1155341665 J * Viper0482 ~Viper0482@p54977A49.dip.t-dialin.net 1155341724 Q * Viper0482 1155341845 J * Viper0482 ~Viper0482@p54977A49.dip.t-dialin.net 1155341897 Q * comfrey Ping timeout: 480 seconds 1155341897 J * shedi ~siggi@inferno.lhi.is 1155342340 Q * ekc Ping timeout: 480 seconds 1155342504 J * liquid3649_ ~Viper0482@p54977A49.dip.t-dialin.net 1155342660 Q * liquid3649_ 1155342788 Q * Viper0482 Quit: one day, i'll find this peer guy and then i'll reset his connection!! 1155342816 J * Viper0482 ~Viper0482@p54977A49.dip.t-dialin.net 1155342841 Q * Viper0482 Remote host closed the connection 1155343770 J * ntrs_ ~ntrs@68-188-51-87.dhcp.stls.mo.charter.com 1155344052 Q * ntrs Ping timeout: 480 seconds 1155344529 J * ntrs__ ~ntrs@68-188-51-87.dhcp.stls.mo.charter.com 1155344529 Q * ntrs_ Read error: Connection reset by peer 1155345452 N * Bertl_oO Bertl 1155345458 M * Bertl evening folks! 1155345517 M * matti Bertl: :-) 1155346292 Q * jkl Ping timeout: 480 seconds 1155346455 M * Bertl hey matti! everything fine? 1155346483 M * matti Bertl: Yes, almost :) 1155346509 M * Bertl everything is _almost_ fine, or almost everything is fine? :) 1155346545 M * matti Everything is almost fine. 1155346547 M * matti :) 1155346670 M * ekc2 I noted that the cpu lines in /proc/virtual//sched will not update to reflect a new scheduler parameters (tokensmin) while in the 'on hold' state. 1155346677 M * ekc2 is that behaviour unusual? 1155346733 M * Bertl nope 1155346739 M * Bertl that is called lazy update 1155346750 M * Bertl and it is the default, you can force an update though 1155346766 M * ekc2 yes, that's what i'd like. how can I force an update? 1155346786 M * Bertl what tools are you using to set it, atm? 1155346792 M * ekc2 vsched-0.2 1155346809 M * ekc2 i don't mind using something else, though. what do you suggest? 1155346839 M * Bertl nah, just asking, I guess I have to update the tool soon 1155346885 M * Bertl give me a minute to figure out the magic 1155347095 M * ekc2 also, minor documentation bug in vsched-0.2: the '-i' operator needs to appear before the '-R' and '-I' operators to work. help text doesn't state that, however. 1155347128 M * Bertl right, that's what I meant with update :) 1155347146 M * Bertl well, it wasn't designed as production tool, but enrico didn't catch up yet 1155347318 M * Bertl ekc2: ah, here it is ... 1155347323 M * Bertl #define VXSM_FORCE 0x0400 1155347354 M * ekc2 and re-compile the kernel? 1155347362 M * Bertl no, userspace only 1155347386 M * Bertl two options here, wait a few minutes for vsched-0.03, or add an option which sets this :) 1155347404 M * ekc2 haha. ok. cool 1155347407 M * ekc2 thanks! 1155347472 M * matti ekc2: ;-) Hi there. 1155347473 M * matti ;] 1155347480 M * ekc2 howdy 1155348656 Q * ekc2 1155348672 J * ekc2 ~EKC@netblock-66-245-252-180.dslextreme.com 1155348680 M * Bertl wb ekc2! 1155348686 M * Bertl ekc2: http://vserver.13thfloor.at/Experimental/TOOLS/vsched-0.03.tar.bz2 1155348690 M * ekc2 trying vsched-0.3 now 1155348741 M * ekc2 works like a charm. thanks! 1155348752 M * Bertl you're welcome! 1155348786 M * ekc2 also, is there any way to have idle time advance when there are no active guest processes, as opposed to looking at host processes as well? 1155348863 M * Bertl no, that doesn't make much sense, because a) the cpu isn't idle in that case, so no cpu time for guests, and b) it would be hard to tell them apart, for the scheduler there are only processes 1155348896 M * Bertl i.e. you would have to walk all processes on the run queue every time and check if they are 'host' processes or not 1155348914 M * Bertl would be possible, but I do not see a gain there 1155348956 M * ekc2 ok. that makes sense 1155349064 J * NiTrO|Tel ~NiTrO@ALyon-152-1-142-175.w86-209.abo.wanadoo.fr 1155349074 M * Bertl welcome NiTrO|Tel! 1155349084 M * NiTrO|Tel :) 1155349130 P * NiTrO|Tel 1155349585 J * skram ~skram@the.telephreak.org 1155349622 M * Bertl wb skram! 1155349686 M * skram thanks 1155349691 M * skram uh 1155349697 M * skram im really confused abotu something 1155349704 M * skram how can i get the ram usage of a vps 1155349724 M * skram i used to have a script to find it.. now i am on 2.6.17.7 and it just seems wtong 1155349730 M * skram *wrong 1155349739 M * skram Linux hercules 2.6.17.7-vs2.0.2-rc27 #1 SMP Wed Aug 9 12:37:15 CDT 2006 i686 Intel(R) Xeon(TM) CPU 3.20GHz GenuineIntel GNU/Linux 1155349753 M * skram vserver 0.30.210 -- manages the state of vservers 1155349754 M * skram This program is part of util-vserver 0.30.210 1155349760 M * Bertl well, /proc/virtual//limits should show 1155349793 M * skram RSS="`cat /proc/virtual/$XID/limit | grep RSS | awk '/RSS/ { printf "%.1f %u", $2 / 256, $4 / 256; }'`" 1155349797 M * skram is what my script does 1155349809 M * skram CURRENT, LIMIT, Hit, Observed Max 1155349837 M * Bertl format changed slightly, double check 1155349845 M * skram thats probably why 1155349845 M * skram hehe 1155349848 M * Bertl we have min/max observed now too 1155349855 M * skram Oh!!! 1155349855 M * skram damn 1155349861 M * skram i totally forget how to use awk 1155349872 M * skram want me to post my script and someone could maybe help me? 1155349901 M * skram :? 