1296346736 Q * yarihm Quit: This computer has gone to sleep 1296348718 Q * bonbons Quit: Leaving 1296350596 J * rgl ~Rui@a89-154-33-145.cpe.netcabo.pt 1296350599 M * rgl hello 1296351509 M * Bertl_oO hello 1296351606 Q * FireEgl Quit: Leaving... 1296352724 M * daniel_hozac chrissbx: make dist 1296355283 Q * rgl Ping timeout: 480 seconds 1296359236 J * FireEgl ~FireEgl@173-25-19-139.client.mchsi.com 1296364134 M * chrissbx Thanks Daniel, that did it (make dist failed with not finding ChangeLog, but created debian/changelog before the failure) 1296364704 M * chrissbx daniel_hozac: where is PACKAGE_VERSION coming from? It's set to 215, whereas the published package is called 216-preXXX, 1296364720 M * chrissbx and since Debian has some 216 package it would override my locally created one. 1296364730 M * chrissbx So I'm tempting to change it. 1296364820 M * chrissbx (I was meaning to say: I'm attempting to change it. Well ok I'm tempting you to change it, too :), but I'll gladly send you my patch, too, if I get it to work.) 1296365636 M * chrissbx daniel_hozac: some problems when installing the new util-vserver +dependent debs: 1296365638 M * chrissbx update-rc.d: using dependency based boot sequencing 1296365638 M * chrissbx update-rc.d: warning: vservers-default start runlevel arguments (none) do not match LSB Default-Start values (2 3 4 5) 1296365638 M * chrissbx insserv: Service vprocunhide has to be enabled to start service vservers-default 1296365657 M * chrissbx (on Debian testing) 1296366290 J * Romster ~romster@202.168.100.149.dynamic.rev.eftel.com 1296368184 M * chrissbx Bertl_oO: I've got some issue on 2.6.36.3-vs2.3.0.36.38.2 / 0.30.216pre2930 (fresh installation): 1296368216 M * chrissbx created a new vserver, enter works, I do some stuff, exit, then when I try to enter again: 1296368222 M * chrissbx # vserver testing enter 1296368222 M * chrissbx vlogin: fork(): Try again 1296368234 M * chrissbx then bash: fork: retry: Resource temporarily unavailable 1296368303 M * chrissbx vserver x stop gives cannot fork failures, too, but succeeds anyway. vserver x start, works again until some time later the same thing happens. 1296368452 M * chrissbx Is this some resource limit? I didn't change from the defaults as created by vserver build. 1296369544 M * Bertl_oO with 2.6.36.3? 1296369612 M * Bertl_oO ah, .38.2, well, get .38.4 instead :) 1296369693 M * Bertl_oO it's a known problem with credential reuse, we hope it will be fixed soon, the latest patch removes the recently introduced credential code for now 1296369726 M * Bertl_oO (i.e. restores the state we had before credential patches) 1296370393 M * Bertl_oO off to bed now ... have a good one everyone! 1296370397 N * Bertl_oO Bertl_zZ 1296370666 J * yarihm ~yarihm@178-83-213-36.dynamic.hispeed.ch 1296371061 M * arekm er 1296372478 M * nkukard daniel_hozac, around? 1296372779 Q * Piet Ping timeout: 480 seconds 1296373420 J * Piet ~Piet__@04ZAACC94.tor-irc.dnsbl.oftc.net 1296376115 M * arekm Bertl_zZ: ready for new patches ;) 1296378226 J * petzsch ~markus@dslb-088-075-124-240.pools.arcor-ip.net 1296378249 J * BenG ~bengreen@cpc12-aztw24-2-0-cust146.aztw.cable.virginmedia.com 1296378296 J * Piet_ ~Piet__@04ZAACDB8.tor-irc.dnsbl.oftc.net 1296378600 Q * Piet Ping timeout: 480 seconds 1296380391 J * bonbons ~bonbons@2001:960:7ab:0:2c0:9fff:fe2d:39d 1296382132 Q * BenG Quit: I Leave 1296382178 J * BenG ~bengreen@cpc12-aztw24-2-0-cust146.aztw.cable.virginmedia.com 1296384357 Q * BenG Quit: I Leave 1296384960 J * pmenier ~pmenier@ACaen-152-1-56-244.w83-115.abo.wanadoo.fr 1296385004 Q * Piet_ Read error: Connection reset by peer 1296386413 J * rgl ~Rui@a89-154-33-145.cpe.netcabo.pt 1296387050 N * ensc Guest2116 1296387059 J * ensc ~irc-ensc@p5DF2D25B.dip.t-dialin.net 1296387203 M * Guy- how does CONFIG_SCHED_AUTOGROUP play with vserver? 