1255998393 J * imcsk8 ~ichavero@189.155.146.147 1256000073 Q * imcsk8 Quit: This computer has gone to sleep 1256002460 N * Bertl_oO Bertl 1256002463 M * Bertl back now ... 1256005179 M * Bertl off to bed now .. have a good one everyone! 1256005183 N * Bertl Bertl_zZ 1256005883 Q * orzel_ Remote host closed the connection 1256007658 J * saulus_ ~saulus@d004103.adsl.hansenet.de 1256007887 Q * balbir_ Ping timeout: 480 seconds 1256008068 Q * SauLus Ping timeout: 480 seconds 1256008072 N * saulus_ SauLus 1256008709 J * balbir_ ~balbir@122.172.7.32 1256008714 J * scientes ~scientes@174-21-161-91.tukw.qwest.net 1256010075 J * derjohn_foo ~aj@c192190.adsl.hansenet.de 1256010501 Q * derjohn_mob Ping timeout: 480 seconds 1256010750 Q * fback_ Read error: Connection reset by peer 1256010778 J * fback fback@red.fback.net 1256012321 J * Piet_ ~piet@04ZAABMLI.tor-irc.dnsbl.oftc.net 1256012563 Q * Piet Ping timeout: 480 seconds 1256012811 Q * balbir_ Ping timeout: 480 seconds 1256013046 Q * ard Ping timeout: 480 seconds 1256013168 Q * Piet_ Ping timeout: 480 seconds 1256013191 Q * nou Ping timeout: 480 seconds 1256013329 J * nou Chaton@causse.larzac.fr.eu.org 1256013477 J * balbir_ ~balbir@122.172.42.145 1256013775 J * Piet_ ~piet@04ZAABML3.tor-irc.dnsbl.oftc.net 1256013799 N * Piet_ Piet 1256013804 Q * scientes Ping timeout: 480 seconds 1256015415 Q * Mark17 Ping timeout: 480 seconds 1256015828 J * Mark17 ~mark@195.184.64.194 1256016867 J * geb ~geb@earth.gebura.eu.org 1256016883 J * kir ~kir@swsoft-msk-nat.sw.ru 1256017147 J * imcsk8 ~ichavero@189.155.146.147 1256017220 M * Alteisen 1256017839 Q * Snow-Man Ping timeout: 480 seconds 1256019001 J * Snow-Man ~sfrost@tamriel.snowman.net 1256019518 J * doener ~doener@i59F56417.versanet.de 1256019683 Q * balbir_ Ping timeout: 480 seconds 1256019776 J * ard ~ard@shell2.kwaak.net 1256019894 N * DoberMann[ZZZzzz] DoberMann 1256019920 J * sharkjaw ~gab@90.149.121.45 1256020218 J * lucrus ~papo@host-84-223-100-32.cust-adsl.tiscali.it 1256020763 Q * derjohn_foo Ping timeout: 480 seconds 1256021381 J * ghislainocfs2 ~Ghislain@LPuteaux-151-41-11-129.w217-128.abo.wanadoo.fr 1256022011 J * balbir_ ~balbir@122.248.163.1 1256022052 J * derjohn_foo ~aj@80.85.196.112 1256022217 Q * derjohn_foo Remote host closed the connection 1256022924 J * derjohn_mob ~aj@80.85.196.112 1256027145 M * tokkee Heyha ;-) 1256027186 M * tokkee According to http://lists.debian.org/debian-devel-announce/2009/10/msg00003.html Linux-VServer support in Debian will be deprecated (possibly) in favor of Linux containers. 1256027209 M * tokkee Has that been discussed with any of the upstream developers? 1256027230 M * tokkee Are there any plans to merge the Linux-VServer features into Linux containers? 1256027781 M * arekm did you ever saw any attempt of merging vserver stuff upstream? I did not, so it likely will never happen. 1256027822 M * arekm containers will get more and more features and vserver will become obsolete in the end 1256027851 M * arekm (that's my guess... too bad that containers don't have such cool thing as util-vserver package ;) 1256028396 M * ghislainocfs2 bertl told on the list that they will continue until mainline is on par with vserver and that they will then continue the tools only 1256028445 M * ghislainocfs2 the issue is that bertl way is to have performance and simple solutions where i fear mainline will use "do too much things" solution with less performances 1256028451 M * ghislainocfs2 anyway future will tell 1256029110 M * Alteisen as long as debian provides a util-vserver package, i am fine, as i am using my own vanilla kernel with needed patches (such as linux-vserver) 1256029562 Q * lau Ping timeout: 480 seconds 1256029654 J * laurent ~laurent@AToulouse-157-1-48-92.w86-201.abo.wanadoo.fr 1256029855 M * ghislainocfs2 for the package i work on a debian for the last tools 1256029882 M * ghislainocfs2 i just spend a lot of time fighting the debian tools and learning packaging so it takes longer than expected to release 1256030496 M * pmjdebruijn ghislainocfs2: thus far lxc isn't feature bloated at all 1256030819 M * sid3windr feature meager! 