1314059559 J * ser ~ser@host1.tldp.ibiblio.org 1314059773 Q * ser_ Ping timeout: 480 seconds 1314062816 J * aaagirl ~aaagirl@212-198-248-35.rev.numericable.fr 1314062900 Q * aaagirl 1314066164 Q * fback Ping timeout: 480 seconds 1314069504 M * Bertl off to bed now ... have a good one everyone! 1314070246 N * Bertl Bertl_zZ 1314073060 J * imcsk8 ~ichavero@h163.140.40.162.dynamic.ip.windstream.net 1314074318 J * sannes ~ace@cm-84.209.106.118.getinternet.no 1314076763 Q * imcsk8 Quit: This computer has gone to sleep 1314077283 Q * hparker Quit: Quit 1314079009 J * ncopa ~ncopa@3.203.202.84.customer.cdi.no 1314080129 J * harobed ~harobed@arl57-1-82-231-110-14.fbx.proxad.net 1314081022 Q * ncopa Ping timeout: 480 seconds 1314081102 J * ncopa ~ncopa@ti0143a340-0173.bb.online.no 1314081168 Q * harobed Ping timeout: 480 seconds 1314081532 J * imcsk8 ~ichavero@h11.104.22.98.dynamic.ip.windstream.net 1314081594 M * arekm hmm, root 164338 ERR 2.0 0.0 4320 980 ? D 05:02 4:26 /usr/bin/sxid 1314081621 M * arekm what does "ERR" mean? "couldn't lookup name" or something else? 1314081952 Q * derjohn_mob Ping timeout: 480 seconds 1314083750 M * nospoonuser_ arekm: is this a ps output ? 1314083798 M * arekm vps 1314083928 M * nospoonuser_ i thinks it's output is equal to ps. keep in mind that you see /proc/$pid/cmdline in the list and the process can change this by it's own 1314084090 M * arekm it's not equal 1314084262 M * nospoonuser_ ok then it's an special output from vps / nothing in the manpage ? 1314085266 Q * Guest5990 Remote host closed the connection 1314085408 J * josef_ ~josef@gw-gbg.ilait.se 1314086283 Q * imcsk8 Quit: This computer has gone to sleep 1314086398 J * derjohn_mob ~aj@ip-81-210-228-102.unitymediagroup.de 1314087197 J * mike__ mike@94.23.159.232 1314088698 J * kir ~kir@swsoft-msk-nat.sw.ru 1314088720 P * kir 1314092328 J * SwenTjuln_ ~SwenTjuln@77-111-2-36.ipv4.tusmobil.si 1314092422 Q * SwenTjuln Ping timeout: 480 seconds 1314092422 N * SwenTjuln_ SwenTjuln 1314092935 J * FireEgl ~FireEgl@2001:470:e056:1:4d59:a939:7dcc:9e41 1314095630 N * Bertl_zZ Bertl 1314095649 M * Bertl morning folks! 1314095774 M * daniel_hozac arekm: yeah, ERR means the process probably disappeared. 1314095902 M * arekm # vps aux|grep ERR |wc -l 1314095903 M * arekm 67 1314095936 M * daniel_hozac could be your kernel is simply broken. 1314095980 M * arekm could be, patch-3.0-rc7-vs2.3.1-pre7.diff 1314096545 M * SwenTjuln daniel_hozac: i've a problem running command: "vattribute --xid 3238 --get" 1314096555 M * SwenTjuln i get an errormsg: "vattribute: vc_get_umask(): Function not implemented" 1314096566 M * SwenTjuln kernel 2.6.38.6-vs2.3.0.37-rc15/ utils 0.30.216-pre2982 1314097549 J * BenG ~bengreen@cpc12-aztw24-2-0-cust146.aztw.cable.virginmedia.com 1314098615 M * daniel_hozac SwenTjuln: should be fixed now... 1314099777 Q * derjohn_mob Ping timeout: 480 seconds 1314100191 M * Bertl off for now ... bbl 1314100199 N * Bertl Bertl_oO 1314100210 M * SwenTjuln daniel_hozac: tnx! 1314100461 J * derjohn_mob ~aj@213.238.45.2 1314101700 J * ghislain ~AQUEOS@adsl2.aqueos.com 1314101733 M * ghislain hello there 1314101793 M * ghislain following my issue with quota and tagging i still meet some issue in my main partition that is shared for all vservers ( /vservers ) 1314101856 M * ghislain i use hashification on this one and tagging. Each vserver has a different directory /vserver/guest1.../ 1314101882 M * ghislain when i upgrade webmin for exemple i get: unable to create `./usr/share/webmin/cluster-shell/run.cgi': No space left on device 1314101882 M * ghislain but quota show 50% of space left and dlimit is fine too 1314101920 M * daniel_hozac inodes? 