1296387444 Q * Guest2116 Ping timeout: 480 seconds 1296390631 J * Piet_ ~Piet__@659AAB7EX.tor-irc.dnsbl.oftc.net 1296390696 N * Piet_ Piet 1296393479 Q * ensc|w Remote host closed the connection 1296394585 J * ensc|w ~ensc@www.sigma-chemnitz.de 1296396084 Q * petzsch Quit: Leaving. 1296396536 J * petzsch ~markus@dslb-088-075-124-240.pools.arcor-ip.net 1296397136 J * hijacker_ ~hijacker@87-126-142-51.btc-net.bg 1296398908 J * JonB ~NoSuchUse@77.75.164.169 1296398917 N * ensc Guest2127 1296398927 J * ensc ~irc-ensc@p5DF2E2A5.dip.t-dialin.net 1296398940 M * JonB did anyone ever figure out how to enable netatalk inside a vserver guest? 1296399016 Q * Guest2127 Ping timeout: 480 seconds 1296399260 N * ensc Guest2128 1296399270 J * ensc ~irc-ensc@p5DF2CBBA.dip.t-dialin.net 1296399419 Q * Guest2128 Ping timeout: 480 seconds 1296399781 N * ensc Guest2129 1296399790 J * ensc ~irc-ensc@p5DF2FF9C.dip.t-dialin.net 1296399908 Q * Piet Remote host closed the connection 1296399956 J * Piet ~Piet__@659AAB7H5.tor-irc.dnsbl.oftc.net 1296400129 Q * Guest2129 Ping timeout: 480 seconds 1296400401 Q * ensc Ping timeout: 480 seconds 1296400452 Q * JonB Quit: Leaving 1296400631 J * ensc ~irc-ensc@p5DF2E797.dip.t-dialin.net 1296401351 N * Bertl_zZ Bertl_oO 1296402454 Q * eyck_ Ping timeout: 480 seconds 1296402639 J * kirodar ~Administr@p4FCDF2AA.dip.t-dialin.net 1296402770 M * kirodar hi everybody - could somebody help me to debug an issue? 1296402780 M * kirodar I've got the following situation: 1296402798 M * kirodar I did run Debian Lenny with a backports vserver kernel so far 1296402820 M * kirodar linux-image-2.6-vserver-amd64 (2.6.32+27~bpo50+1), util-vserver was probably 0.30.216~r2772-6 1296402870 M * kirodar last night I dist-upgrade'd to squeeze - however, I did also install a new kernel: 2.6.37 + latest vserver patch 1296402916 M * kirodar from then on I couldn't vserver start my vservers any more because I did run into this issue: http://www.freak-search.com/de/thread/2814383/vserver_failing_to_start_vserver_since_patch-2.6.36-vs2.3.0.36.38 1296402931 M * kirodar therefore I compiled the latest util-vserver and I've got the following error message now: 1296402967 M * kirodar /home/util-vserver/sbin/vserver webserver start 1296402968 M * kirodar vcontext: execvp("/home/util-vserver/sbin/vspace"): No such file or directory 1296402968 M * kirodar ... 1296403008 M * kirodar util-vserver is util-vserver-0.30.216-pre2926 1296403072 M * kirodar I assume that vspace is used to start other processes? currently the error message is not very helpful - therefore I'd like to know if there's a way to extend the debug output to better trace down the problem 1296404118 J * eyck ~eyck@77.79.198.68 1296404609 M * arekm does the file exist? 1296404625 M * kirodar yes, it does 1296404639 M * arekm ldd on the file 1296404706 M * Bertl_oO kirodar: vserver-info - SYSINFO 1296404708 M * Bertl_oO (please use paste.linux-vserver.org for everything longer than 3 lines) 1296404733 M * kirodar sorry for the spam :-) 1296404737 M * kirodar I guess I found the problem 1296404743 M * kirodar or at least the cause ;) 1296404761 M * kirodar dmesg did contain lines like '[ 3965.220201] vxW: [▒vcontext▒,15188:#23|23|23] did hit the barrier.' 1296404798 M * kirodar probably because the actual vdir was located at /home/webserver/ and the util-vserver stuff was installed to /home/util-vserver/ 1296404816 M * kirodar just for testing I installed util-vserver to /util-vserver/ and it works now 1296404852 M * Bertl_oO interesting places where you install your stuff :) 1296404870 M * Bertl_oO but yes, you can't put util-vserver config or tools below a barrier 1296404970 M * kirodar /home/ is my encrypted partition :D 1296404972 M * kirodar and why is the barrier (probably?) set to /home/ ? did I have to do that in the past to get vserver running? 