1142121780 M * Bertl okay, I guess I'm off for today ... will be back tomorrow
1142121801 M * Bertl have a nice whatever everyone ... and cya!
1142121809 M * daniel_hozac good night!
1142121814 N * Bertl Bertl_zZ
1142122006 J * lilalinux__ ~plasma@dslb-084-058-193-186.pools.arcor-ip.net
1142122288 M * waldi find: /usr/lib/util-vserver/check-unixfile terminated by signal 11
1142122385 M * waldi mmap2(NULL, 232, PROT_READ, MAP_PRIVATE, 3, 0x431434) = -1 EFAULT (Bad address)
1142122392 M * waldi hihi
1142122444 Q * lilalinux_ Ping timeout: 480 seconds
1142126717 Q * vrwttnmtu Quit: Leaving
1142129207 Q * lilalinux__ Remote host closed the connection
1142129571 Q * matta Ping timeout: 480 seconds
1142131427 Q * michal` Ping timeout: 480 seconds
1142131818 J * michal` ~michal@www.rsbac.org
1142135244 Q * mnemoc Ping timeout: 480 seconds
1142141492 J * Smutje_ ~Smutje@xdsl-84-44-247-89.netcologne.de
1142141599 Q * Smutje Ping timeout: 480 seconds
1142141599 N * Smutje_ Smutje
1142141779 Q * PerlLaptop Read error: Connection reset by peer
1142146659 M * Hollow daniel_hozac: great, thx!
1142154704 J * pzYsTorM schak@dslc-082-082-166-094.pools.arcor-ip.net
1142155654 J * bonbons ~bonbons@83.222.39.180
1142156116 M * daniel_hozac Hollow: does it look ok? or did i mess it up?
1142156808 M * bonbons Bertl, Hollow: Just run basic start/stop test with v-u 1.0.4[HEAD] on 2.1.1-rc12 and 2.0.2-rc12, working fine
1142156865 M * daniel_hozac bonbons: cool.
1142156909 M * Hollow great!
1142156920 M * Hollow daniel_hozac: didn't test it yet, but by looking at the code it looks good
1142156934 M * bonbons Now I'm looking to adapt testme.sh, for some more testing
1142157005 M * daniel_hozac well, i did some basic testing already, i'm installing utils with vlogin on my laptop right now to do some real testing.
1142157063 M * Hollow oki, let me know then... i'm a bit afk (baking birthday cake for myself ;)
1142157078 M * daniel_hozac is it your birthday today?
1142157081 M * Hollow tomorrow
1142157087 M * daniel_hozac ah.
1142157127 M * daniel_hozac i'll be very afk soon, i might not get around to real life tests until tonight.
1142157222 M * Hollow daniel_hozac: friend of mine jus tested the vlogin patch (it is in gentoo portage already) seems to work fine!
1142157237 M * daniel_hozac wow, you work fast.
1142157242 M * Hollow :)
1142157261 M * Hollow well, it's not good in general to put untested patches into portage, but well..
1142157261 M * Hollow :P
1142157269 M * daniel_hozac hehe.
1142157327 M * daniel_hozac i'll file it in savannah then.
1142157453 M * Hollow daniel_hozac: good... phreak``  also included your uid and bcaps patch
1142157458 M * Hollow in gentoo
1142157523 M * daniel_hozac hehe, you apply my patches faster than i do :)
1142157545 M * Hollow heh
1142158135 M * bonbons where is the most recent/complete testme script? I have one that claims 'V0.15'
1142158143 M * Hollow yup
1142158147 M * Hollow latest
1142158218 M * bonbons hm, not that easy to understand the script :)
1142158507 M * daniel_hozac FYI: https://savannah.nongnu.org/patch/index.php?func=detailitem&item_id=4969
1142158992 M * waldi ah, 2.6.16-rc6 is out ...
1142159946 J * meandtheshell ~markus@85-125-229-158.dynamic.xdsl-line.inode.at
1142162194 Q * phreak`` Ping timeout: 480 seconds
1142162379 J * phreak`` ~phreak``@styx.xnull.de
1142164628 J * phedny ~mark@volcano.p-bierman.nl
1142165489 J * UKLiNUX ~UKLiNUX@82-39-80-136.cable.ubr03.jarr.blueyonder.co.uk
1142165519 P * UKLiNUX 
1142167637 Q * Dr4g Read error: Connection reset by peer
1142168020 M * pzYsTorM # cat /proc/virtual/1001/limit
1142168020 M * pzYsTorM PROC:          155             164              -1           0
1142168020 M * pzYsTorM VM:         316463          323715              -1           0
1142168020 M * pzYsTorM can anyone explain to me the "316463"? the pagesize is 4kb. and the whole server has 512mb ram. how can that be?
