[00:00] so you could load balance mail out to a pair of servers all delivering to the same nfs spool [00:00] mrbawb: i want those moved as well [00:00] I am not aware of anything in existance that will do that [00:00] but obviously the demand is there :) [00:01] mrbawb: i'm not talking about in existance, i'm looking for something that i can do as my master project [00:01] ah [00:01] mrbawb: and i was thinking about something that would allow me to move a vserver to a different physical machine, and reroute the network packets somehow [00:02] re-routing the packets isn't a problem [00:02] mrbawb: thinking a iptables that delayed/froze the packets, move of vserver, restart packets [00:02] that is actually what you could do with LVS ... [00:02] it's the state of the applications [00:02] tcp will retransmit [00:02] mrbawb: but move of vserver takes time [00:02] you don't need to queue the packets [00:02] true [00:02] mrbawb: what about new connections while i move it ? [00:03] yes, so you need to start it before bringing down the old vserver :) [00:03] mrbawb: i want to catch those as well [00:03] just rerouting isn't a good solution ... [00:03] new connections can't be handled at the new vserver till all the staet is moved over [00:03] Bertl: i know, i have to move the future packets as well [00:03] i know [00:04] but think again like load balancing ... [00:04] assume that you have 2 hosts and you are able to 'adjust' how much goes to A and B [00:05] adjusting this to 100%/0% is the 'start' situation ... 0%/100% your 'moved' version ... [00:05] as we know that load balancing works, and will not hurt existing connections ... that's what you want ... [00:06] just keep in mind that anything you do will require lots of kernel modifications :) [00:06] mrbawb: i know [00:07] Bertl: oh, i get your point [00:07] i was thinking of putting the userland processes to sleep [00:07] and restarting them on the new computer, and then just open for the packets [00:07] yeah [00:08] freeze everything, move the state, unfreeze [00:08] and since moving might take time [00:08] that won't allow migration ... [00:08] why not ? [00:09] i would catch any packets for new connections and then send them when the vserver runs the new place [00:09] because you have to make sure that everything, down to the smallest detail is the same when the applications have been moved ... [00:09] that's part of the state :) [00:09] JonB: just simulate a network failure by dropping the packets :) [00:09] Bertl: there is already a patch that makes you suspend to disk [00:09] mrbawb: *grin* ofc [00:10] Bertl: i just want to suspend a vserver only [00:10] well, then the first part would be the ability to migrate a task (including files/ports/sockets) to another machine ... [00:10] if that works, what seems not easy to me (not even *Mosix does this) ... [00:11] you can address the 'simple' packet rerouting stuff ... [00:11] Bertl: yes, and since the vserver uses a 192.168.x.y ip, then giving it the same files/ports/sockets would be easy [00:11] and the lo device belongs to the specific vserver, so you are sure that all ports are free [00:11] same with the eth device [00:13] you don't know much about files and sockets yet, right? [00:14] Bertl: is it that obvious ;-P [00:14] Bertl: they have a unique number, right ? [00:25] Last message repeated 1 time(s). [00:25] well, if that would be the only thing ... [00:26] netrose (~john877@CC3-24.171.21.47.charter-stl.com) joined #vserver. [00:26] you have partially mapped parts for each file, and structures/buffers/inodes ... etc ... [00:26] not to talk about file descriptors and that you have to match/position them for the task ... [00:27] Bertl: suppose that the files are not mapped into memory ? [00:27] on sockets there are bindings and packet queues ... and such stuff ... the task might be waiting on such things via select ... for example [00:28] Bertl: untill it is needed on the new host [00:28] the files could be mapped, in which case you would have to do the exact same mapping ... [00:28] position... isnt that saved in the program ? [00:28] Bertl: but yeah i get your point [00:29] have a look at the 'kernel' structures storing this information, and think about how you could move that to a different machine ... (I don't say it's undoable, it's just much work to do) [00:30] okay, have to leave now ... will be back later ... [00:30] Bertl: work for 2 people for 6 months ? [00:30] Bertl: aka a master project ? [00:30] i have to write a repport about it too [00:31] depends on the people (Linus and Al Viro, maybe ;) [00:31] Nick change: Bertl -> Bertl_oO [00:31] ouch [01:12] serving (~serving@213.186.191.194) joined #vserver. [01:16] mhm [01:17] i found a problem with an iptables accounting programm [01:17] maharaja: wong channel ? [01:17] it uses pthreads and somehow things it cannot be initialised inside the vserver [01:17] it perfectly works outside [01:17] are there any capabilities i need to have inside the vserver for a working iptables setup? [01:18] iptables currently isn't virtualized per vserver [01:18] jonb: nope ;) [01:18] so what does that mean? [01:18] so you'd either need to grant access to all of iptables or none [01:18] and it's at none right now [01:18] i c [01:19] thank you :) [01:19] probably just a capability to grant will give the vserver access to it [01:20] it will now or it will in the future? [01:21] CAP_NET_ADMIN should do it [01:21] like: am i able to accomplish it somehow with v1.1.6 [01:21] or do i have to wait/hack it by myself [01:21] that has a few other side effects [01:22] morning [01:22] worth considering if you want to give a vserver access to mess with the network config [01:23] which capability is it? [01:23] CAP_NET_ADMIN [01:24] and what are the sideeffects? [01:24] the vserver is allowed to change network settings, iptables [01:26] affecting all the processes on the whole box [01:31] ic [01:31] thnx :) [01:42] netrose (~john877@CC3-24.171.21.47.charter-stl.com) left irc: Ping timeout: 480 seconds [01:43] kestrelw (~athomas@o2rosock0a.optus.net.au) left irc: Quit: brb, apt-get dist-upgrade time [02:23] Bawb (abob@swordfish.drown.org) joined #vserver. [02:23] mrbawb (abob@sparky.dok.org) left irc: Quit: -> bawb [02:23] Nick change: Bawb -> MrBawb [02:46] kestrelw (~athomas@o2rosock0a.optus.net.au) joined #vserver. [03:05] Nick change: Bertl_oO -> Bertl [03:05] cool, vskel is quite useful [03:06] MrBawb: what does it do ? [03:07] JonB: makes a copy of the parent server as a skeleton for new vservers [03:07] @dan thanks for the output! [03:07] Bertl: np [03:07] @dan you are evaluating vskel? [03:07] yup [03:07] I approve :) [03:08] MrBawb: i prefer debiannewserver.sh from mr. Sladen [03:08] JonB: ah, does it handle unification? [03:08] @dan please verify for me that the 'resulting' server has sane/secure /dev entries ... [03:08] MrBawb: no [03:09] $ ls /dev | wc -l [03:09] 11 [03:09] MrBawb: but, you have to handle unification at a later point when you upgrade [03:09] full, log, null, ptmx, pts, random, reboot, shm, tty, urandom, zero [03:09] JonB: ah, ok [03:10] I like the idea of just cloning an on-disk install [03:10] MrBawb: how would you do that ? [03:10] hmm reboot and shm is? [03:10] shm is a dir [03:10] reboot is a socket from rebootmgr [03:10] I added shm [03:10] ahh okay ... [03:11] so people still use rebootmgr ... [03:11] JonB: it takes the parent install and copies it into /vserver/.skel/[name] [03:11] MrBawb: copies or hardlinks ? [03:11] JonB: copies [03:11] the vserver installs use hardlinks [03:12] MrBawb: okay, i thought it hardlinks [03:12] so you have one copy for your parent server and one copy for all of your unified vservers [03:12] Bertl: is there a replacement for rebootmgr? [03:12] MrBawb: what when you there is a security update? [03:13] JonB: that's tricky [03:13] you update the skel and all the vservers [03:13] and then you have to re-unify them [03:13] MrBawb: why not just run that script continuesly ? [03:14] @dan, sure, people bugged me that they want the userspace helper replacement ... I added it in vs1.1.4 or so ... there was a posting how to use it ... [03:14] as in, install the vserver again? [03:14] MrBawb: especialy if it is customers that install the security updates them selves [03:14] Bertl: ah, ok. I'll go look for it [03:14] MrBawb: no, the script that unifies [03:14] oh, the unification process is disk-intensive [03:15] running it every night or so would be okay if you didn't have that much content [03:16] i would seperate content and system [03:16] like /vserver//var/www is not processed [03:16] same with log [03:16] and ... [03:16] but /usr is [03:17] yeah, that'd be a good policy [03:19] MrBawb: i suppose that there is a difference between whether or not you expect the vservers to be customers that updates them, or yours [03:19] if they were yours, you would know when you needed to run the re-unify script [03:20] anyway, i think i'll go to bed [03:20] JonB (~jon@kg188.kollegiegaarden.dk) left irc: Quit: zzz [03:34] netrose (~john877@CC3-24.171.21.47.charter-stl.com) joined #vserver. [03:39] Nick change: apw -> apw-zzz [04:50] Nick change: Bertl -> Bertl_zZ [04:56] grepmaster-afk (~chatzilla@66-101-59-73.oplnk.net) left irc: Quit: ChatZilla 0.8.31 [Mozilla rv:1.4/20030813] [07:10] iceberg (~infowolfe@pcp04891550pcs.frnkmd01.md.comcast.net) left irc: Quit: Trillian (http://www.ceruleanstudios.com) [07:10] iceberg (~infowolfe@pcp04891550pcs.frnkmd01.md.comcast.net) joined #vserver. [07:11] iceberg (~infowolfe@pcp04891550pcs.frnkmd01.md.comcast.net) left irc: Client Quit [07:13] iceberg (~infowolfe@pcp04891550pcs.frnkmd01.md.comcast.net) joined #vserver. [07:50] [1]James (~James@ip68-96-180-27.lv.lv.cox.net) joined #vserver. [07:50] <[1]James> hello all [08:11] kestrel (~athomas@dialup51.optus.net.au) left irc: Ping timeout: 480 seconds [08:22] micah (micah@micha.hampshire.edu) left irc: Ping timeout: 500 seconds [08:31] iceberg (~infowolfe@pcp04891550pcs.frnkmd01.md.comcast.net) got netsplit. [08:31] MrBawb (abob@swordfish.drown.org) got netsplit. [08:31] sannes (ace@sannes.org) got netsplit. [08:31] MedivhWrk (ck@netops.multimedia-centrum.de) got netsplit. [08:31] Medivh (ck@62.93.217.199) got netsplit. [08:31] Zoiah (Zoiah@matryoshka.zoiah.net) got netsplit. [08:32] iceberg (~infowolfe@pcp04891550pcs.frnkmd01.md.comcast.net) returned to #vserver. [08:32] MrBawb (abob@swordfish.drown.org) returned to #vserver. [08:32] Medivh (ck@62.93.217.199) returned to #vserver. [08:32] MedivhWrk (ck@netops.multimedia-centrum.de) returned to #vserver. [08:32] Zoiah (Zoiah@matryoshka.zoiah.net) returned to #vserver. [08:32] sannes (ace@sannes.org) returned to #vserver. [08:53] sannes (ace@sannes.org) left irc: Ping timeout: 480 seconds [08:53] sannes (ace@sannes.org) joined #vserver. [09:00] Zoiah (Zoiah@matryoshka.zoiah.net) got netsplit. [09:00] MedivhWrk (ck@netops.multimedia-centrum.de) got netsplit. [09:00] Medivh (ck@62.93.217.199) got netsplit. [09:00] MrBawb (abob@swordfish.drown.org) got netsplit. [09:00] iceberg (~infowolfe@pcp04891550pcs.frnkmd01.md.comcast.net) got netsplit. [09:00] serving (~serving@213.186.191.194) got netsplit. [09:00] apw-zzz (~apw@212.104.150.41) got netsplit. [09:00] CosmicRay (~jgoerzen@glockenspiel.complete.org) got netsplit. [09:00] click (click@gonnamakeyou.com) got netsplit. [09:00] say (~say@212.86.243.154) got netsplit. [09:00] maharaja (maharaja@ipax.tk) got netsplit. [09:00] matta (matta@tektonic.net) got netsplit. [09:00] linas (~linas@h-67-100-217-179.HSTQTX02.covad.net) got netsplit. [09:00] [1]James (~James@ip68-96-180-27.lv.lv.cox.net) got netsplit. [09:00] netrose (~john877@CC3-24.171.21.47.charter-stl.com) got netsplit. [09:00] Bertl_zZ (~herbert@MAIL.13thfloor.at) got netsplit. [09:00] sannes (ace@sannes.org) got netsplit. [09:00] kestrelw (~athomas@o2rosock0a.optus.net.au) got netsplit. [09:00] mcp (~hightower@wolk-project.de) got netsplit. [09:00] ensc (~ircensc@ultra.csn.tu-chemnitz.de) got netsplit. [09:01] click (click@gonnamakeyou.com) returned to #vserver. [09:01] say (~say@212.86.243.154) returned to #vserver. [09:01] maharaja (maharaja@ipax.tk) returned to #vserver. [09:01] linas (~linas@h-67-100-217-179.HSTQTX02.covad.net) returned to #vserver. [09:01] matta (matta@tektonic.net) returned to #vserver. [09:01] Bertl_zZ (~herbert@MAIL.13thfloor.at) returned to #vserver. [09:01] #vserver: mode change '+o Bertl_zZ ' by arion.oftc.net [09:01] netrose (~john877@CC3-24.171.21.47.charter-stl.com) returned to #vserver. [09:01] Topic changed on #vserver by ChanServ!services@services.oftc.net: http://linux-vserver.org/ || latest stable 1.00, devel 1.1.6 [09:01] iceberg (~infowolfe@pcp04891550pcs.frnkmd01.md.comcast.net) returned to #vserver. [09:01] MrBawb (abob@swordfish.drown.org) returned to #vserver. [09:01] Medivh (ck@62.93.217.199) returned to #vserver. [09:01] MedivhWrk (ck@netops.multimedia-centrum.de) returned to #vserver. [09:01] Zoiah (Zoiah@matryoshka.zoiah.net) returned to #vserver. [09:01] Topic changed on #vserver by !arion.oftc.net: http://linux-vserver.org/ || latest stable 1.00, devel 1.1.6 [09:01] [1]James (~James@ip68-96-180-27.lv.lv.cox.net) returned to #vserver. [09:01] #vserver: mode change '-o Bertl_zZ' by ChanServ!services@services.oftc.net [09:02] serving (~serving@213.186.191.194) returned to #vserver. [09:02] apw-zzz (~apw@212.104.150.41) returned to #vserver. [09:02] CosmicRay (~jgoerzen@glockenspiel.complete.org) returned to #vserver. [09:02] sannes (ace@sannes.org) returned to #vserver. [09:02] kestrelw (~athomas@o2rosock0a.optus.net.au) returned to #vserver. [09:02] mcp (~hightower@wolk-project.de) returned to #vserver. [09:02] ensc (~ircensc@ultra.csn.tu-chemnitz.de) returned to #vserver. [09:15] kestrel (~athomas@dialup51.optus.net.au) joined #vserver. [09:37] [1]James (~James@ip68-96-180-27.lv.lv.cox.net) left irc: Ping timeout: 512 seconds [09:42] micah (micah@micha.hampshire.edu) joined #vserver. [09:51] iceberg (~infowolfe@pcp04891550pcs.frnkmd01.md.comcast.net) left irc: Ping timeout: 480 seconds [09:52] infowolfe (~infowolfe@pcp04891550pcs.frnkmd01.md.comcast.net) joined #vserver. [10:10] [1]James (~James@208.186.78.254) joined #vserver. [10:32] JonB (~jon@kg188.kollegiegaarden.dk) joined #vserver. [10:53] micah (micah@micha.hampshire.edu) got netsplit. [10:53] micah (micah@micha.hampshire.edu) returned to #vserver. [10:58] micah (micah@micha.hampshire.edu) left irc: Ping timeout: 512 seconds [11:23] Zoiah (Zoiah@matryoshka.zoiah.net) got netsplit. [11:23] MedivhWrk (ck@netops.multimedia-centrum.de) got netsplit. [11:23] Medivh (ck@62.93.217.199) got netsplit. [11:23] MrBawb (abob@swordfish.drown.org) got netsplit. [11:23] MrBawb (abob@swordfish.drown.org) returned to #vserver. [11:23] Medivh (ck@62.93.217.199) returned to #vserver. [11:23] MedivhWrk (ck@netops.multimedia-centrum.de) returned to #vserver. [11:23] Zoiah (Zoiah@matryoshka.zoiah.net) returned to #vserver. [11:24] <[1]James> hello? [11:35] hi [11:43] Morning guys. [11:44] Did they rename sys_call_table[] in 2.4.22 or have I missed something? [11:51] <[1]James> no idea [11:58] I've wrote a workaround module for recent mm/mmap.c/do_brk vulnerability and it loads into 2.4.20 perfectly, but for 2.4.22 it complains of undefined symbol sys_call_table. [12:12] [1]James (~James@208.186.78.254) left irc: Quit: HydraIRC -> http://www.hydrairc.com <- The dawn of a new age [12:18] Zoiah (Zoiah@matryoshka.zoiah.net) got netsplit. [12:18] MedivhWrk (ck@netops.multimedia-centrum.de) got netsplit. [12:18] Medivh (ck@62.93.217.199) got netsplit. [12:18] MrBawb (abob@swordfish.drown.org) got netsplit. [12:18] MrBawb (abob@swordfish.drown.org) returned to #vserver. [12:18] Medivh (ck@62.93.217.199) returned to #vserver. [12:18] MedivhWrk (ck@netops.multimedia-centrum.de) returned to #vserver. [12:18] Zoiah (Zoiah@matryoshka.zoiah.net) returned to #vserver. [12:37] Zoiah (Zoiah@matryoshka.zoiah.net) got netsplit. [12:37] MedivhWrk (ck@netops.multimedia-centrum.de) got netsplit. [12:37] Medivh (ck@62.93.217.199) got netsplit. [12:37] MrBawb (abob@swordfish.drown.org) got netsplit. [12:38] MrBawb (abob@swordfish.drown.org) returned to #vserver. [12:38] Medivh (ck@62.93.217.199) returned to #vserver. [12:38] MedivhWrk (ck@netops.multimedia-centrum.de) returned to #vserver. [12:38] Zoiah (Zoiah@matryoshka.zoiah.net) returned to #vserver. [12:45] TamaPanda (~a@193.173.84.237) joined #vserver. [12:45] morn [12:45] hey TamaPanda [12:45] hm i just tried something in vserver.. maybe someone could clarify [12:46] yes ? [12:46] after ssh'ing to a vserver, i did fdisk... it could write the partition, but not read (but of course, after writing the damage is already done) [12:46] or did it fake 'write' it? [12:46] dunno, try it [12:46] stop using swap [12:47] change swap from swap to a linux [12:47] how do i stop swap? :) [12:48] swapoff [12:48] but, before you try this [12:48] tell me what capabilities your vserver has [12:48] i specified no additional capabilities [12:50] let me see your .conf file please [12:51] 6 lines, want it in here? [12:52] sure [12:52] ok i changed swap partition type to linux [12:52] lets see if