1296404995 M * Bertl_oO the barrier should be placed at /path/to/guest/.. 1296405012 M * Bertl_oO so as your guest is on /home/webserver, /home got a barrier 1296405049 M * Bertl_oO I'd suggest to use /home/vservers for guests, this way the barrier will end up on /home/vservers 1296405136 M * kirodar thanks for the hint 1296405158 M * Bertl_oO np 1296405375 M * kirodar btw: do debian's plans, to possibly stop integrating linux vserver after squeeze, influence your work in any way? 1296405410 M * kirodar I'm wondering if I should have a look at LXC - so far I've been always a linux vserver user 1296405618 M * arekm lxc miss few scripts to compete with util-vserver scripts 1296405719 M * kirodar but it's main line and that saves a lot of trouble :D 1296405727 M * Bertl_oO kirodar: not really, at some point, util-vserver will work without kernel patches (in a few years), but not much will change for the end users 1296405740 M * kirodar for the same reason I stopped using loop-aes in favour of dm-crypt / luks 1296405849 M * kirodar and for what reasons will be no kernel modifications needed in the future? is the LXC kernel code more general and can be used by other projects, too? 1296405872 M * Bertl_oO we already use whatever is available from mainline 1296405889 M * Bertl_oO that always was the philosophy behind Linux-VServer 1296405939 M * arekm still, vserver patch is gigantic somehow 1296405946 M * kirodar true ;-) 1296406021 M * arekm lxc has one ugly thing afaik, host sees all guest processes 1296406037 M * Bertl_oO gigantic in what respect? 1296406069 M * kirodar almost 1MB in size? 1296406072 M * arekm 700kb of size, 17 000 inserted lines 1296406084 M * cehteh arekm: there are no process namespaces yet? 1296406086 M * Bertl_oO did you check the last kenel delta? 1296406115 M * arekm cehteh: afaik parent namespace sees own childs 1296406167 M * arekm what kernel delta have to do with patch implementing some feature? you could compare to other patches adding some features but not to kernel delta 1296406173 M * cehteh i didnt played with it yet, i am looking forward that lxc can take over vserver functionality, but we are not there yet 1296406195 M * cehteh and its a bit sad for some people here then :P .. but great for the users 1296406233 M * arekm cehteh: linux vserver power IMO comes mainly from great scripts that manage all that. Which feature do you miss in lxc that is in vserver? 1296406270 M * arekm (beside network isolation of course... but network namespace is even nicer IMO ;) 1296406273 M * cehteh yes .. i think vserver will stay alive even if no kernel patch is needed anymore .. just because the userspace tools and infrastructure 1296406318 M * Bertl_oO I hate it if folks consider the Linux-VServer patch huge without comparing it to e.g. kernel changes 1296406326 M * cehteh can recent vserver use network namespaces? 1296406336 M * Bertl_oO yes 1296406349 M * cehteh (i am always to lazy to follow development and using way too old kernels P) 1296406363 M * cehteh cool thanks 1296406380 M * arekm Bertl_oO: kernel change == tons of different things inside. vserver == one thing. How can you compare that? 1296406436 M * arekm ceneth. it not really can't. kernel can, userspace doesn't support it in a usable way (but there a hacky way to make it work) 1296406451 M * Bertl_oO one thing? you have no idea what needs to be done to get a secure jail :) 1296406563 M * cehteh lxc is not there yet :) .. 