1256030821 M * ghislainocfs2 pmjdebruijn: no but i wonder if the cgroup system is as performant as the vserver was, i can argue that the features seems interesting but i wonder what is the cost (if i had time to test ahhh time.. :) ) 1256030933 M * pmjdebruijn I could live with a little less performance 1256030941 M * pmjdebruijn not having to patch a kernel, is worth A LOT to me 1256030965 M * ghislainocfs2 :) 1256030965 M * pmjdebruijn patching a kernel, always sucks by definition 1256030980 M * ghislainocfs2 default distrib kernels are not tuned but oh well 1256030987 M * pmjdebruijn "tuned" 1256030999 M * pmjdebruijn we build our own as well... but not because of "tuning" 1256031290 M * ghislainocfs2 os many ways of doing things :) we "tune" it but with our limited knowledge so .. mainly it is used to remove the 204574857398498435 unused drivers in the kernel :) 1256031303 M * ghislainocfs2 s/os/so/ 1256031683 M * _Shiva_ ghislainocfs2: that reminds me of my idea about a "probeconfig" target in the Kbuild .. ;-) i.e. take a full-featured distro-Kernel and let probeconfig generate an optimized .config for the machine it's running on ;-) 1256031710 M * _Shiva_ at least for the hardware that is 1256031710 N * Bertl_zZ Bertl 1256031714 M * Bertl morning folks! 1256031726 M * pmjdebruijn ghislainocfs2: we do that as well... but that's not really a valid reason so build your own kernels :) 1256032014 M * Bertl tokkee: Linux-VServer wasn't merged because of political reasons, it is now slowly re-invented in mainline ... 1256032108 M * Bertl and yes, mainline is is focusing on more heavy and general isolation than Linux-VServer, and I fear, the kernel already got about 2-5% slower with those virtualizations (but no actual test results here either) 1256032164 M * Bertl pmjdebruijn: what's your problem with patching a kernel (or custom tailoring it)? 1256032236 M * Alteisen pmjdebruijn: hm, if you have "small" hardware, you are not going to waste performance 1256032732 J * Pazzo ~ugelt@reserved-225136.rol.raiffeisen.net 1256032821 M * pmjdebruijn Bertl: you're making it work well... 1256032832 M * pmjdebruijn Bertl: not all external patches are as well maintained 1256033274 M * Bertl so, it's not patching the kernel you hate, it's patching with non-well maintained patches you hate? 1256033431 Q * elesouef Ping timeout: 480 seconds 1256033442 M * pmjdebruijn yeah 1256033463 M * pmjdebruijn Bertl: but almost all patches are ill-maintained 1256033476 M * sid3windr well 1256033483 M * sid3windr problem is also that they don't always play nice *with eachother* ;) 1256033508 M * pmjdebruijn that too 1256034047 M * tokkee Bertl: Okay - so, I suppose, non of the Debian Kernel team members has approached you so far? 1256034081 J * elesouef ~elesouef@212.51.177.50 1256034103 M * tokkee arekm: Well, I guess, auxiliary tools will emerge eventually, when / if Linux containers become more commonly used. 1256034126 M * tokkee Bertl: That speed-down should be a reason to re-consider that decision … 1256034229 M * Bertl tokkee: the debian folks and I are working on a plan, the depreciation is due to Bastian Blank favoring LXC over Linux-VServer 1256034299 M * sid3windr so I read that debian is also deprecating xen :] 1256034313 M * sid3windr in favour of kvm I presume 1256034319 M * Bertl yes, korrekt 1256034320 M * sid3windr pity my cpu's don't do VT then 1256034547 M * tokkee sid3windr: Yep - that's another decision I dislike quite a bit :-/ 1256034565 M * tokkee Imho, Xen is much nicer technology than KVM … 1256034596 M * pmjdebruijn almost everybody is depreciating Xen 1256034617 M * Bertl the problem with Xen is that it never was really 'well-maintained' 1256034619 M * tokkee It's a pity that upstream is kinda dead / working in the background … 1256034780 J * Guest1705 ~user@143.Red-83-51-217.dynamicIP.rima-tde.net 1256035190 Q * Guest1705 Quit: Saliendo 1256036091 J * Shinsaku ~Shinsaku@chello089076140236.chello.pl 1256036470 J * thierryp ~thierry@zanzibar.inria.fr 1256036964 J * elesouef_ ~elesouef@212.51.177.50 1256037373 Q * elesouef Ping timeout: 480 seconds 1256039919 J * doener_ ~doener@i59F57083.versanet.de 1256039939 J * tanjix tanjix@77.37.0.150 1256039943 M * tanjix hi @ll 1256039961 M * tanjix when i try to start a guest, i get: 1256039962 M * tanjix vcontext: mkdir(): No space left on device 1256039968 M * tanjix but there is space left? 1256040023 Q * doener Ping timeout: 480 seconds 1256040222 M * _Shiva_ tanjix: maybe space - what about Inodes? (df -i) 1256040503 M * pmjdebruijn tanjix: quota? 1256040823 M * ard ext3 and an old version of vserver (2.3.0.14 something..)? 1256040993 P * kir Leaving. 