it holds after a reboot [12:52] IPROOT=eth0:192.168.2.152 [12:52] IPROOTDEV=eth0 [12:52] S_HOSTNAME=steelsoul [12:52] S_NICE=15 [12:52] S_FLAGS=lock sched [12:52] ULIMIT=-H -u 1000 [12:52] okay, no extra capabilities [12:53] okay, now, use top to see if you have any swap [12:53] it has to boot first :) [12:53] use top in the real server [12:53] of course [12:54] hm takes longer than usual [12:55] can't ssh yet [12:56] if this the real server or the vserver ? [12:56] real [12:56] ah tehre is the login prompt :) [12:56] the vserver doesnt auto-start yet [12:56] ok i have a top.. what is the swap thing? [12:57] hm [12:57] it held [12:57] what number is just left of swap ? [12:57] what held ? [12:57] no swap, and cfdisk shows 2 linux partitions [12:57] are they in usage ? [12:57] well i rebooted the entire server [12:57] so i guess so... [12:58] mount [12:58] are they used ? [12:58] which vserver are you running ? [12:59] /dev/hdc2 on / type ext2 (rw) [12:59] devpts and proc [12:59] no hdc1 ? [12:59] no, but top still says it has 512MB swap (the size of the original partition) [12:59] maybe if i completely removed the partition... [13:00] no [13:00] it never really showed swap in mount for me anyway [13:00] look in /etc/fstab and tell me where swap is [13:00] hdc1 [13:01] /dev/hdc1 swap swap defaults 0 0 [13:01] how far are you from the server ? [13:01] hm.. a city away. It is sitting in my bedroom, i'm at work :) [13:02] not really a biggy if it 'poop's [13:02] is it your ass if it doesnt come back up again ? [13:02] okay [13:02] fdisk -l /dev/hdc [13:02] in a verserver or in root? [13:03] s/ver/v [13:03] root [13:03] /dev/hdc1 1807 1868 498015 83 Linux [13:03] /dev/hdc2 * 1 1806 14506663+ 83 Linux [13:04] swapoff /dev/hdc1 [13:04] partitions are not in disk order.. swap is the last 512MB of the disk [13:04] top shows swap 0 [13:04] swapoff /dev/hdc1 [13:04] (after swapoff) [13:04] okay [13:05] mhepp (~mhepp@r72s22p13.home.nbox.cz) joined #vserver. [13:05] and now? [13:06] enter the vserver [13:06] there [13:06] fdisk /dev/hdc [13:06] mind if i cfdisk? i like it better :) [13:07] just delete swap? [13:07] no [13:07] change the partition type of the swap device from linux to swap [13:07] swap is number 82 (if i remember correctly) [13:08] ok done [13:08] it now has linux ext2 and linux swap [13:08] save [13:08] Wrote partition table, but re-read table failed. Reboot to update table [13:09] reboot [13:09] the real server [13:09] when it comes up again, let me know the output of fdisk -l /dev/hdc [13:09] its rebooting.. [13:10] *pompompom* [13:10] ;) [13:11] /dev/hdc1 1807 1868 498015 82 Linux swap [13:11] /dev/hdc2 * 1 1806 14506663+ 83 Linux [13:11] NOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO [13:11] seems to have been changed [13:11] heh [13:11] 'bye bye security' [13:11] ;) [13:11] okay, are you _SURE_ you were inside the vserver ? [13:11] yes [13:11] okay [13:11] you want to have a go yourself/ [13:12] i can give real root and vserver root if you want [13:12] no, i want to know something else [13:12] Zoiah (Zoiah@matryoshka.zoiah.net) got netsplit. [13:12] MedivhWrk (ck@netops.multimedia-centrum.de) got netsplit. [13:12] Medivh (ck@62.93.217.199) got netsplit. [13:12] MrBawb (abob@swordfish.drown.org) got netsplit. [13:12] sure [13:12] enter your vserver [13:12] Jonny (~oseufer@interner.SerNet.DE) joined #vserver. [13:12] ls -la /dev/hdc [13:12] vserver web enter [13:12] New