1142168065 M * phedny swapping/
1142168154 J * Viper0482 ~Viper0482@p549755C4.dip.t-dialin.net
1142168165 M * pzYsTorM ah ok
1142168178 P * Viper0482 
1142168190 N * Bertl_zZ Bertl
1142168194 M * Bertl morning folks
1142168284 M * Bertl pzYsTorM: it is not even required that your system has swapping to get over the physical ram 
1142168348 M * Bertl pzYsTorM: the default allocation policy in linux gives out arbitrary memory (that's called overcommitment) even if it cannot be put into ram/swap afterwards
1142168374 M * Bertl as long as the pages are not used, no ram is required
1142168444 M * pzYsTorM and the RSS value is the actual occupied memory?
1142168452 M * pzYsTorM +ly
1142168470 M * Bertl yep, that's the RAM used (resident set size)
1142168524 M * pzYsTorM thanks for the enlightenment :)
1142168668 M * pzYsTorM is there a possibility to mount a swapfile inside a vserver? ("swapon: swapfile: Operation not permitted")
1142168688 M * Bertl nope, but that would not make sense ...
1142168710 M * Bertl the host is handling the swapping and such, so you only want to add swap space to the host
1142168846 M * Bertl what currently isn't done, because it adds too much overhead, is limiting the swap space a guest can use ...
1142169467 J * Viper0482 ~Viper0482@p549755C4.dip.t-dialin.net
1142169702 M * pzYsTorM d'oh. thats not good. i have a guest, which uses so much memory, that the host hangs up every few days. so i want to limit swap and memory
1142169833 J * hw ~Hartmut@85-124-100-166.dynamic.xdsl-line.inode.at
1142169852 M * Bertl you can set (with recent devel) an RSS soft and hard limit, which is somewhat compareable to ram and swap limit
