[00:03] Hm I am cleaning up some this doc [00:05] Tamama: thx. [00:06] is there somebody willing to test this stuff? [00:07] test what? [00:07] the memory limits ;) [00:08] Oh.. its an addon patch? or not? [00:09] yup, it _was_ an addon, and it _might_ be included in the next development _if_ somebody is willing to test ;) [00:09] heh [00:09] sure i can test it [00:09] if you can tell me what it does exactly (or at least is supposed to do) [00:09] well, it imposes a virtual memory limit for a context ... [00:13] usefull [00:14] hm.. [00:14] i am wondering how vrsetup ended up in /sbin/ [00:14] cp -va vrsetup /sbin/vrsetup maybe? [00:15] it wasnt me [00:15] nm found it [00:15] install vrsetup /sbin/vrsetup [00:15] well, as my makefile doesnt include any install target ;) [00:15] right [00:15] well i am documenting so... [00:15] :) [00:16] I think i have it down now.. [00:19] let me upload it :) [00:22] http://www.vangog.net/docs/vs-quota.simple-walkthrough.txt [00:22] care to see if i missed something? :) [00:22] sure .. [00:23] it's a 'dummy guide', except for the actual first kernel compile with vservers [00:23] oh.. missed the cq make [00:23] let me fix :D [00:24] send it to me via email, I comment that line by line ;) [00:24] oops, forgot: please be so kind and ... and .. thanks in advance :( [00:25] lol [00:25] give me your mail address then/ [00:25] herbert@13thfloor.at [00:26] sent [00:26] i think i missed some parts but maybe thats because it was already present in my system.. wouldnt know heh [00:29] i guess i could re-order some parts [00:29] heh [00:29] save some vserver x enter lines [00:29] oh well, good practice ;) [00:32] Is it normal behaviour of bochs to eat up 100% CPU all the time? [00:32] ccooke (~ccooke@80.1.164.238) joined #vserver. [00:32] i wouldnt know, but i guess it could [00:32] that is an x86 emu right? [00:33] virtuoso: yes, because there is no way to idle in userspace .. [00:34] I thought I compiled some kernel module for it.. [00:34] that is plex [00:35] bochs doesn't use kernel modules ... [00:35] Hm. [00:35] (at least it didn't when I last tried it) [00:35] qemu is a good alternative ... [00:35] It's not based on plex86/bochs? [00:35] QEMU 0.5.1 was released yesterday ... [00:36] nope qemu is a completely independant development [00:36] And it's GPLed? [00:36] heh Bertl what did you think of it besides the comments? ;) [00:36] of course, would I use it if not? [00:36] Tamama: short but usable .. you could explain a little more of the actual magic ... [00:37] short? i thought it was longer than the one existing :D [00:37] yup, that for sure ... [00:37] but oh well, it is a walk-through.. what can you expect :P [00:38] it is not really ment as a 'do all know all' doc at this time [00:38] if you want to improve it, add some explanations after each command ... [00:38] but anyway, publish it via the wiki ... [00:38] hm such as? [00:39] mknod /dev/vroot/0 b 4 0 [00:39] ah yes [00:39] nobody knows what do do if he needs vroot/1 for example ... [00:39] well dont even know what devfs is :D [00:40] and really that mknod is uber-magic to me lol [00:40] man mknod ? [00:40] don't you think the others could do that too? ;) [00:40] yes, but they won't figure out what arguments to use for vroot/1, right? [00:41] hm i guess.. but i dont know either :D [00:41] and i only have 1 vroot [00:41] heh [00:41] well, that's simple, ask somebody who knows ... [00:41] Hi! :) [00:41] mknod /dev/vroot/1 b 4 1 [00:42] basically it's tequiv. to the loop device ... do you know that