1155349980 M * skram http://paste.linux-vserver.org/238 1155349995 M * Bertl what is the $2/256, $4/256 for? 1155350015 M * skram it divides the number by 256 so it gets the ram in megabytes 1155350050 M * Bertl well, that is only correct on certain archs ... 1155350068 M * skram such as? 1155350081 M * skram cat /proc/virtual/17661/limit | grep RSS 1155350082 M * skram RSS: 4436 8852 -1 0 1155350087 M * skram that changed from what? 1155350113 M * Bertl such as those which have a page size of 4k I'd say 1155350122 M * Bertl it is now like this: 1155350146 M * Bertl RSS: 667 0/ 3962 -1/ -1 0 1155350163 M * skram why is mine still the old version? and if so.. my script should still work 1155350180 M * Bertl no, your script looks like it uses the new format :) 1155350187 M * skram new as in? 1155350197 M * Bertl as in the one I pasted here :) 1155350203 M * skram on both my servers i see 5 columns.. 4 of data 1155350211 M * skram weird.. then did it revert back? 1155350216 M * skram yeah, actually you are right 1155350223 M * skram so why did it change when i UPGRADED kernels? 1155350240 M * Bertl hmm, maybe you switched from a devel kernel 2.1.x to a stable one 2.0.2? 1155350259 M * skram i was on 2.0.1 1155350264 M * skram and am now on 2.0.2 1155350271 M * Bertl sure about that 2.0.1? 1155350290 M * skram drwxr-xr-x 19 root root 1368 May 13 19:53 linux-2.6.14-vserver-2.0.1 1155350309 M * Bertl check the makefile, if that concurs ... 1155350320 M * skram make file for the kernel? 1155350320 M * skram okay 1155350330 M * Bertl yep, should show an extra version 1155350341 M * skram EXTRAVERSION = -vs2.0.1-gentoo 1155350371 M * Bertl interesting .. well, I guess that's a bug for the gentoo folks then 1155350377 M * skram okay 1155350389 M * Bertl I do not remember that we had this format in stable, especially not 2.0.1 1155350389 M * skram so.. right now i guess i will just need to adapt to what /proc is showing me 1155350402 M * Bertl yep, make that $4 a $3 1155350406 M * skram yeah 1155350423 M * skram thats not all though 1155350494 M * skram is it? 1155350606 M * Bertl well, it would give you similar results to the 'other' script, why? 1155350649 M * Bertl I don't see a hit or observed max in that line 1155350654 M * skram $4 to $3 1155350670 M * Bertl s/observed max/limit/ 1155350670 M * skram and then REACHED=".." to $4 right? 1155350688 M * Bertl ahem, we are talking about different scripts I guess 1155350703 M * Bertl I was talking about the one-liner you pasted here 1155350780 M * Bertl old format is $2=current, $3=max-observed, $4=limit, $5=hits 1155351100 M * ekc2 I'm resyncing a RAID array on the host, and idle time is stuck at 0 for all the guests even though all guests are 'on hold'. Is this correct behaviour? 1155351119 M * waldi good morning 1155351131 A * waldi should finaly test the bind-mount patch .... 1155351137 M * Bertl ekc2: kind of, unless your raid sync threads are limited 1155351154 M * ekc2 they are not limited 1155351157 Q * skram Ping timeout: 480 seconds 1155351175 M * Bertl so the guests will only get what they are assigned then 1155351181 M * Bertl i.e. no idle benefits 1155351183 M * ekc2 so this brings me back to my motivation for asking about differentiating between guest processes and host processes 1155351200 M * ekc2 I guess I'm trying to use idle time for something it was not intended for 1155351213 J * Skram ~mark@HERCULES.sentiensystems.net 1155351214 M * Skram sorry 1155351223 M * Skram i was using someone else's shell server 1155351230 M * Skram Bertl: will ya still help me out? 1155351234 M * Bertl ekc2: I don't think the 'host is busy' scenario is that typical, at least I don't help so 1155351238 M * Bertl *hope 1155351247 M * Skram well.. i think its working.. but the ram usage makes no sense 1155351267 M * Bertl Skram: let's see the 'original' values and your results 1155351291 M * ekc2 bertl: unforuntately, it's typical for my use case. i guess i'll have to use the non-idle bucket and/or play with limiting raid rhreads 1155351306 M * Skram Bertl: its showing what is shown in free -m via the guest (after entering) 1155351310 M * Skram but i dont think its right 1155351314 M * Skram it just.. doesnt make sense 1155351329 M * Skram my Limit column isnt right 1155351359 M * Skram column 3 of limits is limit? 