1256041001 M * tanjix yes it seems to be the inodes 1256041010 M * tanjix it's 100% used 1256041152 M * mnemoc ext* suck :p 1256041370 Q * Shinsaku 1256041426 M * Bertl mnemoc: yeah, reiserfs is the future ... wait, maybe not :) 1256041453 M * mnemoc atleast until btrfs gets mature :p 1256041487 M * pmjdebruijn I don't want to whine 1256041495 M * pmjdebruijn but it's pretty hard to run out of inodes on ext3 1256041522 M * pmjdebruijn tanjix: you should probably look into how they got exhausted 1256041532 M * mnemoc get a production mail server using maildir, wait some weeks and you are done =) 1256041533 M * pmjdebruijn but btfs will be great 1256041554 M * Bertl pmjdebruijn: depending on the filesystem configuration it can be very easy 1256041566 M * Bertl (running out of inodes that is) 1256041608 M * mnemoc on os build server it's also easy if you don't *4 the default number of inodes 1256041655 M * pmjdebruijn still 1256041658 M * mnemoc sometimes even *4 is not enough :< 1256041661 M * pmjdebruijn he should analyze why he ran out 1256041695 M * mnemoc sure, but ext*'s inodes limitation still sucks :p 1256041719 M * mnemoc and enabling huge_disk by default also sucks :p 1256041724 M * Bertl there isn't really a limitation (within reasonable values) for inodes 1256041733 M * ghislainocfs2 ext has done the job for years, its just a little aged now 1256041802 M * ghislainocfs2 btrfs seems very promising really, even better than zfs it seems 1256041840 M * ghislainocfs2 we just have to let 5 years of real life testing before we can use it ;p 1256041875 M * mnemoc i hoped ext4 would support to extend the number of inodes after formating, but it doesn't.... you need to create another partition and move the data :| 1256041979 M * mnemoc (god bless lvm) 1256042208 M * cehteh mnemoc: there are mkfs options to set the number of inodes 1256042395 M * mnemoc cehteh: sure.... but sometimes you don't know off-hand what's a good number 1256042422 M * cehteh make notes about the old numbers which ran out of space 1256042422 M * mnemoc one learns... but it should be the job of the fs to handle it, not yours 1256042427 M * cehteh yeah 1256042431 M * cehteh but btrfs is not ready 1256042440 M * mnemoc reiserfs3 is :) 1256042447 A * cehteh follows developement and tries it sometimes 1256042464 M * Bertl mnemoc: ready and crashing :) 1256042467 M * cehteh basically works, a lot of features work too, but there are some rather big issues open 1256042469 M * mnemoc :p 1256042486 M * cehteh nilfs looks also promising for some kinds of workloads 1256042518 M * cehteh means mostly writing, where seeking is expensive and you have a memory for caching reads 1256042540 M * cehteh but its also not ready and its performance is still not as good as it could be 1256042548 M * cehteh and sometimes crashes :P 1256042609 M * Bertl btw, changing the inode numbers on an unmounted (offline) filesystem for ext2/3 should not be too hard, I wouldn't be surprised if there was a tool to exaclty do that 1256042681 M * cehteh uhm doesnt store ext* the inodes at fixed locations? 1256042702 M * cehteh cant remember exactly 1256042739 M * Bertl well, you know resize2fs, what do you think does that tool do with the inode tables when resizing the filesystem? 1256042752 M * cehteh well, give btrfs half a year then it becomes useable for non critical stuff 1256042808 M * cehteh i am not sure, i just thought there was a defined interval where inode blocks and datablocks appear on disk and resizefs just appends there 1256042813 M * cehteh but maybe i am wrong 1256042827 M * cehteh and iirc there was some provision to extend inodes only 1256042839 M * cehteh i dont care, i never run into inode limits yet 1256042866 M * cehteh /dev/md3 24223744 1317893 22905851 6% /data 1256042872 M * cehteh i doubt i will :) 1256042901 A * _Shiva_ has just yesterday.. on rootfs.. duh 1256043756 M * mnemoc Bertl: resize2fs keeps the number of inodes... and when i asked in their irc channel they said the only way was mkfs 1256043928 M * Bertl I didn't say that resize2fs does what you want, I said, it moves stuff around to place the inode tables/superblock backups 1256043944 M * Bertl the very same would work for changing the #inodes ... 1256043962 M * mnemoc yes, makes sense 1256044075 J * davidkarban_ ~david@199.123.broadband11.iol.cz 1256044075 Q * davidkarban Read error: Connection reset by peer 1256045056 P * Mark17 1256045150 Q * Pazzo Remote host closed the connection 1256045235 J * Shinsaku ~Shinsaku@chello089076140236.chello.pl 1256045453 N * elesouef_ elesouef 1256045969 M * Hollow Bertl: i have updated helios and had some problems with apache, but i have figured it out now and the setup is a bit cleaner now. it is also monitored with nagios now. if you come across any problems please drop a note 1256046041 Q * geb Ping timeout: 480 seconds 1256046172 M * Bertl Hollow: k, will do, thanks! 1256046234 J * Pazzo ~ugelt@reserved-225136.rol.raiffeisen.net 1256046596 M * yang Bertl: is it possible that kernel doesn't boot properly becouse of the enabled klogd ? 