one? [00:42] s/tequiv/equiv/ [00:43] hm [00:43] the tar 'j' option helps :D [00:44] though i dont see how the 'j' relates directly to 'bzip2' :D [00:44] at least the 'z' is in 'gzip' heh [00:45] first they had Y then I now it's j ... nobody knows ... [00:45] guess it's the battle for the free commandline character options :D [00:46] hm what is a command line to check the quota tools version? [00:47] sec [00:47] quotacheck -V [00:47] quota -V [00:48] k, added [00:49] now could you explain devfs? [00:49] apperantly i have it, but i dont.. well i dunno.. lol [00:49] okay, devfs is a replacement for /dev [00:49] you do not have to create devices anymore with devfs [00:50] if you, for example, plug in an usb device, devfs creates the required device node ... [00:50] and if you unplug it, devfs removes it again ... [00:50] ok.. and how do you get it? [00:50] heh [00:51] it's in the kernel, just enable it ... [00:51] under filesystem ... devfs ... [00:51] i didn't _disable_ it.. [00:51] well, the issue is, it is unmaintained since more than a year, developer disappeared ... the other kernel developers don't like it ... [00:52] /dev/ file system support (EXPERIMENTAL) [00:52] that one? [00:52] so it got depreciated with 2.6 and replaced with udev filesystem, which isn't able to do that stuff, but it's better than nothing ... [00:52] yes that one! [00:52] experimental == not in my kernel (well when i can help it :D) [00:53] I'll just add that to my assumptions :D [00:53] no devfs [00:53] :D [00:56] even so, if you DO have devfs.. this still works i think :D [00:56] if you do not want to address devfs, you should change from /dev/vroot/0 to /dev/vroot0 ;) [00:57] well, the mknod will fail ... [00:57] hrm [00:57] the old docs said /dev/vroot/0 :D [00:58] because I assumed devfs ;) [00:58] note the / [00:58] hrm [00:58] nm [00:58] heh [00:58] why should i make it /dev/vroot0 then [00:58] i could name it /dev/vroot/lalalala_i_am_bugging_Bertl [00:58] heh [00:59] because that is the 'old' dev style like /dev/loop0 [00:59] yes, you could also name it /tmp/I_don't_like_you ;) [00:59] true [01:01] good idea [01:01] :D [01:03] hm, to change the device i can just do: [01:03] mknod /dev/vroot0 b 4 0 [01:03] adn change the vrsetup line in rc.local right? [01:03] yup, maybe you should mention the difference and the alternate form in ( ... ) [01:04] maybe like, with devfs, this would be /dev/vroot/0 ;) [01:05] Assumptions: [01:05] You do not have devfs [01:05] If you do, all vroot devices are /dev/vroot/device_number [01:05] and are auto-created, so no need for the mknod later on. [01:06] /dev/vroot/ instead of /dev/vroot (note the additional '/' ) [01:08] hm? [01:08] is that for me or the doc? ;) [01:08] because in either case i dont know where to fit that in lol [01:08] doc: If you do, all vroot devices are named /dev/vroot/ instead of /dev/vroot (note the [01:08] additional '/' ) [01:09] ok works for me [01:09] AND they dont need to mknod :D [01:09] yup .. [01:10] forgot that one ;) [01:15] http://www.13thfloor.at/vserver/d_release/v1.3.1/ [01:21] hm let me send you and updated vsersion [01:21] make it so .. [01:21] 'you have mail' [01:22] oops, left out vroot1/2/3 ;) [01:22] oh well [01:26] [ You can in fact choose other methods, but this one [01:26] +gives 65536 [01:27] you _always_ get 64k contexts, what changes is the size of UID and GID from 2^32 to 2^24 in this case ... [01:27] the other option is 2^32 UID and 2^16 GID [01:27] right.. heh [01:27] and you can store it ina n unused place on the disk (which gives UID/GID 