1155351380 M * Bertl ekc2: so you constantly have raid sync running? isn't that a bottle neck? 1155351408 M * Bertl ekc2: it might be a good idea to limit that to one CPU on a multi cpu system then 1155351429 M * Bertl Skram: old format, yes, -1 means unlimited 1155351436 M * Skram pl 1155351437 M * ekc2 bertl: not constantly. but i have a large raid array and disks fail regularly. so, i don't want guests to stall on disk failure 1155351438 M * Skram *ok 1155351463 M * ekc2 bertl: good point. i'll look into putting some limits on raid sync 1155351664 M * Skram current and limit are working.. 1155351671 M * Skram and observed max seems to be too 1155351861 M * Skram brb 1155351862 Q * Skram Quit: leaving 1155352109 J * Skram ~Mark@hermes.sentiensystems.com 1155352117 M * Bertl wb :) 1155352122 M * Skram thanks 1155352124 M * Skram golly 1155352127 M * Skram okay 1155352141 M * Skram if i total up RSS for all vpses.. it doesnt equal what free -m on the host says 1155352243 M * Skram is that norma? 1155352255 M * Skram also i think vpses are taking more ram than RSS is reporting 1155352268 Q * weasel Quit: brb 1155352291 J * weasel weasel@asteria.debian.or.at 1155352598 Q * Bertl Ping timeout: 480 seconds 1155352637 M * ekc2 hrmm. all of the processes on the host are idle, but idle time is still not advancing on the guest: cpu 0: 711 13 253392 5624 0 HI 22 50 500 1/1000 50/300 1155353118 J * virtuoso ~s0t0na@shisha.spb.ru 1155353243 Q * DreamerC Quit: leaving 1155353327 J * DreamerC ~dreamerc@59-112-7-8.dynamic.hinet.net 1155353390 Q * DreamerC 1155353473 J * DreamerC ~dreamerc@59-112-7-8.dynamic.hinet.net 1155353652 J * Bertl herbert@IRC.13thfloor.at 1155353680 M * Bertl oops, seems like I timed out somehow 1155353693 M * Skram oh noes 1155353712 M * Bertl checking irc logs now ... 1155353722 M * Skram 00:59 < ekc2> hrmm. all of the processes on the host are idle, but idle time is still not advancing on the guest: cpu 0: 711 13 253392 5624 0 HI 22 50 500 1/1000 50/300 1155353730 M * Skram that was all that was said (in here), Bertl 1155353800 M * Bertl ekc2: are you sure you did enable the idle time feature? 1155353820 M * Bertl ekc2: also sure that the guests in question have the flag set? 1155353833 M * Bertl hey virtuoso! 1155353852 M * Bertl Skram: ok, tx! 1155353901 M * Skram sure 1155354499 M * waldi hrm 1155354565 M * waldi next try, I should use the correct kernel ... 1155354680 M * Skram :) 1155354736 M * waldi but my patch does not work ... 1155354791 M * waldi ah, i should apply it also ... 1155354838 M * Bertl indeed a good idea :) 1155355860 P * meandtheshell 1155356250 M * waldi okay, works 1155356262 M * Bertl excellent! 1155356319 M * waldi http://194.39.182.225/linux/vserver-bindmount.patch 1155359881 M * Bertl good night folks! I'm off to bed .. have a good one everyone! 1155359906 N * Bertl Bertl_zZ 1155360863 M * virtuoso Hey Bertl_zZ ! 1155363043 Q * ekc2 Remote host closed the connection 1155363374 Q * sladen Ping timeout: 480 seconds 1155365628 Q * matti Ping timeout: 480 seconds 1155365710 J * dna ~naucki@p54BCD49C.dip.t-dialin.net 1155367794 J * bonbons ~bonbons@83.222.36.236 1155370111 J * rgl Rui@217.129.151.190 1155370112 M * rgl hi 1155370149 M * Skram Hello 1155371602 Q * s0undt3ch Read error: Connection reset by peer 1155372525 J * Viper0482 ~Viper0482@p54975FA2.dip.t-dialin.net 1155372878 Q * Viper0482 Quit: one day, i'll find this peer guy and then i'll reset his connection!! 1155372952 J * Viper0482 ~Viper0482@p54975FA2.dip.t-dialin.net 1155373292 Q * mnemoc Ping timeout: 480 seconds 1155373665 J * mnemoc ~amery@kilo105.server4you.de 1155374160 J * pisco ~pampel@p50878881.dip0.t-ipconnect.de 1155375357 Q * Viper0482 Remote host closed the connection 1155375377 J * Viper0482 ~Viper0482@p54975FA2.dip.t-dialin.net 1155375426 P * Viper0482 1155375666 J * s0undt3ch fbcyjio@bl7-247-47.dsl.telepac.pt 1155375681 M * phedny s0undt3ch: who are you? :P 1155375724 M * rgl a tuga! :D 1155375730 A * rgl is not aloneee 1155375736 M * s0undt3ch aparently another one :) 1155375754 M * s0undt3ch phedny: why!? 1155375769 M * phedny 11:41:06 -!