1142169856 M * Bertl welcome hw!
1142169872 M * hw servus! funkt soweit alles?
1142169913 M * Bertl hw: hmm, channel language is english :)
1142169918 M * hw sry
1142169920 M * hw :)
1142169937 M * pzYsTorM recent devel means 2.1.x and not 2.0.2-rc12?
1142169969 M * Bertl yup
1142169978 M * pzYsTorM okay
1142170073 M * pzYsTorM do those RSS limits not work with 2.0.2 at all?
1142170122 M * Bertl the hard limit yes, the soft limit isn't present there
1142170146 M * pzYsTorM ah ok.
1142170361 N * Wenix Dopehead_
1142170377 N * Dopehead_ Wenix
1142170652 M * pzYsTorM if i set a hard RSS limit, would it be possible to occupy all the swap?
1142170752 M * Bertl if there is no reasonable VM limit, yes
1142170975 M * Bertl thing is, you problem is a little more complicated than it seems
1142170993 M * Bertl let's assume you have perfect ram and swap limits
1142171009 M * Bertl what would happen to your guest?
1142171032 M * Bertl first it would start using up ram, then swap, then it hits the limits and something _will_ go wrong
1142171082 M * Bertl an application inside your guest is probably leaking memory 
1142171090 M * pzYsTorM yes. thats at the moment. the server load goes to 80.0
1142171111 M * Bertl so, what can be done against that?
1142171123 M * Bertl - you can fix the leakage
1142171133 M * Bertl - you can periodically restart the guest/app
1142171144 M * pzYsTorM at the momentan i dont know the name of the naughty app
1142171176 M * Bertl you should be able to find that with ps auxwww (or vps auxwww)
1142171196 M * pzYsTorM restarting doesnt work, if the load is so high (80.0 or so)
1142171260 M * Bertl well, you can a) limit the processes inside and b) restart earlier
1142171260 Q * marl Read error: Connection reset by peer
1142171268 M * pzYsTorM vps only finds some a handful processes with 4% mem each.
1142171272 M * pzYsTorM -some
1142171275 J * marl ~matt@84.92.193.226
1142171361 M * Bertl pzYsTorM: hmm, have you checked the vm sizes?
1142171511 M * pzYsTorM you mean the /proc/virtual/xxx/limit files?
1142171650 M * phedny what do the numbers on the last line of /proc/virtual/*/sched mean?
1142171653 M * phedny like: cpu 0: 175 146 162
1142171662 M * Bertl pzYsTorM: well, more the vm output column in vps
1142171672 M * Bertl phedny: those are the ticks
1142171770 M * pzYsTorM Bertl: there is only a MEM, a VST and a RSS column
1142171778 M * phedny Bertl: like real,user,sys?
1142171800 M * pzYsTorM VSZ - not VST
1142171816 M * bonbons Bertl: related to sendfile() issue: failed for me on 2.6.16rc5-vs2.1.1-rc10, working on rc12 (x86)
1142171855 M * bonbons I also tested guest start/stop and service restart in guests on vs2.0.2rc12 and vs2.1.1rc12, both working fine
1142171877 M * Bertl phedny: yeah, but it is user,sys and hold
1142171889 M * bonbons Finally I started a port of testme.sh to vserver-utils (1.0.* branch) [same v-u used for my tests above]
1142171934 M * Bertl pzYsTorM: yeah, VSZ is the interesting
1142171991 M * phedny Bertl: thanks, now I see those numbers make sense in my tests :)
1142172001 M * Bertl bonbons: can you follow up to gerardi and report that your apache's sendfile feature is working?
1142172013 M * pzYsTorM Bertl: is this a size in MB or a pagecount?
1142172034 M * Bertl pzYsTorM: that's never sure, you have to consult the manual ...
1142172038 M * bonbons I can do, what the delta-s390-fix01.diff he's referring to?
1142172047 M * Bertl sec
1142172073 M * Bertl http://vserver.13thfloor.at/Experimental/delta-s390-fix01.diff
1142172397 M * pzYsTorM Bertl: it seems to be the pagecount. the apache on the host has 200000 pages (pagesize is 4kb). and then some /bin/php5-fcgi processes on a guest which 30000 pages each. and 5 spamd processes with 30000 pages each. the rest are only little pagecounts.
1142172544 M * pzYsTorM but: if i stop the apache, the mem/swap values on `top` change only little (10mb less)
1142172585 M * Bertl how much swap do you have?
1142172630 M * pzYsTorM 500mb partition and a temporary 1gb swapfile (to avoid those hang ups)
1142172667 M * Bertl well, if your numbers are right, then apache (host) is using 800MB VM and each of your SPAMDs has another 120M
1142172714 M * Bertl not accounting the fcgi processes
1142172721 M * pzYsTorM very much i think. and there a two apache childs. both have ~200000
1142172737 M * Bertl which is very likely shared to some extend
1142173045 M * phedny does a page exist on the site explaining the values in several /proc/virtual/* files?
1142173058 M * phedny since I can't seem to find one
1142173103 M * Bertl not yet, it is kind of work in progress
1142173128 M * Bertl phedny: but feel free to work on that, let me dig out the apge which is supposed to show that
1142173261 M * Bertl http://linux-vserver.org/HowTo+Read+ProcFS
1142173308 M * Bertl it's also still in flux as we didn't decide if it is a good interface (proc) for user communication
1142173402 M * phedny when I find some spare time, I'll try to update the page a little
1142173505 M * Bertl feel free to make suggestions for useful proc data/format too
1142173579 J * liquid3649_ ~Viper0482@p5497667A.dip.t-dialin.net
1142173765 J * doener ~doener@i5387DC93.versanet.de
1142173778 M * phedny Bertl: I think they're nice like they are, the values can be processed with a shell script and are human readable (except maybe for column names)
1142173787 M * Bertl welcome doener!
1142173796 M * doener hi!
1142173829 M * Bertl doener: got some spare time?
1142173858 M * doener about how much time are we talking? ;)
1142173870 M * Bertl maybe a few minutes if you are good :)
1142173964 M * Bertl nah, not that urgent, but we have some issues with the sendfile stuff _again_
1142173999 M * Bertl I fixed a _real_ bug in rc12 and reverted your changes, as they didn't make much sense to me
1142174029 Q * Viper0482 Ping timeout: 480 seconds
1142174032 M * Bertl the result is working fine for bonbons, but seems to fail for gerardi
1142174033 M * doener they were taken from patch-2.6.16-rc5 from kernel.org
1142174079 M * doener rw_verify_area now returns a postive number if the area is too big or sth. like that, so the count has to be adjusted
1142174084 M * Bertl yeah, probably something in the logic is still wrong
1142174113 M * Bertl the real fix was to change the FLOCK*READ/WRITE to READ/WRITE
1142174152 M * Bertl but I think the return value needs more thinking (or maybe our tests were just wrong)
1142174307 M * Bertl you say rc5 .. interesting ...
1142174346 M * doener i just took that patch, because the rc10 was against that version, dunno where the change was introduced actually