1256046609 M * yang i know that guests will not boot becouse of that 1256046620 M * yang not sure about the host 1256046870 M * Bertl very unlikely, the host still has kmsg after all 1256046879 M * Bertl what exactly are you seeing? 1256047254 M * yang I don't see anything, I am just breaking my head trying to boot the machine - I can only see rescue mode 1256047271 M * yang it boots well with an ordinary kernel 1256047274 M * Bertl what about the serial console? 1256047283 M * yang I can't access it (datacenter) 1256047306 M * Bertl what datacenter/setup is that ... 2009 and no remote console? 1256047319 M * yang it has only rescue mode 1256047350 M * Bertl well, probably you forgot some driver or misconfigured the initrd 1256047364 M * Bertl (initramfs actually) 1256047449 M * yang http://pastebin.com/m494c1689 1256047517 M * yang now I should probably run also "update-initramfs" 1256047518 M * ghislainocfs2 bertl: most hoster bill the console 30/40€/month so on a marquet where price is important most do not take it (i do not) 1256047547 M * yang ghislainocfs2: that is the price of a monthly hosting fee 1256047578 M * ghislainocfs2 yang: see why i do not take it :) 1256047579 M * Bertl ghislainocfs2: but, if you have two machines hosted, you can get a serial console for free 1256047625 Q * davidkarban_ Quit: Ex-Chat 1256047628 M * ghislainocfs2 bertl : if i take a dedicated box whith hosting company i cannot ask them to serial link them, at least not with the one i know :) 1256047656 M * ghislainocfs2 bertl: high level enterprise hosting do it but i work with little web shop they cannot afford the big hosting guys 1256047680 M * yang Bertl: update-initramfs -k linux-image-2.6.26-1-vserver-686 doesn't seem to be working 1256047732 M * Bertl did you try? usually providers with dedicated machines have no problem with daisy-chaining serial lines 1256047754 M * Bertl yang: debian issue, check with #debian 1256048629 Q * Pazzo Quit: Bye! 1256048652 M * yang Bertl: The lilo will spit out 1256048655 M * yang Warning: LBA32 addressing assumed 1256048655 M * yang Fatal: Setup length exceeds 31 maximum; kernel setup will overwrite boot loader 1256048668 M * yang becouse of a long name 1256048673 M * yang then it cannot boot 1256048789 M * yang Bertl: http://pastebin.com/m515e006e is the output of lilo.conf 1256048975 M * yang Bertl: just see this - http://pastebin.com/m566a7762 1256049008 M * yang I think this should suffice, for anything extra, I would really need a serial console to see why kernel stops 1256049367 Q * balbir_ Ping timeout: 480 seconds 1256049800 Q * derjohn_mob Ping timeout: 480 seconds 1256049821 J * derjohn_mob ~aj@80.85.196.112 1256050324 M * yang Bertl: still refuses to boot 1256050619 Q * imcsk8 Quit: This computer has gone to sleep 1256050647 Q * jrdnyquist Ping timeout: 480 seconds 1256050660 M * PowerKe yang: did it boot before? 1256050800 Q * sharkjaw Remote host closed the connection 1256050846 J * jrdnyquist ~jrdnyquis@slayer.caro.net 1256051168 M * yang PowerKe: I can't recall it had a preety long uptime 1256051215 M * PowerKe did you try rebooting before making changes? 1256051378 M * Bertl yang: can't be that a long time when it was 2.6.26-1 :) 1256051751 Q * jrdnyquist Quit: bye 1256052194 J * grharry ~root@ppp-94-65-253-220.home.otenet.gr 1256052274 M * grharry hi I have an image tar of a vhost ( I had to reinstall all ) how am I suppose to re-install it, thanks ?? 1256052287 M * fback good evening! 1256052297 M * grharry hi 1256052363 M * Bertl grharry: best use the template build method 1256052501 M * grharry Bertl: Is it that hard to bring it back to life ?? ( the tgz image ) ?? 1256052529 M * Bertl not at all, shouldn't be a problem .. 1256052563 M * grharry Bertl: A howto avail ?? 1256052591 M * Bertl it's a single build command line ... should be trivial 1256052662 M * grharry So the tar image serves as a template ... ok ... what concerns me is the /dev/* stuff included in the image !! will it be ok ? 1256052704 M * Bertl was it from an existing guest (the tar) or from a real machine? 1256052712 M * grharry guest 1256052734 M * Bertl then it should be fine, no? 1256052790 M * grharry tnx .... by the way .. do I have to specify the -d option ?? 1256052836 M * Bertl it might help, specifically with cleanups and similar 1256052876 J * geb ~geb@earth.gebura.eu.org 1256052920 J * mrfree ~mrfree@host1-89-static.40-88-b.business.telecomitalia.it 1256053135 J * grharry1 ~root@ppp-94-65-233-232.home.otenet.gr 1256053295 Q * grharry Ping timeout: 480 seconds 1256053640 Q * derjohn_mob Ping timeout: 480 seconds 1256053751 M * yang well the datacenter does not provide a serial console, so how the heck can I boot into the system now 1256053774 M * yang 2.6.27 boots up, while I cannot get -vserver to boot 1256053846 M * pmjdebruijn 2.6.27.37? 