- s0undt3ch is "Why don't you just ask me?" on #vserver 1155375773 M * phedny you realname ;) 1155375809 M * rgl some tugas are afraid of the dark ;) 1155375816 M * s0undt3ch Pedro 1155375832 M * s0undt3ch I live at night so, no, I'm not afraid of the dark 1155375845 M * s0undt3ch I am however carefull with too much information :) 1155375847 M * phedny afraid of the light then? 1155375888 M * s0undt3ch k, that "live at night" thing wasnt that correct, I also have to work with the daylight :) 1155375902 M * phedny so afraid of the sleep it will be... 1155376549 J * debugger_ Rui@217.129.151.190 1155376839 Q * mnemoc Ping timeout: 480 seconds 1155377020 Q * rgl Ping timeout: 480 seconds 1155377190 J * mnemoc ~amery@kilo105.server4you.de 1155377494 J * me1 ~mef@c-68-39-177-97.hsd1.nj.comcast.net 1155377501 P * me1 1155378006 J * Wonka produziert@chaos.in-kiel.de 1155378405 J * sladen paul@starsky.19inch.net 1155379672 J * LiNuX[a] ~Staff@85.104.63.65 1155379934 Q * LiNuX[a] Quit: < Klavye [ 5.FinaL ] > [ http://www.klavye.web.tr ] - irc.klavyescript.org - Lee 1155380189 M * Hollow mnemoc: around? 1155380236 M * mnemoc Hollow: hi 1155380316 M * Hollow mnemoc: hi :) http://svn.linux-vserver.org/viewvc.py/vcd/trunk/src/vxpasswd/ 1155380322 M * mnemoc :D 1155380712 M * mnemoc so there is no vserver-utils anymore? it was replaced by three independent packages? 1155380738 M * Hollow right 1155380779 M * Hollow should solve the confusion, and ease for downstream maintainers 1155380786 M * mnemoc cool :) 1155380842 M * Hollow also vstatd and vwrappers should work with util-vserver too 1155380851 M * Hollow at least in theory, didn't test everything yet 1155381064 M * mnemoc have you tested on 2.0.2-rclast ? 1155381104 M * Hollow no, vcd needs at least 2.1.1_rcsomething 1155381136 M * Hollow and once it is released it will need 2.2.0 1155381243 M * mnemoc :( 1155381264 M * Hollow many features in 2.1.1 were introduced because of vcd ;) 1155381312 M * mnemoc =) 1155381355 Q * |coocoon| Quit: KVIrc 3.2.0 'Realia' 1155381357 N * debugger_ rgl 1155381360 M * rgl Hollow, cool :) 1155381373 M * rgl Hollow, btw, just looked at your code, and saw this: asprintf(&sqlbuf, "UPDATE user SET password = '%s' WHERE name = '%s'", 1155381373 M * rgl sha1_password, username); 1155381382 M * rgl I hope you'll change that code ;) 1155381405 M * rgl or better, I hope that there is no code like that in other places :) 1155381428 M * mnemoc if username is sanitized, there is no problem :p 1155381474 M * rgl but it isn't. 1155381482 M * mnemoc yet :) 1155381487 M * rgl yes :) 1155381522 M * mnemoc at least until you send the patch 1155381528 M * rgl anyways, you don't want to only rely on sanitize :) 1155381662 M * Hollow ah yes.. the xmlrpc methods have validation functions... 1155381693 M * Hollow but well, i guess the admin will not do sql injection :P 1155381781 M * rgl intentionally, I hope not :) 1155382227 M * Hollow rgl: fixed 1155382373 M * rgl Hollow, cool :) 1155382559 J * ekc ~ekc@netblock-66-245-252-180.dslextreme.com 1155383193 J * coocoon ~coocoon@p54A05492.dip.t-dialin.net 1155383515 Q * michal` Quit: chilling out 1155383530 J * michal_ ~michal@www.rsbac.org 1155383543 M * michal_ hey :) 1155383749 J * debugger_ Rui@217.129.151.190 1155383987 J * daniel15 ~dansoftau@220-244-251-210.tpgi.com.au 1155384012 M * daniel15 Linux-VServer is awesome! :) 1155384017 M * FaUl yes, indeed 1155384043 M * daniel15 I previously tried to install OpenVZ (couldn't get the kernel working), and Xen (couldn't get networking to work) 1155384054 M * daniel15 VServer was the only one that worked for me 1155384071 M * daniel15 And even better, there's a Debian package for it (no need to compile kernel) 1155384074 M * daniel15 :D 1155384188 M * daniel15 The only problem I ran into was I needed to add some rules to my IPTABLES firewall (it was too restrictive, and didn't allow data from the VServers to go to the internet) 1155384205 Q * rgl Ping timeout: 480 seconds 1155384457 M * daniel15 Anyways, talk to you guys later 1155384462 Q * daniel15 Quit: Bye! 1155384552 Q * DreamerC Quit: leaving 1155384598 J * DreamerC ~dreamerc@59-112-7-8.dynamic.hinet.net 1155384745 M * ekc Will having CONFIG_VSERVER_IDLELIMIT enabled prevent vserver idle-time scheduling from working? 1155385585 J * Viper0482 ~Viper0482@p54975FA2.dip.t-dialin.net 1155388660 Q * olilo Quit: brb 1155389147 Q * DreamerC Quit: leaving 1155389182 J * DreamerC ~dreamerc@59-112-7-8.dynamic.hinet.net 1155389199 M * mnemoc OT: do you know any good c library for invoquing remote methods using binary communications? (i.e. non-xml) 1155389437 J * olilo hiddenserv@tor.noreply.org 1155389524 N * Bertl_zZ Bertl 1155389537 M * Bertl good morning folks! 1155389603 J * meandtheshell ~markus@85-124-175-201.dynamic.xdsl-line.inode.at 1155389615 M * Bertl wb meandtheshell! 