1142174413 M * Bertl yeah, I see what you mean ...
1142174443 M * Bertl maybe we should retest with the READ/WRITE fixed?
1142174466 M * doener you mean with both changes applied?
1142174466 M * Bertl have to reread the irc logs if we tested that ...
1142174481 M * Bertl yeah, with your change and the R/W change
1142174504 M * Bertl IIRC; we first reverted your change, which made it work for phreak
1142174529 M * Bertl hmm, no, I have to dig zthat out
1142174646 M * Bertl http://irc.13thfloor.at/LOG/2006-03/LOG_2006-03-11.txt ~ 1142099536
1142174709 M * Bertl doener: the thing is I was comparing stable against devel version, and I didn't see your checks in my stable branch (which might be the reason for 'reverting' it in the first place)
1142174986 M * doener Bertl: i guess READ != FLOCK_VERIFY_READ, right?
1142175116 M * doener cause then the other fix affects locks_mandatory_area which is the function actually returning the positive int... so it might depend on read_write whether it works, when only one of the fixes is applied
1142175275 J * coocoon ~coocoon@p54A079F5.dip.t-dialin.net
1142175285 M * coocoon hello
1142175341 M * Bertl doener: yep, READ and WRITE use 0 and 1 while the FLOCK versions use 1 and 2 :)
1142175351 M * Bertl welcome coocoon!
1142175455 M * doener so we always used the write case... ok doesn't help me... :)
1142175485 M * coocoon bertl: now I know the --KEEP command
1142175487 M * coocoon ;-)
1142175580 M * Bertl coocoon: congrats! :)
1142175703 M * coocoon now u have workes out rc12
1142175709 M * coocoon worked
1142176076 J * phycho ~phycho@ext-gw.darktech.org.uk
1142176078 M * phycho hey guys
1142176086 M * phycho anyone remember the vserver control panel? what was the name of it?
1142176162 Q * rs Quit: rs
1142176177 M * phycho We're Sorry.
1142176177 M * phycho The SourceForge.net Website is currently down for maintenance.
1142176177 M * phycho We will be back shortly
1142176178 M * phycho ffs
1142176228 M * doener Bertl: btw, do you remember that vserver control panel thing I showed you about a week ago? they added a "opensource group" now :)
1142176257 M * phycho ooh
1142176258 M * phycho link
1142176265 M * phycho im looking for one of those :p
1142176292 M * doener http://vserver.webman-tool.de/vserver/ -- current version is free, but only as in beer... and it's all german
1142176296 J * matta ~matta@c-68-32-239-173.hsd1.pa.comcast.net
1142176331 M * phycho boo
1142176341 M * phycho any screenshots?
1142176456 M * doener didn't find any
1142176478 M * phycho k
1142176580 M * phycho does vserver use folders or does it use filesystem images?
1142176582 A * phycho forgets
1142176585 M * phycho has been that long since ive used it
1142176619 M * doener the former
1142176624 M * phycho k
1142176726 M * phycho suppose i could use filesystem images if i mounted them
1142176729 M * phycho and then copied the files out of it
1142176769 M * doener if the image is writable, just mounting it should be enough i guess
1142176784 M * phycho yeh
1142176808 M * phycho i use UML for my gateway, but i dont fancy the kernel overheads etc for my other server.
1142176821 M * doener :)
1142176834 M * phycho what kinda interfaces does vserver use? just the host interface?
1142176839 M * phycho i might try setup vlans
1142176864 M * doener just the host interface. vserver network virtualization is based on ip addresses, not interfaces
1142176875 M * phycho k
1142176886 M * phycho so i cant use vlans then? or can i?
1142176920 M * phycho the idea is, i want to keep the network secure between the vservers, so they cant sniff each others traffic
1142176944 M * doener unless you grant additional capabilities, they cannot sniff at all
1142176959 M * phycho k cool
1142176964 M * phycho ty :)
1142176967 M * doener yw
1142176987 M * phycho and i take it theres no way to break out into the host system?
1142177000 M * phycho i had that problem with UML, had to use grsec + chroot and run as limited user
1142177006 M * doener no known one at least ;)
1142177009 M * phycho lol k
1142177025 M * phycho good enough for me :-)
1142179267 Q * michal` Ping timeout: 480 seconds
1142179504 Q * phycho Ping timeout: 480 seconds
1142179615 J * michal` ~michal@www.rsbac.org
1142179940 J * mmouse ~ab@office.haefft.de
1142179950 M * mmouse hi
1142180236 M * Bertl welcome mmouse!
1142180388 M * mmouse Bertl: reboot funktioniert :-) Danke nochmal
1142180398 M * mmouse aeh thanks
1142180414 M * Bertl good to hear :)
1142180416 M * coocoon Bertl: a short question if set in the DiskLimit.sh space_used=0, is it right that that the space of the vserver guest isn't used, although the guest OS (with space) is loaded