10 years catching up but still not 1296406655 M * cehteh Bertl_oO: btw, kernel question, do you see some way to make file timestamps trusted by removing the user ability to alter them with utimes() possibly by some capability drop? 1296406785 M * arekm many things could go from vserver or be implemented differently like xattrs instead of patching each fs (yeah, slower but at least usable and doesn't change on disk structure every few releases), network isolation (but many things prefer isolation over net namespace). These two things likely make 1/2 of the patch size 1296406879 A * arekm uses vserver on almost every server and is happy with it (some problems like that creds one arise but well...) 1296406913 M * Bertl_oO arekm: patches for xattr special attribute support are welcome 1296406951 M * Bertl_oO (actually ea, but that's what you meant anyway) 1296406952 M * cehteh i use vserver too and be happy with it, but there are always ideas how it could be smoother 1296406980 M * arekm I know patches are welcome. Patches are welcome everywhere. I'm saying that it could be stripped off from the patch since it works only by luck or deep knowledge of what changes between each vserver patch 1296406984 M * cehteh while currently i am mostly out of business .. building a new kitchen for our new flat, finally somthing not programming related :P 1296407013 M * Bertl_oO arekm: it is the only way to provide the expected performance 1296407045 M * Bertl_oO arekm: why do you think are the other xattrs (in mainline) implemented in this way and not via ea? 1296407114 M * kirodar arekm: which creds problem? 1296407171 M * cehteh some notes about vserver wishes: allow bind mounts from inside a vserver, allow mknod of a device already existing (i am possibly just write a mknod script hardlinking to an existing device), and this utimes thing from above, allowing to run aide much faster only revisiting changed/new files 1296407224 M * arekm Bertl_oO: performance doesn't help if on disk format changes and you can't upgrade at all (xfs case) 1296407242 M * arekm Bertl_oO: I understand your point while I'm sure you understand mine 8) 1296407278 M * Bertl_oO there are proper upgrade procedures for that 1296407292 M * arekm kirodar: SO_PEERCRED fails to work properly (not sure what are other implications of that bug). Bertl works on that 1296407322 M * arekm Bertl_oO: like dumping and restoring filesystem? that doesn't really work 1296407364 M * Bertl_oO kirodar: well, no problem with creds, unless you use 'vserver - enter' which will not get the same credentials as the init process inside the guest 1296407415 M * Bertl_oO (this has been so for many years, we recently tried to improve that because of problems some folks have with managing guest services via 'enter') 1296407461 M * Bertl_oO arekm: you can just dump and restore the flags 1296407508 M * Bertl_oO daniel_hozac did a script/tool for that when debian made the funny 2.6.26 kernels :) 1296407521 M * arekm and would have to somehow zeroe old flags space on disk (because it's used for other things) 1296407559 M * Bertl_oO which is a single comman, so? 1296407575 M * Bertl_oO +d 1296407643 J * kir ~kir@swsoft-msk-nat.sw.ru 1296407666 P * kir 1296407679 M * arekm and how many people here even know what we are talking about? ;) no one knows the issue and procedures 1296407706 M * Bertl_oO and how many people here use xfs? :) 1296407716 A * arekm . 1296407816 M * cehteh i try hard to drop it .. iirc i dont have it on any vserver machine anymore 1296407861 M * kirodar I only use xfs for filesystems >1TB 1296407862 M * arekm Bertl_oO: selinux uses ea for thing. I guess the penalty for vserver using ea would be the same as it is now for selinux, correct? 1296407887 A * arekm uses for (almost) everything. IMO best fs under Linux 8) 1296407893 M * cehteh xfs has some ugly performance behavior .. and old xfs's tend to become slow .. 