1314101937 M * daniel_hozac is it using /tmp? 1314101942 M * daniel_hozac do you have an strace? 1314101983 M * ghislain no i do not have a strace, aptitude is configured not to use the /tmp in ram but /var/tmp on disk 1314101994 M * ghislain i will try to get a strace 1314102036 M * ghislain inode is 8% used 1314102767 Q * FireEgl Read error: Connection reset by peer 1314103557 Q * jeroen__ Ping timeout: 480 seconds 1314103559 Q * Mr_Smoke Read error: Connection reset by peer 1314103577 J * Mr_Smoke smokey@layla.lecoyote.org 1314103881 J * jeroen__ ~jeroen@imap.powerinternet.eu 1314104939 Q * click Remote host closed the connection 1314105106 Q * kshannon_ Remote host closed the connection 1314105111 J * kshannon ~kris@122.252.14.166 1314105144 J * click click@ti0127a340-0368.bb.online.no 1314106324 M * ghislain you knwo murph'y rules ? i retry with strace and it works... tsss 1314109291 J * thierryp ~thierry@zanzibar.inria.fr 1314110827 J * dowdle ~dowdle@scott.coe.montana.edu 1314111793 Q * BenG Quit: I Leave 1314113357 Q * ncopa Quit: Leaving 1314114629 Q * ghislain Quit: Leaving. 1314116664 J * bonbons ~bonbons@2001:960:7ab:0:1dbb:8d1a:1322:8265 1314117012 J * nkukard ~nkukard@41-133-248-130.dsl.mweb.co.za 1314117041 M * nkukard hrmmm, if I tar up a centos system and install it as a template, it seems I run into an init issue and lots of sperg on my screen, has anyone encountered this? 1314117167 M * nkukard i guess initstyle plain is the only way 1314117469 M * Bertl_oO shouldn't be too hard to clean up the initscript mess for a guest with sysv style, but be careful, if you used tar with the wrong options, your guest might have bad ownership/permissions 1314118010 J * ghislain ~AQUEOS@adsl2.aqueos.com 1314118458 Q * derjohn_mob Ping timeout: 480 seconds 1314118866 Q * dowdle Remote host closed the connection 1314118886 J * dowdle ~dowdle@scott.coe.montana.edu 1314122658 Q * thierryp Remote host closed the connection 1314123599 J * imcsk8 ~ichavero@h11.104.22.98.dynamic.ip.windstream.net 1314126411 J * SkyNet2000 ~SkyNet200@71-81-25-51.dhcp.gwnt.ga.charter.com 1314126837 N * Bertl_oO Bertl 1314126842 M * Bertl back now ... 1314127051 J * hijacker_ ~hijacker@cable-84-43-136-96.mnet.bg 1314127654 J * hparker ~hparker@2001:470:1f0f:32c:beae:c5ff:fe01:b647 1314128450 Q * Marillion Quit: leaving 1314129922 J * Marillion ~dirk@178.63.150.30 1314130156 Q * hijacker_ Quit: Leaving 1314130932 M * nkukard Bertl, thanks 1314130956 M * nkukard Bertl, one more question .... is there anything stopping me technically wise from mounting a loopback FS and using quotas inside a vserver? 1314130980 M * nkukard I followed the howto on the site, but seem to be getting this failure in the vserver using strace ... 1314130982 M * nkukard quotactl(Q_GETFMT|USRQUOTA, "/dev/hdv1", 0, 0x7fff542c5a7c) = -1 ENXIO (No such device or address) 1314130996 M * Bertl nope, but it is a little inefficient and you probably want to use the vroot device to make it secure 1314131026 M * Bertl does /dev/hdv1 exist and map the loop device? 1314131061 M * nkukard yea, I used vroot , exactly as the site says, but using /dev/loop0 1314131073 M * daniel_hozac did you load the module? 1314131079 M * daniel_hozac did you run vrsetup? 