security context is 49152 [13:12] root@steelsoul:/# ls -la /dev/hdc [13:12] brw-rw---- 1 root disk 22, 0 Jun 9 2002 /dev/hdc [13:13] okay, now, rm -rf /dev/hdc [13:13] done [13:13] then mknod hdc b 22 0 [13:13] inside the vserver [13:14] in /dev dir or anywhere? [13:14] root@steelsoul:/dev# mknod hdc b 22 0 [13:14] mknod: `hdc': Operation not permitted [13:15] Hello, I think there is a wrong behaviour in the vserver patch. When you connect from a vserver to 127.0.0.1 the destination is set to the vserver adress, but the source is still 127.0.0.1. I think it's a problem in route.h. Where can I send a patch or a mail? [13:15] vserver@list.linux-vserver.org [13:15] or talk it over with bertl here [13:16] TamaPanda: okay, so the problem is that you had the entry in /dev [13:16] hm so what devices are 'safe' to have there? [13:16] MrBawb (abob@swordfish.drown.org) returned to #vserver. [13:16] Medivh (ck@62.93.217.199) returned to #vserver. [13:16] MedivhWrk (ck@netops.multimedia-centrum.de) returned to #vserver. [13:16] Zoiah (Zoiah@matryoshka.zoiah.net) returned to #vserver. [13:16] TamaPanda: my suggestion is that you write an email to the list where you ask if the only way to prevent a vserver from doing this fdisk is by removing the entry in /dev [13:17] do you run debian ? [13:17] slack [13:17] 9.0 with 2.4.23 kernel, vserver 1.1.6 [13:17] okay, i think you should use the new vskel util [13:17] it makes a skeleton from your root server [13:18] in mine i have full log null ptmx pts random shm tty urandom xconsole zero [13:18] created with Mr. Sladens debiannewserver.sh [13:18] i'm not sure that the vserver should have access even if the /dev entry is there [13:18] ... [13:19] well it can not read.. [13:19] apperantly [13:19] heh [13:19] no, it can read [13:19] else you wouldnt know what the partition table was [13:20] right, but the fdisk utility says it cant [13:21] it says it can re-read it [13:21] it says this because it tells the kernel that the partition has been updated [13:22] and the kernel then either says "okay' using new" or "sorry, error, reboot" [13:22] ah so it can't 'activate' it [13:22] but it says re-read [13:23] well, that doesnt matter, because if you did this in the real server, it would say the same [13:23] ok [13:24] the kernel just says "sorry, error" if it has mounted partitions [13:26] anyway, email, but dont say it is a bug [13:26] would it be possible to make a device node without mknod? [13:26] just ask if it is supposed to be like this [13:26] i dont think so [13:27] also ask what about devfs [13:27] Nick change: apw-zzz -> apw [13:29] vserver@list.linux-vserver.org? [13:29] yes [13:29] do i need to be on the list? [13:30] apw (~apw@212.104.150.41) left #vserver (Client exiting). [13:31] i think so, but you could also just ask on this channel later tonight [13:31] serving (~serving@213.186.191.194) left irc: Ping timeout: 492 seconds [13:33] i'll just register on the list [13:33] okay, i think that the option is just to remove the dev entries, but i'm not sure [13:35] mount inside a vserver does show a different "device" mounted as / [13:35] i cant see why it should have access to any other blockdevices [13:35] hm it doesnt here [13:36] root@steelsoul:/# mount [13:36] devpts on /dev/pts type devpts (rw,gid=5,mode=620) [13:36] proc on /proc type proc (rw) [13:36] /dev/hdc2 on / type ext2 (rw) [13:36] /dev/sda1 on / type ext2 (rw,errors=remount-ro) [13:36] this