1256053855 M * pmjdebruijn 2.6.27.37-vserver I mean? 1256053863 M * pmjdebruijn yang: sure you kernel config is ok? 1256053875 M * yang nope 1256053883 M * yang 2.6.27 plain (non vserver) 1256053884 M * pmjdebruijn yang: is your disk controller built into the kernel... or does your ramdisk include the disk controller driver? 1256053890 M * pmjdebruijn yang: which -vserver then? 1256053954 M * yang http://pastebin.com/m6f2c8762 1256053971 M * pmjdebruijn Debian stock? 1256053974 M * pmjdebruijn ugh 1256053977 M * pmjdebruijn build your own 1256053989 M * pmjdebruijn :s 1256054000 M * pmjdebruijn LILO 1256054014 M * yang yes, its lilo, since grub 2 refused to work 1256054018 M * pmjdebruijn I think stock -vserver in Debian is severely broken 1256054029 M * pmjdebruijn yang: Lenny doesn't have GRUB2 does it? 1256054030 M * yang but it should still boot 1256054042 M * yang no, I am using squeeze 1256054045 M * pmjdebruijn oh 1256054046 M * pmjdebruijn :s 1256054049 M * pmjdebruijn seriously... 1256054050 M * pmjdebruijn build your own 1256054069 M * pmjdebruijn 2.6.27.37 works great, with the 2.6.27.33 patch (except you need to change the EXTRAVERSION) 1256054179 Q * grharry1 Read error: Connection reset by peer 1256054194 J * grharry ~root@ppp-94-65-251-239.home.otenet.gr 1256054423 M * Bertl grub refused to work? 1256054456 M * yang well, in the same way as lilo...it refused to boot the debian-vserver kernel 1256054480 M * yang I cannot see at which point the boot stops 1256054525 M * Bertl you could try with network console 1256054559 M * yang ? 1256054577 Q * sardyno Ping timeout: 480 seconds 1256054578 J * grharry1 ~root@ppp-94-67-122-91.home.otenet.gr 1256054623 M * yang what do you mean ( a remote output via the network) 1256054625 M * Bertl well, build your own (I presume the debian kernel doesn't have it enabled) kernel, and use the network console to send the output to you 1256054633 Q * mrfree Ping timeout: 480 seconds 1256054633 M * yang good idea 1256054666 M * Bertl might need some testing and configuration, so don't give up easily if you do not get anything right away 1256054680 M * yang do you know the name of the package? 1256054681 M * Bertl also read the kernel documentation for the netconsole and google for examples 1256054694 M * Bertl it is a kernel feature, no package 1256054765 Q * grharry Ping timeout: 480 seconds 1256054827 Q * Shinsaku 1256054849 M * geb geb: see dmesg 1256054853 M * geb tks :) 1256054867 M * geb EXT3-fs: dm-2: tagging not permitted on remount. :( 1256054969 M * Bertl FYI: the reason for that is quite simple, the tagging happens when inodes get instantiated, enabling that while the filesystem is life would result in inodes without tag 1256054996 M * Bertl when those inodes get written back, a guest might gain untagged inodes 1256055015 J * bonbons ~bonbons@2001:960:7ab:0:2c0:9fff:fe2d:39d 1256055045 M * Bertl but there is no need to reboot, as you did, unmounting and mounting (with tagging enabled) is sufficient 1256055049 M * yang Bertl: do you think this would be worth trying - http://ser2net.sourceforge.net/ 1256055098 M * Bertl only if you run it on a second machine :) 1256055128 M * Bertl you want to get the console logs, at that time userspace is not active (kernel boot) 1256055142 M * yang Bertl: well yes, I can run 2 parts server-side on the machine with kernel issues and client side on the other node 1256055164 M * Bertl with a serial line between them? 1256055171 M * geb ok thanks Bertl for the explanation :) the problem is that i don't want to stop vservers ... but i'll have to do :) 1256055176 M * yang Bertl: well that would still require me to create a new kernel with netconsole option built ? 1256055208 M * Bertl geb: to enable tagging, yes ... 1256055235 M * Bertl yang: netconsole is your best option, unless you have a remote (serial) console 1256055549 M * yang maybe adding "netconsole" into /etc/modules would trigger it in the start, if it gets to the userspace 1256055674 M * yang I found also this reference - http://www.cyberciti.biz/tips/linux-netconsole-log-management-tutorial.html 1256056012 M * yang Bertl: how would you be able to see the logs before networking starts? 1256056135 M * Bertl you need to configure the kernel to start networking early, like e.g. for nfs-root, with network card drivers compiled in 1256056184 J * Shinsaku ~Shinsaku@chello089076140236.chello.pl 1256056437 J * grharry ~root@ppp-94-67-116-91.home.otenet.gr 1256056553 Q * grharry1 Ping timeout: 480 seconds 1256056757 J * scientes ~scientes@174-21-161-91.tukw.qwest.net 1256056898 M * PowerKe yang: did you configure the -vserver kernel from scratch, or did you start by copying the .config and using make oldconfig? 