1155389630 M * mnemoc good morning Bertl 1155389681 M * Bertl ekc: no, having it _enabled_ makes it possible 1155389726 M * Bertl mnemoc: hey, I remember you had a vserver question yesterday, is it still unanswered? 1155389747 M * mnemoc .oO 1155389755 M * mnemoc ah, about the barrier? 1155389772 M * mnemoc at least i didn't see an answer 1155389787 M * Bertl Hollow: I did an update to the logos, please check the usual location for fixed size and favicon.ico :) 1155389823 M * Bertl mnemoc: okay, so the Q was what happens with the barrier when you have separate mounts per guest, right? 1155389834 M * mnemoc yes 1155389858 M * Bertl well, the rule basically stays the same, make sure that /path/to/guest/.. carries a barrier 1155389870 M * Bertl note that the '..' is literally 1155389899 M * Bertl the barrier itself might end up on different filesystems then 1155389901 M * mnemoc so i have to make them a parent directory 1155389928 A * mnemoc is quite slow today 1155389938 M * meandtheshell Bertl: hello everybody 1155389939 M * Bertl if your mount point is /x for example, then /x/.. probably is the root dir 1155389963 M * Bertl mnemoc: but certain filesystems have other semantics regarding parent dir 1155389968 M * mnemoc ah, even if it's a differernt filesysmte 1155389975 M * mnemoc reiser :( 1155390012 M * Bertl yep, you have to check that when you are inside a guest, '..' (whatever that might be) carries a barrier 1155390041 M * mnemoc ic 1155390124 M * mnemoc should lsattr on the guest show the barrier? 1155390181 Q * DreamerC Quit: leaving 1155390190 M * mnemoc ------------- /.. :\ 1155390210 J * DreamerC ~dreamerc@59-112-7-8.dynamic.hinet.net 1155390409 M * Bertl mnemoc: you did say reiser, yes? try mounting with 'attrs' option 1155390439 M * Bertl reiser is the only (linux) filesystem which requires a mount option for ext attributes 1155390490 M * renihs hi Bertl! :) 1155390498 M * Bertl hey renihs! 1155390604 M * renihs hmm say you happen to know when *next* linuxtag is? somehow i cant figure that 1155390607 M * renihs :) 1155390616 A * renihs hides in corner 1155390788 M * Hollow Bertl: ok, updated in the wiki 1155390815 M * Hollow uf 1155390818 M * Hollow the logo is too big 1155390831 M * Hollow one sec.. 1155390845 M * Bertl it comes in all sizes, just use the smaller one 1155390865 M * Hollow i used the smallest 1155390870 M * Bertl *t? 1155390873 M * Hollow yes 1155390881 M * Bertl hmm .. interesting :) 1155390911 M * Bertl okay, should I do an even smaller one? 1155390921 M * Hollow i'm at it already 1155390965 J * rgl ~Rui@217.129.151.190 1155390969 M * Bertl Hollow: I'm using some trickery to get the scaling right 1155391029 M * Hollow well.. using the eps in photoshop seems ok ;) 1155391043 M * Hollow but one cannot read (r) and (tm) anymore in this size 1155391065 M * Bertl who uses photoshop? :) 1155391089 M * Hollow i do 1155391093 M * Hollow all linux graphic apps suck 1155391127 M * rgl Bertl, hey :D 1155391190 M * Bertl hey rgl! 1155391193 M * Hollow Bertl: maybe you can do one in small size with these symbols still readable? 1155391205 M * Bertl Hollow: try the *x one :) 1155391223 M * rgl Bertl, I did some experiment for testing quota in ruby, its at http://ruilopes.com/tmp/test_vserver_quota.tgz, its just a simply one, just tests the more basic stuff I could find. the only basic stuff that I can think its missing is the xid stuff, but I didn't try that yet. 1155391244 M * Hollow Bertl: it's exactly the same size 1155391265 M * Bertl oops, sec 1155391293 M * rgl Bertl, the file you want to see is the test_quota_on_host.rb 1155391364 M * Bertl Hollow: reload 1155391402 M * Bertl rgl: okay, do you have some output to look at (i.e. what does it show?) 1155391415 Q * debugger_ Ping timeout: 480 seconds 1155391432 M * rgl Bertl, 11 tests, 22 assertions, 0 failures, 0 errors :D 1155391461 M * Bertl okay, so no details atm? what if something fails? it then simply bails out? 1155391493 M * rgl Bertl, it doesn't have any output, if it fails, you will get a message telling you where it failed 1155391517 M * Hollow Bertl: ok, now it looks good 1155391540 M * Bertl rgl: okay, I'd prefer to get results of _all_ tests even when it fails 1155391548 M * Bertl rgl: is that possible in the future? 1155391561 M * rgl like, if the test creates a file with 2K, and it doesn't get accounted you'll get an error like: 1155391563 M * rgl 1) Failure: 1155391564 M * rgl test_simple_file_quota(TestQuotaOnHost) 1155391564 M * rgl [./vserver_test_helper.rb:197:in `assert_quota_diff' 1155391564 M * rgl ./vserver_test_helper.rb:172:in `assert_user_quota_diff' 1155391564 M * rgl test_quota_on_host.rb:10:in `test_simple_file_quota']: 1155391566 M * rgl <[4126, 0, 0, 3, 0, 0]> expected but was 1155391568 M * rgl <[4127, 0, 0, 3, 0, 0]>. 