2^32) [01:28] [ You can in fact choose other methods, but 24/24 looks ok to me [01:28] if you don't know how much 2^32 and 2^24 is, ask google ;) [01:28] :) [01:28] Action: Tamama rolls eyes [01:29] 2^32 is 34 of course [01:30] in that case, better ask google ;) [01:30] 2 xor 32 really is 34... [01:30] heh [01:30] Bertl: It looks like damn fast (QEMU) in comparsion to bochs. :) [01:31] that's why I use it, and because I can attach with the debugger, quite easily ... [01:31] Yeap, I noticed that feature. [01:31] so other than that 24/24 thing... [01:31] and the best of all, with the right config, you do not even need a graphical display ;) [01:32] a terminal is enough to get linux up and running ;) [01:32] Action: virtuoso dances around the table. [01:32] Cool. [01:40] heh [01:41] Tamama: s/Make it/Build it/g [01:43] again, it's not necessary to 'enter' the vserver to write the /etc/mtab ;) [01:43] i know, but i re-arranged the commands so that you can enter 3 in a row [01:44] and don't forget to actually _enable_ the quota after all that testing, 'quotaon -vaug' [01:44] inside or outside the vserver? [01:44] inside of course ... [01:44] everything is now available from 'inside' ... [01:44] root@powerhouse:/# quotaon -vaug [01:44] quotaon: using //aquota.group on /dev/hdv1 [/]: No such device or address [01:44] quotaon: using //aquota.user on /dev/hdv1 [/]: No such device or address [01:45] ahh, forgot the cp /dev/vroot0 /my/vserver/dev/hdv1 ? [01:45] hm no, but i did kill the original vroot/0 device.. *rolls eyes* [01:45] heh [01:45] or didn't give the CAPQUOTATL ? [01:46] CAP_QUOTACTL ;) [01:46] cp -ap /dev/vroot0 /vservers/web/dev/hdv1 [01:47] tanjix1 (ViRu_@c-180-204-192.n.dial.de.ignite.net) joined #vserver. [01:48] hm actually i forgot that copy in the doc ;) [01:48] tanjix (ViRu_@c-180-204-223.n.dial.de.ignite.net) left irc: Read error: Connection reset by peer [01:51] Nick change: tanjix1 -> tanjix [01:52] Madkiss_ (madkiss@madkiss.org) got netsplit. [01:53] Madkiss_ (madkiss@madkiss.org) returned to #vserver. [01:56] root@powerhouse:/# quotacheck -maug [01:56] quotacheck: Can't create new quotafile //aquota.user.new: Read-only file system [01:56] quotacheck: Cannot initialize IO on new quotafile: Read-only file system [01:56] Last message repeated 1 time(s). [01:56] quotacheck: Can't create new quotafile //aquota.group.new: Read-only file system [01:56] ok.. that looks bad :D [01:57] hmm, what changed? [01:58] dunno [01:58] i just removed /dev/vroot/0 and re-created /dev/vroot0 [01:59] okay, what did you change/do? (history) [01:59] then copied the new /dev/vroot0 to the vserver /dev/hdv1 [01:59] and rebooted :D [01:59] okay, and then you did the vroot setup with the wrong device, right? [01:59] (after changing the vrsetup in my rc.local) [02:00] nope, its /dev/vroot0 [02:00] looks like no quota hash added for that context ... [02:00] hm.. [02:00] is that added to the device or the fs ? [02:01] it is added to the fs [02:01] then it should have been there [02:01] but you have to do it after a reboot ... it's in memory ;) [02:01] DOH! [02:01] so every time i reboot and start a vserver.. i have to run quotacheck? [02:02] or... [02:02] again, ahve a look at http://www.13thfloor.at/old/VServer/HowTo_LVMQS.shtml [02:02] this doesn't use the 'new' tools but the script shows how to 'incorporate' [02:03] although you do not need the vrsetup in the vserver script ... [02:03] but you would add the cqhadd/cqhrem to the vserver script ... accompanied by some quotaon/quotaoff ... [02:03] it doesnt use cqhadd [02:04] anyway after running that myself it worked [02:04] guess i better put that in the startup script [02:04] yup, you should do it on vserver start, and the cqhrem on vserver stop ... [02:05] this can be automated in the same way, as done in this old howto for 'other' tools ... [02:06] so do i need to run 'quotaon' also each time i start a vserver? [02:06] if you remove and add the quota hashes in the meantime, yes, otherwise no ... [02:07] the logic is a simple pairing ... [02:07] a) cqhadd [02:07] b) quotacheck [02:07] c) quotaon [02:07] d) quotaoff [02:07] e) cqhrem [02:08] between c and d is the lifetime of the vserver ... [02:08] right [02:08] Action: Tamama adds to doc :D [02:08] cqhrem will fail when quota is on ... [02:08] quotacheck and quotaon will fail if the cqhadd is missing ... [02:09] and most important: 'read only filesystem' means that the quota hash is missing ;) [02:09] (reminds me that I have to change that soon) [02:09] Action: Tamama smells a vserver startup script coming [02:11] hm quotaoff takes a while [02:18] root@powerhouse:/# quotaoff -a [02:18] quotaoff: quotactl on /dev/hdv1 [/]: No such process [02:18] means? [02:19] it hangs :) [02:19] that is bad, any processes still on that device? [02:19] how to check? [02:19] lsof, /proc/mounts, ps ... [02:20] root@powerhouse:/dev# lsof | grep hdv1 [02:20] lsof: WARNING: can't stat() ext3 file system /vservers [02:20] Output information may be incomplete. [02:20] from inside the vserver though (still funny it got to /vservers [02:21] cat /proc/mounts [02:21] no hdv1 [02:21] yeah, but everything else ... [02:21] devpts /dev/pts devpts rw 0 0 [02:21] none /vservers/web/dev/pts devpts rw 0 0 [02:21] none /vservers/web/proc proc rw 0 0 [02:21] usbfs /proc/bus/usb usbfs rw 0 0 [02:21] proc /proc proc rw 0 0 [02:22] /dev/sda4 /vservers ext3 rw 0 0 [02:22] other / devs are cut by mirc heh (bloody / commands) [02:24] sed 's/^\//\/\ \//g' fixes that for you, and you can even knit with that pattern ;) [02:24] 724 pts/0 D 0:00 quotaoff -vaug [02:24] 843 pts/1 D 0:00 quotaoff -a [02:24] 995 pts/1 S 0:00 /bin/bash -login [02:25] ps -x [02:25] i cant kill those 2 [02:25] why is the bash stopped? [02:25] because it was running a ps -x ? [02:25] sorry ignore that ... [02:25] it's sleeping ... [02:26] 'D' means device I/O something is blocking the quotaoff .. maybe some mount point or opened file ... [02:27] well i killed the server while it was doing that [02:27] guess it couldnt kill em [02:27] heh [02:27] even kill -9 has no help for me now [02:27] should not happen, but we had some ext3 issues with quota in 2.4.22, I don't know if they where resolved ... [02:28] oh goodie [02:28] not vserver specific, only quota+ext3 ... [02:28] right [02:28] still goodie [02:28] heh [02:29] vkill cant kill it either [02:30] something good came from windows [02:30] nothing can kill a process in 'D' state ... [02:30] 'reboot!' [02:30] :) [02:31] can i easily 'downgrade' ext3 to ext2/ [02:31] yup, 'just' remove the journal ... [02:34] tune2fs -O^has_journal [02:34] e2fsck [02:34] ? [02:36] oh well i'll just try :D [02:36] that is, why I used 'just' [02:36] its only an empty vserver, if it screws up, nothing lost :D [02:36] you can mount it as ext2, search for the journal ... [02:40] also a posibility :) [02:40] Action: Tamama kills lost+found [02:40] tanjix (ViRu_@c-180-204-192.n.dial.de.ignite.net) left irc: Read error: Connection reset by peer [02:47] quotaoff is on D again [02:47] ext2 or ext3= [02:48] 2 [02:48] okay, I need a detailed list of your