1314131082 M * nkukard vroot module, yep 1314131120 M * nkukard one sec, I pastebin it 1314131126 M * Bertl so, let's start with ls -la /dev/hdv1 inside the guest 1314131165 M * nkukard http://pastebin.com/HKf7qqsd 1314131171 M * nkukard it is loop0 at the moment, mounted on the host 1314131177 M * nkukard with no quota options 1314131187 M * nkukard brw-r----- 1 root disk 4, 0 Aug 23 12:53 /dev/hdv1 1314131333 M * Bertl loop devices are xid tagged, you have to assign it to the guest 1314131358 M * nkukard aha, so thats why if I use it inside the guest using vrsetup its not working? 1314131363 M * Bertl but I'd strongly suggest to use the vroot for that 1314131372 M * nkukard check my script above, I do 1314131379 M * nkukard /usr/sbin/vrsetup /dev/vroot0 $LOOPDEV 1314131392 M * nkukard or should I mount vroot0 in the host instead of loop0? 1314131527 M * Bertl you should not mount the loop in the host at all 1314131630 M * nkukard ok 1314131645 M * nkukard how do I copy /dev/vroot0 into it? 1314131649 M * Bertl i.e. you want to assign the loop device inside the guest context 1314131660 M * nkukard oh 1314131665 M * nkukard hrmmmm 1314131673 M * nkukard see I need quotas on / it seems 1314131674 M * Bertl then setup the vroot device to map it, and use /dev/vroot0 as /dev/hdv1 (inside the guest) 1314131728 M * nkukard hrmmm 1314131839 M * nkukard yea, this seems way over my head 1314131880 M * Bertl not sure what you are trying to accomplish? 1314131922 M * nkukard I have 1 storage device on the host (11Tb) ..... I want a guest with quotas enabled, so I thought to create a loopback device pointing to a 105Gb file, and use that 1314131948 M * nkukard the docs on the site as far as I can understand say you need a extra partition 1314131962 M * nkukard I only have 1 partition with no way to create another 1314131987 M * Bertl who makes an 11Tb partition with no lvm? 1314131993 M * nkukard SAN 1314132022 M * nkukard thats unfortunately all I have to work with :( 1314132029 M * Bertl but I presume you get that partition as iscsi or ataoe no? 1314132051 M * Bertl or what kind of setup is that? 1314132056 M * nkukard right, its the host root device 1314132068 M * Bertl the host root? 1314132075 M * nkukard yes / 1314132090 M * Bertl I guess your setup is quite flawed already :) 1314132097 M * nkukard o well 1314132108 M * nkukard if its not possible with linux-vserver , then i'm doomed 1314132123 M * Bertl almost everything is possible :) 1314132165 M * Bertl let's get back to my original question, what is the goal of this excercise in bad design? 1314132187 M * nkukard all I want is a guest that has quota support on / 1314132208 M * Bertl nothing else on this host? 1314132230 M * nkukard the host has about 20 vservers on it 1314132250 M * Bertl ah, so you are already running those guests on the host /, yes? 1314132251 Q * sannes Remote host closed the connection 1314132257 M * nkukard yep 1314132302 M * Bertl yeah, we should put some kind of design guide on the wiki, with things you should avoid 1314132407 M * Bertl as I said, it should work if you create the loop inside the guest context, then let util-vserver mount it as guest root with the necessary options 1314132439 M * Bertl configure vroot to this device and add it to the guest as /dev/hdv1 1314132444 M * nkukard yea, I been trying to figure out how to create a loop inside the guest 1314132461 M * Bertl inside the guest context, not necessarily inside the guest 1314132507 A * nkukard googles 1314132515 Q * imcsk8 Quit: This computer has gone to sleep 1314132528 M * Bertl i.e. something like vcontext --create/--migrate --xid -- losetup ... 