33
Romeo-which php-cgi shoul i use`? dahl or jaeger?:)15:34
tilmanuse the one with the higher version15:35
Romeo-jaeger15:36
Romeo-hi tilman15:36
Romeo-i mean how can i tell prt-get to install one of both?:))15:36
*** mike_k has quit IRC15:37
tilmanvim /etc/prt-get.conf :)15:42
*** RedShift has quit IRC15:56
*** percent20_work has quit IRC16:15
Romeo-hm16:29
Romeo-problems whit mysql install:)16:29
treachprt-get readme mysql16:31
Romeo-http://rafb.net/p/Wp0znj65.html16:32
treachif (proper_tarball=1)notify_maintainer(char *md5sum); else (re_download();)16:35
Romeo-http://rafb.net/p/h6Vq4F23.html16:36
Romeo-some:)16:36
treachseems like a fucked up tarball. try another mirror16:38
tilmanapplication/x-tar ... *.tar.gz16:38
tilmanmmh16:38
Romeo-ok, i try now other mirror16:41
Romeo-ok it's work16:43
RyoSyooo Romeo- long time noooo see ^_^16:52
Romeo-hey RyoS:)16:53
RyoSremember me?16:54
Romeo-maby16:55
RyoS?16:55
Romeo-:) if you are this guy wich ners me, no:)16:55
Romeo-nervs16:55
RyoSwhat?16:56
*** treach has quit IRC16:56
*** jdolan_ has quit IRC17:00
tilmano_O17:09
RyoSO_o17:13
jjpkO_O17:13
surroundero_O17:14
RyoSo_o17:14
*** j^2 has quit IRC17:18
rawo_@17:26
*** tilman has quit IRC17:27
Romeo-nice http://rafb.net/p/rOfaBI80.html :))17:44
RyoS>_o17:54
nipuLRomeo-: chroot?18:04
Romeo-ohh18:05
Romeo-i understand18:05
Romeo-tky18:05
Romeo-some:)18:11
*** j^2 has joined #crux18:49
*** mavrick61 has quit IRC19:03
*** Dudde has quit IRC19:03
*** mavrick61 has joined #crux19:04
*** Dudde has joined #crux19:04
*** j^2 has quit IRC19:25
*** lasso|qt has quit IRC19:28
*** lasso|qt has joined #crux19:34
*** lasso has quit IRC20:03
*** j^2 has joined #crux20:17
*** j^2 has quit IRC20:48
*** j^2 has joined #crux20:54
*** j^2 has quit IRC21:03
*** j^2 has joined #crux21:26
*** j^2_ has joined #crux21:47
*** j^2 has quit IRC21:48
*** Romster has joined #crux22:16
*** mrks_ has joined #crux22:32
*** mrks has quit IRC22:46
*** morlenxus has quit IRC23:05
*** morlenxus has joined #crux23:05
*** jaeger has quit IRC23:41
*** Romeo- has quit IRC23:50

Generated by irclog2html.py 2.11.0 by Marius Gedminas - find it at mg.pov.lt!