1256056921 J * dna_ ~dna@p54BC9D7F.dip0.t-ipconnect.de 1256057590 J * derjohn_mob ~aj@c192190.adsl.hansenet.de 1256058141 M * yang PowerKe: I will configure the kernel tomorow, i used a stock debian kernel 1256058404 J * Pazzo ~ugelt@reserved-225136.rol.raiffeisen.net 1256058750 Q * Pazzo Quit: Bye! 1256059155 J * imcsk8 ~ichavero@148.229.1.11 1256059410 Q * grharry Ping timeout: 480 seconds 1256059598 Q * thierryp Quit: ciao folks 1256059988 J * BenG ~bengreen@94-169-110-10.cable.ubr22.aztw.blueyonder.co.uk 1256061016 Q * derjohn_mob Ping timeout: 480 seconds 1256061584 M * geb hum, i have a got a stupid question: what is the purpose of the .default dir in /etc/vserver ? 1256061608 M * geb because, i added some files in, they are not copied when i create a new vserver, nor they are read 1256061642 M * Bertl hehe, no, that dir is used to provide defaults for all guests (not at creation time) 1256061663 M * Bertl check the 'flower page' for details 1256061746 M * geb yeah , but so it looks like this dir is not read when i start some vservers (see http://paste.linux-vserver.org/13538 ) 1256061845 M * geb base was just created, it hasn't any cflag file or rlimits dir, so why it doesn't use values from .defaults ? 1256061854 M * Bertl sure you checked with the 'flower page' because I do not see any cflags there nor rlimits? 1256062023 J * derjohn_mob ~aj@c192190.adsl.hansenet.de 1256062087 M * geb hum you are right... i am "a fool for a minute" (not sure if it is only for a minute :) 1256062134 M * Bertl np, but maybe post a feature request for adding a .skel (or similar) dir with defaults for newly created guests? 1256062334 M * PowerKe yang: using make oldconfig with your previous .config should be safe with regard to hardware support 1256062413 M * geb daniel_hozac, any opinion about a .skel ? :) 1256062414 M * PowerKe If you don't have the .config for your current kernel, it hopefully has /proc/config.gz support and you can copy it from there. 1256062444 M * geb PowerKe, yang, or downloading debian's .config ( http://merkel.debian.org/~jurij/ ) 1256062476 M * Bertl PowerKe: he is using a vanilla Debian kernel, no point in taking the .config from there, when he wants to activate netconsole ... 1256062531 M * PowerKe If he uses the kernel that can boot his system, he should have a good chance using it for his new kernel to have it boot 1256062572 M * PowerKe Starting with something else always has more risk of choosing a wrong network or disk driver 1256062645 M * Bertl I beg to defer .. IMHO the most likely cause for this is a change from /dev/hda to /dev/sda (or similar) and thus, a misconfigured initramfs ... no chance that the old (probably debian too) config will make the new kernel work then 1256064313 J * turin ~turin@213.37.42.133.static.user.ono.com 1256064395 M * Bertl welcome turin! 1256064404 M * turin hi everybody 1256064462 M * turin i'm looking for information about the status of the next stable release of vserver 1256064497 M * turin the web seems not very updated 1256064506 M * Bertl stabilization phase has started, we are on the way to a new devel release, which will, after extensive testing result in a stable release 1256064534 M * Bertl turin: the wiki? it's updated every hour :) 1256064544 Q * Shinsaku 1256064599 M * turin :) ups 1256064608 M * BenG Bertl, any idea of the time scale for those happenings? Development Release and Stable Release? 1256064642 M * Bertl I'm almost done with the devel release, the ipv6 stuff should be the last missing piece 1256064656 M * BenG wow 1256064660 M * BenG exciting times 1256064677 M * Bertl after that, we start with a big testing/code review round 1256064735 M * Bertl turin: where are/were you looking? 