1142180453 M * Bertl no idea what the DiskLimit.sh does?
1142180520 M * coocoon Bertl: hm, it is a script from the vserver.org page  /usr/sbin/vdlimit --xid 100 --set space_total=1000000 --set space_used=0 --set inodes_total=1000000 --set inodes_used=8950 --set reserved=5 /vservers/name
1142180603 M * Bertl well, that means that the 'current' used space (block counter) is set to zero
1142180630 M * Bertl (which means that there are no resources used by that guest, except for shared files)
1142180678 M * coocoon bertl: but if there is something installed new what happens then, the space will grow or not
1142180689 M * coocoon used space i mean
1142180727 M * Bertl yep, kernel will keep the books
1142180737 M * Bertl i.e. it will add/remove space there
1142180753 M * Bertl a better approach is to set the space_used to the actual space used right now :)
1142180788 M * Bertl (similar to the inodes_used value (no idea where that value comes from)
1142181023 M * coocoon bertl: but if I wanted to give the guests 1GB space, so I can give it exactly without the used space of the OS, or can this led to probs
1142181220 M * tokkee What exactly does the VSZ value mean in the vserver-stat output?
1142181554 M * Bertl coocoon: well, the thing is, what is if the guest removes a file (with a default of 0)
1142181583 M * Bertl tokkee: it's the sum of all VSZ values reported by vps for that guest
1142181609 M * tokkee Bertl: What exactly is the VSZ value? ;-)
1142181725 M * Bertl the VSZ value is the number of virtual memory pages per process
1142181740 M * Bertl IIRC the vserver-stat presents that as megabytes or so
1142181788 M * Bertl tokkee: you know the difference between VM and RSS pages?
1142181797 M * tokkee Well... it does prepend a "M" to the actual value... which might be 10^6 pages or 10^6 bytes...
1142181803 M * tokkee Bertl: Nope...
1142181813 M * tokkee s/preprend/append/
1142181834 M * Bertl okay, it is quite simple actually, but complicates the often simplified user view
1142181854 M * Bertl from the physical point of view, you have RAM (memory)
1142181875 M * Bertl from the kernel/software point of view you have VM and RSS pages
1142181892 M * Bertl each process has basically an idependant VM space
1142181914 M * Bertl this space is limited by the architecture, on x86 with 2/2 split it's 2GB for example
1142181941 M * Bertl even if you have only 512MB RAM, this space will still be 2GB :)
1142181962 M * tokkee Yes... with you so far ;-)
1142181982 M * Bertl each 'page' of VM (typically 4k on x86) will require an RSS page to be accessed
1142182001 M * Bertl the VM pages can have different qualities (sources)
1142182021 M * Bertl they can be file based (e.g. executables, binaries, libraries, mmaped)
1142182045 M * Bertl they can also be anonymous (e.g. conventional data memory)
1142182056 M * Bertl they can also be shared between processes
1142182103 M * Bertl every time the data the VM page is referring to is read or written, the page has to be mapped into actual memory 
1142182107 M * tokkee Yes... I'm pretty sure I already know what VM is ... ;-)
1142182125 M * Bertl at this point the VM page gets an RSS page
1142182155 M * Bertl RSS pages are pages which actually consume RAM
1142182184 M * Bertl now while the VM page count for a single process makes perfect sense
1142182211 M * Bertl the VM 'usage' over many processes is not really easy to define
1142182251 M * Bertl is it the number of mappings? or just a sum over all pages?
1142182251 M * Bertl what to do with shared pages?
1142182279 M * Bertl so we decided to go for the least intrusive and most simple way to handle that, just sum up all vm spaces
1142182478 M * tokkee Bertl: Thx...
1142182523 M * tokkee Now, is there a way to get the amount of real (physical) RAM usage of each vserver, similar to what free(1) says?
1142182548 M * Bertl tokkee: the RSS is the real physical RAM
1142182561 M * Bertl you can read that from the /proc/virtual/<xid>/limits
1142182617 M * Bertl the vserver-stat IIRC sums up the RSS values for the individual processes, which is not exactly the same, but usually in the same ballpark