1155391579 M * Bertl thing is, usually the 'errors' show some pattern which simplifies the bug hunt 1155391594 M * Bertl i.e. all allocations are off by 1 e.g. 1155391610 M * Bertl the expected/was is good 1155391634 M * rgl Bertl, it run all tests, if some fail, it will output the failure, and moves on to the next test 1155391649 M * Bertl ah, excellent, so I just got that wrong ... 1155391661 M * rgl (I'm using the std ruby test/unit library) 1155391696 M * Bertl good, and I assume it passed your tests with quota, right? 1155391805 Q * pisco Ping timeout: 480 seconds 1155391831 M * rgl Bertl, humm yes. but the tests I've made are simple one. you can seem what they do in the comments on that test_quota_on_host.rb file 1155391898 M * rgl mainly because I don't known how to subvert the quota mechanism :/ 1155391917 M * rgl like, test its corner cases 1155391933 M * rgl corner/boundary 1155391996 M * Bertl could you do me a favor and try it on a few filesystems, namely ext2/3, jfs, reiserfs and xfs? 1155392020 M * Bertl you can probably make good use of the basic filesystem setup in testfs.sh 1155392078 M * Bertl rgl: this already seems to do quite a nice job, we will (if you help me :) adjust/extend it as we go 1155392114 M * Bertl what do I need to execute ruby code? i.e. what libraries/binaries are required (minimal)? 1155392151 M * rgl I've used the ruby 1.8.2 that ships with debian 3.1 (which is quite old) 1155392153 Q * bubulak Ping timeout: 480 seconds 1155392172 M * rgl its the ruby package 1155392181 M * Bertl rgl: okay, that contains (roughly)? 1155392215 M * Bertl or let me put this the other way round, I have a test image (32MB in size), what do I need to add to that to hae it run? 1155392217 M * rgl what do you mean? it contains the interpreter, ruby, and the std ruby libraries :D 1155392231 J * matti matti@linux.gentoo.pl 1155392234 M * Bertl or can I 'compile' it somehow to plain executable? 1155392265 J * bubulak ~bubulak@whisky.pendo.sk 1155392273 M * Bertl Hollow: yeah, looks quite nice 1155392281 M * Bertl wb matti! bubulak! 1155392441 M * rgl compiling ruby is straitforward (like any normal nix program), but its not a single executable, its a executable and a horde of files , see http://packages.debian.org/stable/libs/libruby1.8 1155392472 M * waldi Bertl: can you take a look at http://194.39.182.225/linux/vserver-bindmount.patch? 1155392478 M * Bertl rgl: hmm, okay, how much does that make in size? 1155392483 M * Bertl waldi: sure, sec 1155392494 M * waldi hmm, it lacks RBIND 1155392509 M * waldi i think this is handled different ... 1155392537 M * Bertl ah, and it fixes some kind of bug I presume? 1155392550 M * Bertl i.e. you can get 'dev' mounts with bind mounting? 1155392561 M * rgl Bertl, ~6MB 1155392564 M * waldi yes 1155392573 M * Bertl rgl: *cough*, thanks :) 1155392596 M * rgl Bertl, but I guess you can trim, I never did, thou :/ 1155392609 M * Bertl okay, will give it a try ... 1155392622 M * rgl its +- the size of python 1155392648 M * rgl why is size a constraint? 1155392661 M * Bertl rgl: just that you get an idea, here is the test disk (complete boot system with a test guest included :) http://vserver.13thfloor.at/Stuff/QEMU/TEST_32M_public2.img.bz2 1155392685 M * Bertl it's 8.4M compressed, and a 32MB disk :) 1155392705 Q * DreamerC Quit: leaving 1155392732 M * Bertl waldi: could/did you test that with the rbind? 1155392741 J * DreamerC ~dreamerc@59-112-7-8.dynamic.hinet.net 1155392742 M * waldi just tested, seems so work also 1155392756 M * Bertl rgl: but it's not important atm, just would be nice to fit it in there 1155392783 M * Bertl rgl: there is a patch which changes from current quota to quota hashes (only one hash actually for now) 1155392785 M * rgl Bertl, whats the freespace on those 32? :D 1155392807 M * Bertl rgl: something about 6 or so 1155392835 M * Bertl rgl: but as I said, it's not top priority ... 1155392891 M * Bertl I'll update the patch tomorrow (to current releases) could you do some testing with and without on beforementioned filesystems then? 1155392929 M * waldi (now i can finaly build chroots within a vserver without external help ...) 