1296407899 M * cehteh (even if defragmented) 1296408057 M * julius <- uses xfs whereever he can 1296408067 M * julius haven't done any performance tests tho 1296408210 M * cehteh the one on my wifes computer is now serveral years old and almost come down to a crawl .. i just buyed new biggier disks and soon move all data over to an ext4 1296408242 M * arekm ask on #xfs what could be reason for that, I'm really interested because I haven't seen such behaviour 1296408257 M * cehteh i dont care, i go for ext4 and someday btrfs 1296408273 M * kirodar how fast is fsck.ext4 of a 1+TB volume? 1296408286 M * arekm eh, ugly attitude ;/ 1296408287 M * Bertl_oO xfs is slow, buggy and alien to linux .. mainly because it was 'hacked' into the kernel 1296408295 M * kirodar I was tired of waiting several hours for a fsck.ext3 1296408302 M * cehteh yeah mostly it feels alien because of its tools 1296408312 M * julius o.o 1296408346 M * cehteh kirodar: i'll try that .. have 3x 2TB disks which will be set up as raid10 1296408350 M * arekm Bertl_oO: maybe it was long time ago. Now it's more Linuxish than SGIsh 1296408383 M * Bertl_oO nah, not really, I know the kernel code with the dual layer indirect everything :) 1296408397 M * cehteh in my tests so far, nilfs looks very promising for cases where you need a very reliable filesystem with a lot of writes 1296408406 M * cehteh nilfs2 1296408436 M * arekm example of such dual layer in xfs would be welcome 1296408443 M * cehteh cant wait for it to be offically released .. i think that makes a nice filesystem for backup servers 1296408443 M * Bertl_oO btrfs has potential to become the new Linux filesystem, but the developer attitude and priorities are somewhat strange 1296408460 M * Bertl_oO arekm: did you read the xfs code at all? 1296408480 M * cehteh Bertl_oO: yes ... and btrfs is not there yet, i always find some weak spots which prevent me to use it even for non serious stuff 1296408506 M * cehteh btrfs will certainly become the defacto general purpose FS in a few years 1296408526 M * cehteh but if you have some write heavy use-case look at nilfs 1296408585 M * cehteh meanwhile chris works on a real working fsck for btrfs .. hehe 1296408619 M * cehteh currently its so much brittle .. grr 1296408652 M * Bertl_oO kirodar: why do you run an fsck on ext3? 1296408775 M * Bertl_oO I mean, it's not a bad thing to run it, just curious why you seem to run it on a regular basis (if you care about the time it takes) 1296408813 M * cehteh Bertl_oO: sometimes fsck's are really nice .. in case the disk did not write what it was told to write (failing sector) or other quirks ... journaling is not an assurance against physical metadata corruption 1296408867 M * cehteh if i reboot servers with a long uptime and i have some patience i usually shutdown -F them .. just to be sure it once a while fscks all filesystems 1296408883 M * cehteh better safe than sorry 1296408909 M * Bertl_oO yeah, I totally agree, nice to have that on _any_ filesystem 1296408944 M * cehteh hehe and my old habbit to run badblocks in write mode over a fresh buyed disks tends to become extensive now .. 1296408982 M * cehteh all 5 passes on my wifes new 2TB disks in a usb enclosure will take more than a week .. times 3 disks .. haha 1296409011 A * cehteh breaks it now after the first pass (write/verify) completed 1296409030 M * Bertl_oO not that it will help anything as bad blocks are remapped nowadays 1296409041 M * cehteh exactly thats the intention 1296409061 M * kirodar that will also happen during daily operation - you don't have to enforce it ;-) 1296409064 M * cehteh find bad blocks before writing vital data to it .. and also see if there any 1296409070 M * Bertl_oO cehteh: dd if=/dev/zero of=/dev/xxx bs=1M 1296409070 M * kirodar a smart long test should be sufficient 1296409085 M * cehteh kirodar: nop smart test doesnt suffice 1296409095 M * cehteh smart check aborts on the first found error 1296409097 M * fback Bertl_oO: can I safely upgrade to latest 2.6.32 with u-v r2834? 