commands ... [02:49] what i type to get that or what i did to get this setup? [02:49] well, from the boot, everything related to the vserver partition ;) [02:50] starting with mount -t ext2 /dev/sdax -o tagctx,usrquota,grpquota /vserver [02:51] its automounted in fstab [02:51] well, then unmount it after boot, and do it by hand ... [02:54] k, but i'll have to wait untill it checked 140GB with fsck (uncleanly unmounted) [02:54] heh [02:54] np [03:07] noel (~noel@pD9E099F2.dip.t-dialin.net) got netsplit. [03:11] noel (~noel@pD9E099F2.dip.t-dialin.net) returned to #vserver. [03:47] okay, enough for me for today ... [03:47] have a nice one, cu 2morrow ... [03:47] Nick change: Bertl -> Bertl_zZ [04:32] xsbyme (xsbyme@D5E0EDEB.kabel.telenet.be) joined #vserver. [04:33] tanjix (ViRu_@c-180-200-254.n.dial.de.ignite.net) joined #vserver. [06:04] shuri (~ipv6@cpu183.adsl.qc.bellglobal.com) joined #vserver. [06:48] xsbyme (xsbyme@D5E0EDEB.kabel.telenet.be) left irc: [08:32] serving (~serving@213.186.191.213) left irc: Ping timeout: 512 seconds [09:04] shuri (~ipv6@cpu183.adsl.qc.bellglobal.com) left irc: Quit: ipv6 [09:52] Doener` (~doener@pD9E129C1.dip.t-dialin.net) joined #vserver. [10:00] doener_aw (~doener@p5082D4A2.dip.t-dialin.net) left irc: Ping timeout: 499 seconds [10:32] noel- (~noel@pD9FFA5AB.dip.t-dialin.net) joined #vserver. [10:40] noel (~noel@pD9E099F2.dip.t-dialin.net) left irc: Ping timeout: 493 seconds [11:22] serving (~serving@213.186.191.213) joined #vserver. [13:12] serving (~serving@213.186.191.213) left irc: Ping timeout: 485 seconds [13:33] xsbyme (xsbyme@D5E0EDEB.kabel.telenet.be) joined #vserver. [13:40] serving (~serving@213.186.189.119) joined #vserver. [13:44] hi @all [13:45] Topic changed on #vserver by noel-!~noel@pD9FFA5AB.dip.t-dialin.net: http://linux-vserver.org/ || latest stable 1.22, devel 1.3.1 "Revolutions". [13:51] LL0rd: hi [14:39] Madkiss_ (madkiss@madkiss.org) got netsplit. [14:40] Madkiss_ (madkiss@madkiss.org) returned to #vserver. [15:42] spunk (spunk@b099203.adsl.hansenet.de) joined #vserver. [15:43] spunk (spunk@b099203.adsl.hansenet.de) left irc: Client Quit [16:10] spunk_ (www-data@62.216.180.231) joined #vserver. [16:12] Hello [16:13] I have a problem with my new vserver installation maybe somebody can help me? [16:15] the problem is that i cant reach any domainname from my vservers. from the hostsystem everything works well.. ping on a domain results in unknown host. ping on an IP does work. did anybody here have the same problem? [16:26] spunk_: correct your /etc/resolv.conf in the vserver [16:26] Nick change: noel- -> noel [16:30] noel: no thats not the problem its the same resolv.conf as in the hostsystem [16:31] the namesservers are in my resolv.conf and i can reach them with ping [16:33] spunk_: hmm. but its an dns Problem. does "dig example.com @yourdns" work? [16:43] i didnt test that but nslookup doesnt work [16:44] i have a dsl connection to the internet may tha cause any problems? [16:53] hmm.. dig did return: no server could be reached [17:06] ok i've got to leave , if i can'T fix that problem i'll post in in the mailinglist [17:06] cu [17:07] spunk_ (www-data@62.216.180.231) left irc: Quit: ...und wech (EOF) [19:00] kestrel (~athomas@dialup51.optus.net.au) left irc: Ping timeout: 485 seconds [19:45] Nick change: Bertl_zZ -> Bertl [19:45] morning! [19:46] good evening :) [20:06] brb. 20min [20:06] Nick change: Bertl -> Bertl_oO [20:36] Nick change: Bertl_oO -> Bertl [20:52] kestrel (~athomas@dialup51.optus.net.au) joined #vserver. [20:52] hi