1155392998 M * rgl Bertl, sure. I'm currently only doing the tests on a loop mounted ext3 file, I guess creating the other fs is more-or-less the same. 1155393041 M * Bertl waldi: hmm, but you are probably compromising the host, you know that? 1155393054 M * waldi Bertl: how? 1155393071 M * waldi i can only bind mount things i'm already able to see 1155393072 M * Bertl rgl: testfs.sh has all the details, probably you just have to 'plug' in the new checks 1155393092 M * waldi hmm, the source mount may be nodev already 1155393094 M * Bertl waldi: with BME bind mounts can have fatal powers 1155393107 M * waldi hmm? 1155393127 M * Bertl just try to mount something without nodev, then --bind mount it with nodev 1155393142 M * Bertl and now think about the other way round :) 1155393152 M * waldi yeah, just realized that 1155393164 M * waldi is it possible to check the flags of the source mount? 1155393174 M * Bertl what I wonder is, what is the actual problem you are trying to solve? 1155393178 M * waldi if it is mounted without nodev, it should be save 1155393202 M * waldi i want to be able to build chroots for buildds 1155393217 M * Bertl okay, fine, where do they fail? 1155393226 M * waldi they need a working /dev 1155393239 M * Bertl well, if you 'allow' that, you have already lost 1155393271 M * Bertl working /dev means you can create device nodes, which in turn means, you can screw the host 1155393274 M * waldi no 1155393290 M * waldi the vserver have already a /dev 1155393303 M * Bertl ah, so you want to --bind mount it into that guest 1155393306 J * |coocoon| ~coocoon@p54A068C3.dip.t-dialin.net 1155393309 M * Bertl s/guest/chroot/ 1155393311 M * waldi yes 1155393318 M * Bertl okay, that's something else 1155393333 Q * |coocoon| 1155393335 M * Bertl so, you might want to keep the nodev if it is already there 1155393346 M * Bertl but allow 'dev' if it is not, right? 1155393358 M * waldi yes 1155393379 M * Bertl so you might want to move that check/option into the actual bind/move mounts 1155393396 M * Bertl (will need some adjustments there) 1155393412 J * |coocoon| ~coocoon@p54A068C3.dip.t-dialin.net 1155393430 M * Bertl wb |coocoon|! 1155393447 M * Bertl okay, I'm off for now, have to do some house cleaning ... back later ... 1155393453 N * Bertl Bertl_oO 1155393463 Q * coocoon Ping timeout: 480 seconds 1155393534 Q * |coocoon| 1155394032 Q * DreamerC Quit: leaving 1155394055 J * DreamerC ~dreamerc@59-112-7-8.dynamic.hinet.net 1155394096 M * waldi Bertl_oO: hmm, is it a bad idea to always force bind mounts to have a subset for secure mounts? 1155394160 M * waldi hmm, it is 1155394532 M * waldi Bertl_oO: i updated the patch. i'm currently testing it 1155395204 J * mire_ ~mire@20-166-222-85.COOL.ADSL.VLine.verat.net 1155395485 J * coocoon ~coocoon@p54A068C3.dip.t-dialin.net 1155395700 Q * DreamerC Quit: leaving 1155395737 J * DreamerC ~dreamerc@59-112-7-8.dynamic.hinet.net 1155395740 Q * mnemoc Ping timeout: 480 seconds 1155396116 J * mnemoc ~amery@kilo105.server4you.de 1155396271 Q * olilo Remote host closed the connection 1155396330 J * olilo hiddenserv@tor.noreply.org 1155397096 M * waldi be works 1155397170 M * waldi Bertl_oO: works 1155397402 J * _fs fs@213.178.77.98 1155397403 Q * fs Read error: Connection reset by peer 1155398167 J * debugger_ Rui@217.129.151.190 1155398610 Q * rgl Ping timeout: 480 seconds 1155398619 N * debugger_ rgl 1155399369 M * bonbons Hollow: ping 1155399459 M * Hollow bonbons: pong 1155399486 M * bonbons Hollow: question concerning vserver x86 profiles in portage... 1155399510 Q * ekc Remote host closed the connection 1155399544 M * bonbons what's the most up-to-day of them, I've found x86/2005.1/vserver qnd x86/vserver/ but none in x86/2006.0 1155399559 M * Hollow use x86/vserver 1155399581 M * Hollow the idea with one per 200x.x was just overkill 1155399625 M * matti Hollow: :)) 1155399657 M * Hollow hey matti ;) 1155399693 M * matti Hollow: What you think about -mm tree? 1155399706 M * Hollow never used it 1155399717 M * matti Hollow: I saw latest patch from Andrew... I am stooned. 1155399722 Q * mire_ Quit: Leaving 1155399750 M * Hollow there are no -mm rediffs for vserver, so i can't use it ;) 1155399830 M * bonbons Hollow: ok, would it make sense to add a sub-profile to vserver with server-oriented default use-flags as is being done for 2006.1? 