1314132614 M * nkukard can I put that into my initialize script? say at the top? 1314132695 M * Bertl yeah, but prepre-start is probably simpler 1314132732 M * Bertl ah, no, initialize is required 1314132749 M * nkukard hrmm 1314132766 M * Bertl prepre-start already has the rootfs mounted 1314132774 M * nkukard http://www.nongnu.org/util-vserver/doc/conf/configuration.html <= is there a more updated doc than that I can use to read up on prepare-start? 1314132791 M * Bertl I don't think so 1314133027 M * nkukard checking at the start script, it seems as you said prepre-start is after rootfs mount 1314133456 M * nkukard ok Bertl , so create & vrsetup the loop in initialize using vcontext, then mount it in the context too as root? 1314133512 M * Bertl the vrsetup is better done in the host context and doesn't need to happen after the loop setup 1314133560 M * nkukard ah 1314133666 M * nkukard ok, so vrsetup first to loop0 1314133675 M * nkukard then vcontext the assignment of the loopback device 1314133701 M * Bertl sounds good 1314133751 M * nkukard damn, the path to the loopback image doesn't appear available within in the context 1314133843 M * Bertl hmm? 1314133947 M * nkukard http://pastebin.com/UEEcXLGL 1314133956 M * nkukard /vservers/web-1-1.virt.iitsp.net.dsk: No such file or directory 1314133957 M * nkukard hrmm 1314134037 M * Bertl sounds strange, try vcontext --create --xid 800 -- /bin/bash 1314134051 M * Bertl then execute the losetup manually (and check for the pathes) 1314134108 M * nkukard ah, /vservers is hidden inside the context it seems 1314134140 M * Bertl barrier most likely, I doubt you are using tags 1314134233 Q * bonbons Quit: Leaving 1314134257 M * nkukard is that a bad thing? 1314134268 M * Bertl the barrier? no 1314134313 M * nkukard http://pastebin.com/VPrPJJva 1314134326 M * nkukard i think the first error is normal because I didn't delete the vroot 1314134339 M * nkukard /dev/hdv1 / ext3 defaults,usrquota,grpquota 0 0 1314134341 M * nkukard my fstab has that 1314134364 M * Bertl nah, you can't mount hdv1, it's just for the quota ioctls 1314134382 M * Bertl i.e. you want to use /dev/loop0 in the guest configs fstab 1314134415 M * nkukard ah 1314134464 M * nkukard ok, i've bust something now ,... urmm 1314134485 M * nkukard http://pastebin.com/YjissF4N 1314134525 A * nkukard removes it 1314134543 M * Bertl well, you probably want to losetup -d in the guest stop scripts 1314134563 M * nkukard yea, but i don't think its reaching the stop scripts because its not starting? 1314134592 M * Bertl i.e. postpost-stop, and yes, you want to do that in the context as well, and in the error case manually 1314134752 M * nkukard http://pastebin.com/rNaq1mtb <= I probably need to grant it the secure mount option? 1314134869 M * Bertl daniel_hozac: would it be possible to have an fstab.context or so, which does the mount with the process context id? 1314134888 M * Bertl (or ist that functionality already available in a different way?) 1314134941 M * nkukard yea, I just tried adding secure_mount to ccapabilities and got the same error, hrmm 1314134959 M * Bertl the problem is not the guest permission 1314135010 M * nkukard ok 1314135010 M * Bertl you probably need to work around by doing the rootfs mount in the initialize script as well 1314135100 M * nkukard like I had before mounting it into /vservers ? 