1296409109 M * Wonka cehteh: can't you run badblocks on your three disks in parallel? ;) 1296409116 M * Bertl_oO fback: unlikely 1296409129 M * cehteh i have only one external usb enclosure .. this is just for testing 1296409153 M * Wonka ah 1296409165 M * cehteh and usb is the bottleneck not badblocks or dd (badblocks has options to increase the buffer size too) 1296409219 M * Wonka well, plug them to different root hubs ;) 1296409242 M * cehteh wont still not max out the disk 1296409256 M * cehteh well i have patience 1296409288 M * cehteh just plugged in the last disk some hours ago .. so somtime next week i can start formatting it and move data over, no urgwe 1296409317 M * Wonka hmm... by the way, haven't you been in some channels with me on IRCnet? 1296409330 M * cehteh yes :P 1296409344 A * arekm back 1296409368 M * cehteh also on freenode 1296409386 M * Wonka ok, then my memory does still wor ;) 1296409388 M * Wonka +k 1296409398 M * Wonka thought I knew that nick 1296409401 M * cehteh lol 1296409444 M * arekm Bertl_oO: I would prefer an answer instead of question but ... I did read and change only small part of xfs code (project id related) and didn't see any weirdness but I'm NOT a kernel developer, so cannot tell whether thing I see is double layer or not 1296409514 M * fback Bertl_oO: I've had very bad experience with ext3 1296409670 M * arekm (I assume we talk about double features like linux having own xyz handling and xfs having own instead of using generic one (and not some typecasting for example)) 1296409883 Q * pmenier 1296409911 M * kirodar I'll go off again - thank you everybody for your help 1296409920 Q * kirodar Quit: Leaving. 1296409989 M * fback Bertl_oO: then r2864 is probably too old too (maybe I should mention it earlier, it seems to work with 2.6.32.10-vs2.3.0.36.29.2) 1296410037 M * Bertl_oO arekm: xfs_vn_lookup -> xfs_lookup -> xfs_dir_lookup -> xfs_dir2_sf_lookup for example 1296410067 M * Bertl_oO (the first one is the 'linux wrapper' for the second one, which is the 'xfs wrapper' for the actual filesystem operation 1296410351 M * arekm xfs_dir_lookup would be gigantic if all xfs_dir2_*_lookup() would be merged into it 1296410377 M * Bertl_oO that's fine, but what about the linux wrapper? 1296410450 M * Bertl_oO or to put it the other way round, do you see a similar structure in ext2/3? 1296410478 M * Bertl_oO (start from the inode_operations) 1296410562 M * arekm xfs_lookup is used by xfs_vn_ci_lookup, xfs_vn_lookup, xfs_fs_get_parent. All some to use single xfs_lookup instead of duplication 1296410582 M * Guy- how does CONFIG_SCHED_AUTOGROUP play with vserver? 1296410820 M * Bertl_oO Guy-: you tell us? I presume it will do fine 1296410921 M * arekm Bertl_oO: looks like they are doing these things to be compatible with some old fses (xfs_sb_version_hasasciici()) 1296410944 M * arekm yeah, irix crapolla 1296411230 M * Bertl_oO yeah, well, all over the place in xfs :) 1296411260 M * Bertl_oO and it's getting even worse when xfs subroutines use the linux interfaces :) 1296411276 M * arekm one of xfs developers works at redhat on ext3/4 and it quite often finding craps in ext4 code for example... so well, both are not perfect 1296411297 M * arekm and now ext4 developer (Tytso) performance testing - xfs wins - http://thunk.org/tytso/blog/2010/11/01/i-have-the-money-shot-for-my-lca-presentation/ 1296411316 M * arekm "We’re now within striking distance of XFS, and there’s more improvements to ext4 which I have planned that should help its performance even further." 