1155399865 M * matti Hollow: Yep. 1155399885 M * matti Hollow: And -mm patch is really heavy ;) 1155399885 M * Hollow bonbons: hm, let me look 1155399888 M * matti Hollow: 8 MB ;p 1155399957 M * Hollow bonbons: well, you can copy the vserver sub-profile to any other (sub-)profile, and it will most likely work.. 1155399978 M * bonbons Hollow: they are somewhere between both currently I would say (things like xorg, pppd, isdnlog, maybe the ttf-fonts should probably be "removed") 1155399999 M * Hollow the question is, if it would make more sense to create a server sub-profile in the vserver profile 1155400011 M * Hollow because one vserver profile per 200x.x is overkill as i said 1155400023 M * bonbons Hollow: but that copy/move makes sync hard 1155400085 M * bonbons well, vserver can be either used in desktop are (e.g. one guest per dm, like vs-kde, vs-gnome, vs-e, ...; think multi-seet) or as servers 1155400171 M * Hollow well, i donÄt have a problem if the vserver profile is copied to desktop/server 1155400184 M * Hollow have to talk with the other profile maintainers first though 1155400193 M * Hollow or you can do if you want 1155400205 M * Hollow or probably file a bug to the x86 arch team and cc vserver devs 1155400362 M * bonbons not sure what is best... the server profile is already bloated, looking into 2006.1/server, much to bloated for vserver where the aproach is more like 1 service per guest 1155400398 M * bonbons for desktop profile a copy would probably be fine 1155400503 M * Hollow well, as i said.. imo it would make more sense to create sub profiles in vserver... there is not so much change going on in the profile, so sync from time to time shouldn't be too hard 1155400735 M * bonbons vserver{/desktop} or vserver/{server,desktop} is also the direction I would prefer, especially taking into account the few things that are "forbidden" in guests 1155400897 M * Hollow yeah, i don't have the time currently, but if you can prepare something i can take a look and we can fine-tune it.. 1155400918 M * Hollow quite busy with vcd currently ;) 1155400936 M * Hollow there are new svn ebuilds in our overlay btw, if you want to test.. 1155400960 M * bonbons No problem :-) I still have to get a deeper look into vcd! I can look at it, will keep you informed 1155401002 M * Hollow off to dinner now, cu later! 1155401015 M * matti Hollow: Enjoy dinner ;] 1155401030 M * bonbons same from me 1155401843 J * ekc ~ekc@netblock-66-245-252-180.dslextreme.com 1155402443 Q * ekc Remote host closed the connection 1155402563 J * ekc ~ekc@netblock-66-245-252-180.dslextreme.com 1155403088 M * ekc what does the new config option CONFIG_VSERVER_PRIVACY do? 1155405357 J * debugger_ Rui@217.129.151.190 1155405815 Q * rgl Ping timeout: 480 seconds 1155407874 Q * ekc Remote host closed the connection 1155408285 Q * yang Quit: KVIrc 3.2.0 'Realia' 1155408722 Q * debugger_ Quit: Fui embora 1155409475 J * ekc ~ekc@netblock-66-245-252-180.dslextreme.com 1155411328 Q * coocoon Ping timeout: 480 seconds 1155411917 J * coocoon ~coocoon@p54A05D97.dip.t-dialin.net 1155412278 M * PowerKe µ 1155412279 M * PowerKe ù$ 1155415129 Q * coocoon Quit: KVIrc 3.2.0 'Realia' 1155415294 J * coocoon ~coocoon@p54A05D97.dip.t-dialin.net 1155416788 Q * Savvy Ping timeout: 480 seconds 1155417158 M * mnemoc Hollow: lucid ? 1155417186 M * Hollow yes, i made it a shared lib.. it was included in vserver-utils before 1155417200 M * Hollow many little helpers 1155417201 M * Hollow ;) 1155417232 M * Hollow http://dev.gentoo.org/~hollow/lucid/doc/modules.html not all are documtented yet 1155417302 M * mnemoc and where can i download it from? :( 1155417324 M * Hollow oh.. http://dev.croup.de/repos/lucid/trunk 1155417326 M * Hollow :) 1155417330 M * mnemoc :) 1155417646 M * mnemoc is your plan to keep lucid as an independent dependency? 1155417837 M * Hollow yep.. i will add all common little helpers to this lib.. 1155417893 M * Hollow it got quite nice over time actually... ;) 1155418015 M * Hollow i'm ioff to bed now.. cu tomorrow 1155418030 M * mnemoc gn8 Hollow :) 1155418062 M * mnemoc Hollow: and please add it a README with the basic info so i can package it :p 1155418557 Q * olilo Remote host closed the connection 1155418600 J * olilo hiddenserv@tor.noreply.org 1155419302 M * bj can anyone give me a hint how I can allow a application (in this case postfix-policyd) the chroot in a vserver cage ? 1155419502 J * alexx ~alexx@proxy.ikse.net 1155419909 M * bj hm maybe the vserver isn't the problem ;$ 1155420044 M * mnemoc try with secure_chroot capability 1155420316 M * bj hm this thing wants to enforce some kind of file limit (Current rlimit: ... file descriptors), hmm 1155420335 M * bj mnemoc: that works somehow with vattribute --set-secure or so right ? 1155421815 Q * coocoon Quit: KVIrc 3.2.0 'Realia' 1155421845 M * mnemoc no idea 1155421860 Q * ekc Remote host closed the connection 1155421890 M * bj thanks anyway mnemoc :) 1155421901 M * mnemoc :) 1155423812 J * Savvy ~SavvyChat@81.215.81.173 1155423979 J * ekc ~ekc@netblock-66-245-252-180.dslextreme.com 1155424079 Q * alexx Quit: bye 1155424460 Q * ekc Ping timeout: 480 seconds 1155424957 J * ekc2 ~EKC@netblock-66-245-252-180.dslextreme.com 1155425906 Q * bonbons Quit: Leaving 1155426032 Q * Viper0482 Remote host closed the connection