1256064756 M * turin i'm a liitle newbee about this, and maybe this is a stupid question... but how far is vserver from getting in the mainline kernel 1256064801 M * Bertl regarding stupid: see topic, regarding mainline: as far as possible, yet quite near :) 1256064879 M * turin thanks :) i am just interested in the status of the project, i have some servers with debian and vservers 1256064930 M * turin and the version of vservers in debian lenny have some problems, looking for newer releases i found the devel for some time now 1256064966 M * turin excuse my english, is not as perfect as i would like 1256065075 M * turin another question i would like to ask, is "live migration" in mind? 1256065120 M * turin i see openvz have this feature, and i would like to see in vserver but dont have the knowlegde for this :) 1256065143 M * Bertl yeah, we know that the debian kernels have some issues, specifically the 2.6.26 branch, this is a debian specific problem, if you build your own kernel/tools the debian host system is fine 1256065188 M * Bertl regarding live-migration: that is something we consider 'marketing feature' and as we like to keep things clean, simple and performant, we do not waste time on that 1256065244 M * turin ok, that's nice, performance in vservers is practicaly native 1256065283 M * Bertl there are only three scenarios I can imaine where one wants to migrate ... a) when you have a scheduled downtime (hardware maintainance), b) when you want to load balance on a per-guest basis, c) when you have a near failure scenario and want to move stuff away 1256065319 M * Bertl a) can certainly be well handled by putting the Linux-VServer kernel into a VM itself (e.g. kvm or xen) 1256065367 M * turin actually i have 2 machines in a cluster with heartbeat (and a nas for the vservers filesystem) and is working perfectly (except for some issues with heartbeat) 1256065368 M * Bertl b) when you do balancing you usually do it to gain performance by utilizing resources optimally, now, migration itself is quite a resource hog (moving a lot of data during migration) 1256065407 M * turin ok, i see 1256065411 M * Bertl c) well, that could be handled like a) but IMHO any HA setup, as for example your setup, works better, is simpler and thus more reliable 1256065459 M * turin :) thanks for the explanation 1256065468 M * Bertl you're welcome! 1256065742 J * sardyno ~me@pool-173-75-5-88.pitbpa.fios.verizon.net 1256065826 M * turin thanks very much for the experience :) 1256065832 M * turin bye 1256065842 M * Bertl feel free to hang around 1256065885 M * turin thanks 1256065888 Q * turin Quit: Konversation terminated! 1256067602 Q * BenG Quit: I Leave 1256068700 A * Alteisen thinks, that linux-vserver is generally a good idea and uses vservers wherever it is possible. even within a vmware guest linux. 1256068718 A * Bertl appreciates that :) 1256068725 M * Bertl nap attack ... bbl 1256068730 N * Bertl Bertl_zZ 1256068766 M * Alteisen good idea, nap attack 1256068885 M * geb http://paste.linux-vserver.org/13539 1256068932 M * geb fell free to use/comment :) 1256069050 M * Alteisen geb: mine has more parameters... -> http://a.ctdo.de/wiki/doku.php?id=linux-vserver 1256069103 M * Alteisen geb: but i like your simple script - does it work? 1256069127 M * geb yeah i just tested it 1256069160 M * geb but i figure that clone don't copy caps etc so i have to add it 1256069319 M * Alteisen i am still thinking about adding another script to my one, that simply copies the template. but installing the template takes only a few minutes, which could be faster if i had a apt-proxy, and also how often do i create a new vserver.... 1256069421 M * geb i think that it is a good option to have a template and build -m clone 1256069450 M * geb (it saves bandwidth etc) 1256069462 M * Alteisen yep, for backup and testing purposes 1256070037 M * tam Anyone around? I have a vserver on a busy host that stopped responding to ssh, so I tried to vserver xyz enter, and it hung... i vserver xyz stop'd and that worked, but now it hangs when I try to start it. vserver -v xyz start outputs 3 lines, added network context, added ip address, added security context then hangs. 1256070065 M * tam Other running guests I'm able to vserver enter on the same host 1256070106 Q * ghislainocfs2 Ping timeout: 480 seconds 1256070126 M * geb try to vkill it ? 1256070167 M * geb vkill --xid -s 15; sleep 2; vkill --xid -s 9 1256070175 M * daniel_hozac tam: vserver --debug xyz start 1256070185 M * daniel_hozac tam: but it doesn't sound like there's anything vserver related causing it to hang. 1256070232 M * tam it's hanging on ionice 1256070262 M * tam Hm! 1256070276 M * daniel_hozac no, that's the command that hands off control to your guest. 1256070316 M * tam hmmm. 1256070325 M * tam I'll poke around the startup stuff inside the guest 1256070341 M * daniel_hozac try --debug-sysv 1256070408 M * tam That hangs with no output 1256070778 M * tam I put an 'echo 1' in the guests /etc/rc.d/rc 1256070782 M * tam and 1 never gets echo'd 1256070802 M * tam so it's hanging before the rc script runs 1256070805 A * tam is puzzled 1256070817 M * daniel_hozac what does the process list look like when vserver ... start hangs? 