is inside my vserver [13:36] hm irc client doesnt like / [13:37] and it is ide disks [13:37] outside vserver: /dev/hdc2 on / type ext2 (rw) [13:37] i know [13:37] it was inside i was thinking at [13:37] inside vserver: /dev/hdc2 on / type ext2 (rw) [13:38] in another it is /dev/hdv1 on / type vfs (none) [13:38] hm werent you using lvm ? :) [13:40] i am [13:40] all vservers has their own LV [13:40] 2 of them are encrypted [13:48] mail sent [13:53] lets see what happens :) [13:58] for now i just removed the nodes i dont need [14:41] mhepp (~mhepp@r72s22p13.home.nbox.cz) left irc: Quit: Tak ja padaaaaM [14:50] JonB (~jon@kg188.kollegiegaarden.dk) left irc: Quit: Client exiting [15:03] BobR (~georg@149.148.78.13) joined #vserver. [15:11] Nick change: Bertl_zZ -> Bertl [15:12] hi everyone! [15:13] hi [15:14] it seems my root from vserver got pooped :) [15:14] hm no.. too hasty.. its just using 100% cpu heh [15:14] 13:17:47 up 2:04, 2 users, load average: 78.29, 376.77, 823.43 [15:15] hmm, what processes are running/active? [15:15] no idea i let someone play with it with the intend to b0rk it :) [15:16] i'm talking to him now heh [15:16] did you use nproc/sched ? [15:16] but i guess this could be avoided by applying quotas [15:16] well i have sched in the config but when i start a vserver it has /usr/local/sbin/vserver: line 11: sched: command not found [15:17] huh? [15:17] could you provide the config? [15:17] maybe i just forgot something.. [15:17] the config of the vserver i run or.. [15:18] yes [15:18] hm i pasted it in here before.. 6 lines [15:19] IPROOT=eth0:192.168.2.152 [15:19] IPROOTDEV=eth0 [15:19] S_HOSTNAME=steelsoul [15:19] S_NICE=15 [15:19] S_FLAGS=lock sched [15:19] ULIMIT=-H -u 1000 [15:19] okay, I search ... [15:19] ahh okay ... you need quotes around "lock sched" [15:19] oh [15:19] and same for the ULIMIT ... [15:20] i just copy/pasted the site think [15:20] the/a [15:20] oh boy... haha [15:20] hmm, funny .... it is normal shell syntax, so X=1 2 gives [15:20] you bash: 2: command not found [15:20] that guy ran a fork/disk thing.. keep forking and filling the disk :) [15:21] exactly the things i knew i didnt look at yet haha [15:23] it seems its not shutting down the vserver [15:23] note: disk full, processes most likely depleted [15:23] you need some 'new' features to do this ... [15:24] Bertl: quota thing? [15:24] vs1.1.6 allows to send a signal to each process of a context ... [15:24] serving (~serving@213.186.190.25) joined #vserver. [15:24] this way you could SIGSTOP or SIGTERM them ... [15:24] i did vserver web stop [15:24] but i guess that tries it the 'nice' way [15:25] how can i shutdown a vserver with a kill command then? the vserver doesnt appear in my ps list [15:26] yeah, as I said, you need the new interface to do that on an overloaded vserver ... [15:27] you can get some 'preliminiary' tools from my site, or enricos latest version should? already contain this ... [15:27] it will be included in the next stable release (in a few days) [15:28] http://vserver.13thfloor.at/Experimental/vkill-0.01.tar.bz2 [15:32] ok i guess i need something else too [15:32] since i cant get the process id to kill :) [15:32] you can get the context id ... right? [15:33] how? [15:34] well look in the /var/run/vservers dir for your vserver name ... [15:34] S_CONTEXT=49163 [15:34] that thing? [15:35] yup, that is you 'dynamic' context id ... [15:35] so i can vkill that? [15:35]