1142182626 M * tokkee Without the buffers and cached mem areas then?
1142182666 M * Bertl yep, buffers and cached mem are not part of the process RSS values
1142182680 M * tokkee Alright... think I got it... thx again :-)
1142182711 M * Bertl you're welcome!
1142182749 M * tokkee 14:00 < pzYsTorM> and the RSS value is the actual occupied memory?
1142182749 M * tokkee 14:00 < pzYsTorM> +ly
1142182749 M * tokkee 14:01 < Bertl> yep, that's the RAM used (resident set size)
1142182752 M * tokkee *lol*
1142182809 M * tokkee You could probably fill tons of real FAQs out of this channel ;-)
1142182857 Q * RoadRunnR Quit: Leaving
1142182924 M * tokkee Bertl: You're german speaking I suppose? How would you translate "resident set size" to german?
1142183084 J * restill ~restill@c-24-11-171-10.hsd1.mi.comcast.net
1142183160 M * tokkee Is there any documentation on the /proc/virtual filesystem?
1142183221 M * Bertl not at all, probably, but if you have to then maybe: "aktiver speicher"
1142183241 J * lilalinux ~plasma@dslb-084-058-193-186.pools.arcor-ip.net
1142183248 M * tokkee ;-) thx
1142183259 M * Bertl there is but it is in an early stage http://linux-vserver.org/HowTo+Read+ProcFS
1142183376 M * tokkee Looks like a good starting point ;-)
1142183624 M * waldi bah, dietlibc have a broken mmap2 implementation
1142183634 M * waldi now its a real error
1142183887 M * Bertl waldi: hmm, and what about the clone() issue?
1142183897 M * waldi Bertl: this is a util-vserver bug
1142183904 M * waldi i have a fix
1142183913 M * Bertl okay, good :)
1142183916 M * waldi mmap2(0x7fd583b8, 256, PROT_READ, MAP_PRIVATE, 3, 0x7f8724a8) = 0x77fff000
1142183916 M * waldi mmap2(0x7fd583b8, 256, PROT_READ, MAP_PRIVATE, 13, 0x7f8724a8) = -1 EBADF (Bad file descriptor)
1142183934 M * waldi the sources specifies both the first and the last parameter as 0
1142183942 M * waldi the others are okay
1142184062 M * waldi the assembler code looks rather similar to the version in glibc, that is what me drives crazy
1142184167 M * tokkee What is the "load_updates" value in /proc/virtual/<xid>/cvirt?
1142184197 M * tokkee And what's the difference between files and file-descriptors in /proc/virtual/<xid>/limit?
1142184260 M * waldi even the comments are the same
1142184331 M * Bertl waldi: don't trust the glibc version :)
1142184346 Q * matta Ping timeout: 480 seconds
1142184370 M * Bertl tokkee: it's how often the virtual load accounting did update
1142184370 M * waldi Bertl: at least it works
1142184381 M * Bertl waldi: no, it work with glibc!
1142184399 M * tokkee Bertl: So, I guess, it's for debugging purposes mainly!?
1142184412 M * Bertl tokkee: file descriptors are slightly different from file structures
1142184420 M * Bertl tokkee: yep
1142184423 M * waldi Bertl: the abi is the same
1142184449 M * tokkee Bertl: Ah... okay... so files means kernel-file structures?
1142184472 M * Bertl waldi: for the ABI double check with my syscall version (which is tested on s390)
1142184500 M * waldi Bertl: mmap2 get one parameter, this is a pointer
1142184590 M * coocoon bertl:  I always got this during running guests http://pastebin.com/598117 it doesn't depend on any xid 
1142184641 M * Bertl waldi: so that can not be a syscall issue then, just a wrong structure, no?
1142184664 M * waldi Bertl: i just try to understand, where the broken value comes from
1142184698 M * Bertl do you have the assembler code at hand?
1142184721 M * waldi yes
1142185128 M * Bertl good for you then :)
1142185219 M * waldi bah, gdb say, the struct is correct
1142185276 M * waldi yeah, the bug don't shows up in gdb
1142185462 M * Bertl doesn't mean much, unless you attach to the kernel and check there
1142185491 M * coocoon Bertl: tesme.sh and vserver-info (tesme.sh shows chbind: vc_net_create(): Invalid argument chbind failed!) http://pastebin.com/598152
1142185574 M * Bertl coocoon: could you redo the testme.sh with the latest version of testme.sh (0.15) please?
1142185601 M * coocoon ok
1142185612 M * waldi and the util-vserver tool uses another function calls __libc_mmap64
1142185674 Q * hw Quit: Client exiting
1142185681 M * Bertl coocoon: but it is very likely that you want to add --enable-apis=NOLEGACY to your ./configure of the tools
1142185744 M * coocoon Bert: hm I have disabled in the kernel
1142185814 M * coocoon bertl: is the latest testme.sh link at the step-by-step guide
1142185890 M * coocoon ok I will try it with this setting
1142185895 M * Bertl then the step-by-step guide links to the wrong file, checking
1142185909 M * coocoon ok
1142185919 M * coocoon i have started the ./configure
1142186001 M * Bertl nope, links are fine, and I get the 0.15 version
1142186010 M * coocoon ok now it works with  wrong file
1142186076 M * coocoon http://pastebin.com/598171, but what is with these errors http://pastebin.com/598117
1142186161 M * coocoon now I got these messages is this ok? http://pastebin.com/598178
1142186212 M * Bertl the testme.sh looks good
1142186236 M * coocoon tehes messages now away vxW: xid=100 messing with the procfs
1142186241 M * coocoon i hope so
1142186261 M * Bertl well, could be, it depends on the guests
1142186278 M * waldi Bertl: the reason is quite easy. dietlibc uses mmap2 directly in its mmap64 implementation, it assumes that the calling conventions for the syscall and function calls matches
1142186283 M * coocoon but now i got these dynamic contexts disabled. but it looks like, because of the new ./configure setting