1256070929 J * dna__ ~dna@p54BC9D7F.dip0.t-ipconnect.de 1256071215 Q * dna_ Ping timeout: 480 seconds 1256071266 M * tam daniel- I'll get a capture in a couple minutes 1256071321 Q * bonbons Quit: Leaving 1256071754 M * tam daniel- http://paste.linux-vserver.org/13542 1256071807 M * daniel_hozac you actually want a vps faux 1256071830 M * daniel_hozac with however many w's your terminal needs. 1256071903 J * Pazzo ~ugelt@sadsl-246059.rol.raiffeisen.net 1256071904 M * tam roger 1256071912 M * tam It's actually starting 1256071916 M * tam just very slowly 1256071919 M * tam hm 1256072550 M * Alteisen killing me softly... 1256073342 M * tam This is really weird. More details... 1256073393 M * tam this is a redhat 5.2 x86_64 host... with my vserver mounted on an iscsi volume mounted as /vservers/ 1256073408 M * tam IO inside the vserver once it boots is very very very slow 1256073415 M * tam but outside the vserver io is fine 1256073422 M * tam i've fsck'd the volume, it's clean 1256073430 M * tam unmounted, remounted the iscsi 1256073438 M * tam start the vserver, and it crawls 1256074373 J * blues_ ~blues@ccz129.neoplus.adsl.tpnet.pl 1256074457 M * tam ok, I just proved it has nothing to do with my iscsi, i copied the volume onto local disk and it's still dog slow to start up 1256074493 Q * blues Ping timeout: 480 seconds 1256074561 M * tam daniel_hozac- I updated my pastebin with the vps faux http://paste.linux-vserver.org/13543 1256075184 M * tam this is so puzzling! 1256075216 M * daniel_hozac have you configured the hard CPU scheduler for that context? 1256075261 M * tam no 1256075268 M * tam Here's my latest steps 1256075281 M * tam I restored an iscsi snapshot of the guest from last night before it started having problems 1256075284 M * tam same startup issue 1256075299 M * tam I copied the /etc/vservers folder for it out of the way, and copied a vserver folder that works and changed the softlinks.... 1256075304 M * tam and it still hangs on startup 1256075359 M * tam the other vserver i copied the /etc/vservers folder from starts and stops just fine 1256076782 Q * lucrus Remote host closed the connection 1256076881 N * DoberMann DoberMann[ZZZzzz] 1256077606 J * BenG ~bengreen@94-169-110-10.cable.ubr22.aztw.blueyonder.co.uk 1256077646 Q * BenG 1256077845 Q * imcsk8 Quit: Leaving 1256079549 Q * dna__ Quit: Verlassend 1256080033 N * Bertl_zZ Bertl 1256080045 M * Bertl back now ... 1256080103 M * tam hey Bertl 1256080245 M * Bertl so you have I/O performance issues over iSCSI? 1256080527 M * tam Well, it happens even when I copy the contents of the iscsi volume onto the local disk 1256080542 M * tam I think that proves it's not with the iscsi layer 1256080560 M * tam it's got to be init corruption or something 1256080584 M * Bertl good point, so, the host is busy? what does 'vmstat 1' output? 1256080622 M * tam 0 0 1804336 1188840 410612 8398748 0 0 48 144 1725 2487 1 0 99 1 0 1256080629 M * tam that's a crummy paste 1256080639 M * tam 19:17:16 up 59 days, 3:07, 5 users, load average: 0.08, 0.17, 0.23 1256080645 M * tam it's not very busy, has plenty of memory 1256080658 M * tam it's a dual quad core 2.66 1256080669 M * Bertl okay, what about 'iostat 10' ? 1256080678 M * tam the rest of the 26 vservers on it are fine 1256080714 M * tam 0 0 1804336 1180640 412848 8402100 0 0 45 640 3603 3533 1 1 95 2 0 1256080724 M * tam oops i did vmstat 10 1256080725 M * tam sorry 1256080733 M * tam I don't have iostat :) 1256080743 M * tam centos 5.2 1256080777 M * Bertl well, install it ... and please use paste.linux-vserver.org for everything longer than 3 lines :) 1256080976 M * tam http://paste.linux-vserver.org/13544 1256081022 M * Bertl looks harmless too ... 1256081040 M * tam yeah 1256081058 M * Bertl do you have any ionice values set in the guest config? 1256081099 M * tam no 1256081137 M * Bertl what about the nice level or the priority bias? 1256081153 Q * Pazzo Ping timeout: 480 seconds 1256081153 M * tam I don't mess with any of that on any of my vservers 1256081214 M * tam It doesn't *seem* like a resource issue 1256081234 M * tam considering other guests on the same host start and stop fine 1256081258 M * Bertl okay, would it be possible to tar the guest up (or dump it) and upload the compressed image somewhere where I could grab it? 1256081292 M * Bertl (you could clone the guest, and strip it down first :) 1256081319 M * tam yeah I can do that 1256081359 M * Bertl okay, then let's do it ... I'l have a look at the guest startup on my test machine 1256082778 M * tam Just to be sure, I extracted that tar and tried to start it with the same results 1256083061 M * tam for fun, I copied it to another host and tried to start it, same issue 1256083188 M * Bertl good!