1296411367 M * Bertl_oO not saying that xfs has bad performance, it mainly suffers from bad design, bad integration and ugly code 1296411395 M * Bertl_oO and of course, it falls apart after some time ... 1296411396 M * arekm suffers as for xfs developer, not xfs user 1296411439 M * arekm nah, using the same 400GB xfs fs created over 5 years ago and no problems with it (on very busy apache server with 3k vhosts) 1296411486 M * arekm that's why I was wondering what's happening for those who see a problem after some time 1296411523 M * arekm Anyway, who cares. I would like to test new creds patch instead of talking about xfs 8-)) 1296411523 M * Bertl_oO well, I tested xfs for some time as alternative to ext3, and the filesystem got slower and slower over time, and eventually got in a state where I had to restore, because the filesystem was simply broken 1296411563 M * Bertl_oO it wasn't that extreme as with reiserfs, which didn't even run for a week or so :) 1296411652 M * Guy- Bertl_oO: thanks 1296411674 M * arekm aka killerfs ;/ 1296411704 M * arekm repair tool for reiserfs is the worse part of reiserfs IMO 1296411772 M * arekm (.39 kernel will have xfs delaylog option turned on by default (after 1 year of testing as a option) - that gives huge performance improvement on metadata handling) 1296411863 M * cehteh i can only agree with bertl .. xfs gets slower and slower 1296412000 A * arekm asked #xfs devs about this. Very interesting problem. 1296412029 M * arekm kernel.org uses xfs for everything btw 1296412087 M * Bertl_oO does it do meta data and data checksumming by now? 1296412166 M * arekm Bertl_oO: no, there are some patches but unfinished and afaik no one works one these at this moment 1296412205 M * arekm does any other fs than (experimental) brtfs do such checksumming anyway? 1296412223 M * Bertl_oO zfs, but that's not really an option on Linux 1296412300 M * arekm dchinner told me few month ago that checksumming is on his list after scalability patches (and scalability patches were merged into after .37) 1296412311 M * arekm so there is a chance he will work on checksuming now 1296412351 M * arekm note that xfs hits scalability problems in ... vfs :) (there was work on this recently; some patches afaik even got merged) 1296412418 N * ensc Guest2151 1296412428 J * ensc ~irc-ensc@p5DF2E797.dip.t-dialin.net 1296412588 Q * Guest2151 Ping timeout: 480 seconds 1296415083 M * cehteh ext4 does at least metadata checksumming 1296415112 M * cehteh (or was that just the journal?) 1296415209 Q * rgl Read error: Connection reset by peer 1296415229 Q * eyck Ping timeout: 480 seconds 1296415240 J * rgl ~Rui@a89-154-33-145.cpe.netcabo.pt 1296415824 J * eyck ~eyck@77.79.198.68 1296417617 Q * manana Remote host closed the connection 1296419330 J * manana ~mayday090@84.17.25.149 1296420981 J * BenG ~bengreen@cpc12-aztw24-2-0-cust146.aztw.cable.virginmedia.com 1296420984 Q * BenG 1296421997 J * eyck_ ~eyck@77.79.198.66 1296422042 Q * eyck Read error: Connection reset by peer 1296423113 Q * petzsch Quit: Leaving. 1296423596 M * Guy- I did some benchmarks recently using iozone, and xfs was far from slow; with delaylog enabled even metadata intensive workloads were fast 1296426063 Q * hijacker_ Quit: Leaving 1296426135 Q * ghislain Quit: Leaving. 1296426455 Q * bonbons Quit: Leaving 1296427991 Q * yarihm Quit: Leaving 1296429584 Q * micah Remote host closed the connection 1296430073 J * micah ~micah@micah.riseup.net 1296431780 M * cehteh Guy-: use that filesystem 24/7 for 5 years and then benchmark again 1296431820 A * cehteh think xfs has some problems with global fragmentation 1296431846 M * cehteh files are defragmented .. but littered all around the disk