1142186286 M * Bertl this is basically a warning that a program inside the guest tried to modify entries in procfs
1142186308 M * Bertl coocoon: I have no idea what the messages mean ...
1142186310 M * tokkee What is anonymous memory btw.?
1142186347 M * Bertl waldi: yep that is a bug I reported a few month ago to fefe for some other syscall/arch
1142186375 M * Bertl tokkee: memory not backed by any file
1142186409 M * coocoon bertl: no it can't be depend on the guest
1142186435 Q * pzYsTorM Ping timeout: 480 seconds
1142186435 M * Bertl coocoon: that was for the xid=100 warning
1142186448 M * coocoon bertl: ah ok
1142186457 M * coocoon bertl: now they are away ;-)
1142186499 M * tokkee Bertl: d'oh... you already told be a couple of minutes ago... *selfslap* ;-)
1142186545 M * tokkee What unit are the values in /proc/virtual/<xid>/limit? Bytes, KB?
1142186551 M * Bertl pages
1142186566 M * tokkee okay... thx
1142186569 M * Bertl btw, this is explained by the limits
1142186586 M * tokkee ?
1142186605 M * coocoon bertl: a question the tagging of the guest disklimit --> " /usr/sbin/vdlimit --xid 100 --set space_total=1000000 --set space_used=0 --set inodes_total=1000000 --set inodes_used=8950 --set reserved=5 /vservers/name" I have put in /etc/vservers/name/scripts/post-start.d, is this right or should it be in /etc/.../pre-start.d
1142186645 M * Bertl tokkee: http://linux-vserver.org/Resource+Limits
1142186674 M * Bertl coocoon: if it works, it is fine ...
1142186702 M * coocoon bertl: it works but I think that these messages can come from this tagxid
1142186706 M * Bertl coocoon: I cannot remember which script it called in what context, and there is no documentation IIRC
1142186746 M * Bertl the vdlimit has to be executed in the host context, if that is true for the script, it's perfectly fine
1142186769 J * jkl eric@c-67-172-156-116.hsd1.co.comcast.net
1142186811 M * Bertl welcome jkl!
1142186837 M * coocoon ah ok thanx bertl good night
1142186854 M * Bertl have a good one :)
1142186869 M * coocoon but one thing I got this from here http://linux-vserver.org/Disk+Limits, so there is a documentation right
1142186881 M * coocoon or have I misunderstood anything
1142186897 M * Bertl nope, if it is written there, probably somebody tested it :)
1142186929 M * coocoon oh I see there has been something changed, ok I will have a look
1142187338 M * waldi Bertl: "fd = open (FILENAME, O_RDWR | O_CREAT);" what is wrong?
1142187441 M * Bertl looks fine, although RDWR and CREAT are unusual
1142187464 M * Bertl assumed that FILENAME is a string constant
1142187475 M * waldi no, the third argument is missing
1142187482 M * Bertl depends
1142187507 M * Bertl from man2
1142187511 M * Bertl        int open(const char *pathname, int flags);
1142187511 M * Bertl        int open(const char *pathname, int flags, mode_t mode);
1142187540 M * waldi with O_CREAT it is necesary
1142187569 M * waldi (if you don't want to end with permissions like 135)
1142187589 M * Bertl well, what if 0135 is fine :)
1142187626 M * waldi *hau*
1142187681 M * Bertl but hey, that's userspace, what am I supposed to know about that :)
1142188175 J * Smutje_ ~Smutje@xdsl-87-78-84-30.netcologne.de
1142188339 Q * Smutje Ping timeout: 480 seconds
1142188339 N * Smutje_ Smutje
1142189958 M * mmouse bertl: are there any known bugs in 2.0.2-rc10 which should prevent me from giving it a try in production environment?
1142190041 M * Bertl well, I'd take rc12, but for that, no, otherweise we would fix them :)
1142190080 M * mmouse Bertl: what about this sendfile issue in rc11/12, or is this 2.1.1 only?
1142190167 M * Bertl this is only because 2.1.1 changes sendfile to make the CoW stuff work
1142190199 M * Bertl (and either some tests were done wrong or it is a really hidden issue)
1142190213 M * Bertl mmouse: but no, it doesn't affect 2.0.2
1142190454 M * restill I just finished compiling kernel-2.6.15-1.1833_FC4.vs2.0.2.0.rc11.2.x86_64.rpm if anyone is interested.