1314135131 M * Bertl well, it will be a little trickier I guess :) 1314135145 M * Bertl you probably want to mount it under e.g. /mnt 1314135193 M * Bertl (from within the context) then add a bind mount from e.g. /mnt to / in the guest configs fstab 1314135482 M * nkukard so the problem here is the combination of loop plus quotas? 1314135503 M * Bertl yep, normally you use a device mapper device for that 1314135526 M * nkukard so without the loopback i would not be having this issue? 1314135625 M * Bertl precisely 1314135687 M * nkukard ok, so say for instance doing something as DUMB as this .... exporting a iscsi on the host, and then imported the block device ... would work? heh 1314135727 M * Bertl probably, but that adds the 'network' complication 1314135739 M * nkukard network complication? 1314135765 M * Bertl i.e. you need at least to use the fstab.remote for that to do the mount in the guest's network context 1314135808 M * nkukard so it couldn't be a host iscsi at all? even if its remote? 1314135825 M * Bertl now you've lost me 1314135849 M * nkukard lets say , I have 2 server right, one is exporting a file over iscsi 1314135863 M * nkukard my host server grabs that and gives me say /dev/mydrive 1314135868 M * Bertl a filesystem stored in a file 1314135882 M * nkukard just as an example, its remote, so it can be a block device even 1314135888 M * Bertl okay 1314135917 M * nkukard would there be an issue because /dev/mydrive is a iscsi and being used as a block device in my vservers fstab config? 1314135960 M * Bertl no, but you have to take into account, that the guest will be limited to certain IPs 1314135979 M * Bertl so normally it is desired to do that iscsi setup/mount in the guest's network context 1314135988 M * Bertl *desireable 1314136002 M * nkukard ok, so its not a "normal" block device then, because the network context would prevent communication? 1314136030 M * Bertl it might prevent and/or might modify the communication 1314136042 M * nkukard ah, ok, this is very useful to know 1314136055 M * Bertl for example, for nfs it is required to use the network context 1314136070 M * Bertl otherwise you end up with strange side effects and permission issues 1314136089 M * nkukard ok ... *thinks* 1314136098 M * nkukard seems i have one very interesting issue here :) 1314136205 M * Bertl well, I'd call it a broken setup/design :) 1314136462 M * nkukard Bertl, so .... if I didn't need quotas on / on the guest, I could just create my loopback inside the guest, mount it and enable quotas and avoid these issues too? 1314136493 M * Bertl yeah, but with less security, as you need to allow the guest to do that 1314136499 M * nkukard right 1314136517 M * nkukard Bertl, thankyou very much for you help man, its more appreciated than you know 1314136711 M * nkukard I'll carry on with this tomorrow and see if I can fix the setup 1314136756 J * imcsk8 ~ichavero@h163.140.40.162.dynamic.ip.windstream.net 1314136793 M * Bertl np 1314139254 M * Bertl off to bed now ... have a good one everyone! 1314139261 N * Bertl Bertl_zZ 1314139394 J * ghislain1 ~AQUEOS@adsl2.aqueos.com 1314139669 J * ghislain2 ~AQUEOS@adsl2.aqueos.com 1314139795 Q * ghislain Ping timeout: 480 seconds 1314139879 Q * ghislain1 Ping timeout: 480 seconds 1314140144 Q * DoberMann Ping timeout: 480 seconds 1314140667 J * DoberMann ~james@2a01:e35:8b44:84c0::2