kestrel! [21:01] kestrel (~athomas@dialup51.optus.net.au) left irc: Ping timeout: 512 seconds [21:03] Kalasa (~Kalasa@dslam160-246-58-62.adsl.zonnet.nl) joined #vserver. [21:03] hi Kalasa! [21:03] Holas [21:03] Tamama here, just someplace else :) [21:04] what does Kalasa/Tamama mean? [21:04] just a nick [21:05] I got per context quota's working heh [21:06] wow, how _unexpected_ ;)# [21:06] just that i dont run quotacheck as it yabbers that it cant mount the file system read only [21:06] of course m overrides.. but still [21:07] ever tried on a 'normal' filesystem? [21:07] works ok there [21:07] let me paste the error [21:08] root@powerhouse:/# quotacheck -vaug [21:08] quotacheck: Quota for users is enabled on mountpoint / so quotacheck might damage the file. [21:08] Please turn quotas off or use -f to force checking. [21:08] uuh [21:08] well, turning quota off, will fix that ... [21:08] wrong message ;) [21:08] since i did quotaon to just test the rest :D [21:09] i am afraid to do quotaoff now since i am really remote now and cant hard-reset the pc once it reboots and hangs on that blocking process [21:09] and then nI do not have anything to play with this evening :D [21:11] brb, got called for dinner :d [21:13] lala [21:34] root@powerhouse:/# quotacheck -vaug [21:34] quotacheck: Cannot remount filesystem mounted on / read-only so counted values might not be right. [21:34] Please stop all programs writing to filesystem or use -m flag to force checking. [21:34] it apperantly is a dinner in etappes :D [21:35] yup, that is normal, as you are not allowed to remount a 'shared' partition (or any other) inside vserver ... [21:35] ok [21:35] just making sure [21:35] :) [21:36] i'll add the script i made to run on boot of the root server to initialise the vservers as well to the doc [21:36] its pretty simple but effective [21:37] great, no more issues with quotaoff? [21:37] uh.. to scaredycat to try while remote :) [21:37] since it hangs the server on shutdown [21:37] huh? [21:38] Madkiss_ (madkiss@madkiss.org) got netsplit. [21:38] Madkiss_ (madkiss@madkiss.org) returned to #vserver. [21:38] if i run quota-off and it enters D state [21:38] the main server cant shutdown either (waits a long long time [21:38] that sounds not good ... I guess I have to investigate that one ... [21:38] yeah it's pretty nasty [21:39] i can strace quotaoff now i guess [21:39] but after that my vserver if effectively unusable untill next reboot [21:40] that doesn't help, it probably hangs inside the kernel, waiting for some inodes to be unlocked ... [21:40] we had that issue a long time ago, but I fixed it ... now it seems back, maybe something in the kernel changed again ... [21:41] guess my stumbling over all small things does help in some way :D [21:41] just not the way everybody wants lol [21:42] another issue [21:42] bring it on ... [21:42] I ran sshd in my vserver [21:42] hm nm that [21:42] it was unused [21:42] ok, situation: [21:42] login ssh root server [21:42] vserver web enter [21:42] lose ssh connection [21:43] users are still logged in in the vserver [21:43] ssh is not 'properly' closed, connection was lost [21:43] hmm, didn't understand that ... please elaborate ... [21:43] i have currently 3 users in my vserver [21:44] of which 2 are lost ssh connections [21:44] 674 pts/0 S 0:00 /bin/bash -login [21:44] 745 pts/1 S 0:00 /bin/bash -login [21:44] 854 pts/2 S 0:00 /bin/bash -login [21:44] 869 pts/2 R 0:00 ps -x [21:44] 0 and 1 are disconnected [21:45] hmm, you mean the entries in /var/log/{w,u}tmp are still there? [21:45]