1142190473 M * restill and kernel-smp-2.6.15-1.1833_FC4.vs2.0.2.0.rc11.2.x86_64.rpm
1142190650 M * Bertl rc11.2 is what precisely?
1142190736 M * Bertl okay, off for a little, back later .. have fun!
1142190739 N * Bertl Bertl_oO
1142190781 M * restill Its from daniel's SRPMS http://rpm.hozac.com/dhozac/fedora/4/vserver/SRPMS/
1142191448 M * restill going to eat.  later
1142191451 Q * restill Quit: Leaving
1142191592 M * daniel_hozac Bertl_oO: rc12 before rc12 appeared :) (i.e. rc11+namespace fix)
1142192086 J * doener_ ~doener@i5387DB4A.versanet.de
1142192500 Q * doener Ping timeout: 480 seconds
1142193350 Q * coocoon Quit: KVIrc 3.2.0 'Realia'
1142193928 P * mmouse 
1142194587 Q * liquid3649_ Quit: bin raus, 
1142194948 M * tokkee In /proc/virtual/<xid>/cacct, there are three columns: one for incoming and one for outgoing traffic. What about the third one?
1142195083 M * daniel_hozac failures, IIRC.
1142195134 M * tokkee And incoming is the first, outgoing the second? That's in fact all I need to know ;-)
1142195465 M * daniel_hozac "Type\t    recv #/bytes\t\t   send #/bytes\t\t    fail #/bytes\n"
1142195721 M * tokkee Thx.
1142196109 Q * comfrey Ping timeout: 480 seconds
1142196944 J * matta ~matta@c-68-81-35-243.hsd1.pa.comcast.net
1142197432 J * phycho ~phycho@ext-gw.darktech.org.uk
1142197439 M * phycho hey guys
1142197443 M * phycho what version of vserver should i go for
1142197448 M * phycho just used harry's patches on 2.6
1142197458 M * phycho root@vps:/home/admin# uname -a
1142197458 M * phycho Linux vps 2.6.14.7-vs2.1.0-grsec-2.1.9 #1 PREEMPT Sun Mar 12 20:04:34 GMT 2006 i686 unknown unknown GNU/Linux
1142197458 M * phycho root@vps:/home/admin#
1142197464 M * phycho i want the latest ;P
1142197509 M * phycho where do i get util-vserver-0.30.210
1142197511 M * phycho i only see 209 on site
1142197557 M * daniel_hozac http://www.13thfloor.at/~ensc/util-vserver/files/alpha/
1142197660 M * phycho aah
1142197669 M * phycho thanks
1142197693 M * phycho take it i need 'vdlimit' as well
1142197696 M * phycho to limit disk usage?
1142197718 M * daniel_hozac vdlimit is part of util-vserver since 0.30.205 or something like that.
1142197730 M * daniel_hozac if not earlier.
1142197808 M * phycho k
1142197809 M * phycho cool
1142197810 M * phycho =)
1142197811 M * phycho ty
1142198029 Q * matta Ping timeout: 480 seconds
1142198066 M * daniel_hozac Hollow, phreak``: i updated the bmask patch, nothing significant, but it's more of a correct solution.
1142198648 M * Hollow ic, will put in the next revision
1142199052 M * phycho where can i get dietlibc?
1142199084 M * daniel_hozac http://www.fefe.de/dietlibc/
1142199154 M * phycho thanks
1142199637 M * phycho fuck it
1142199641 M * phycho ill use the normal libc
1142199643 M * phycho dietlibc sux
1142199648 M * phycho it errors on install/compile
1142199763 M * daniel_hozac glibc will almost certainly not work satisfactory.
1142199783 M * daniel_hozac there have been all sorts of weirdness when using glibc.
1142199791 M * daniel_hozac (but then again, it works fine for some people)
1142199853 M * daniel_hozac you may also want to get a decent distribution that has dietlibc :)
1142199892 M * phycho heh
1142199895 M * phycho slackware is decent..
1142200015 M * phycho :P
1142200032 Q * michal` Ping timeout: 480 seconds
1142200239 J * Aiken ~james@tooax6-213.dialup.optusnet.com.au
1142200240 M * mugwump slackware decent?  nay, it's nude
1142200262 Q * bonbons Quit: Leaving
1142200528 M * phycho i quite like it
1142200657 Q * marl Ping timeout: 480 seconds
1142201004 M * mugwump sure, but if it were decent it would be fully clothed for a start.  slackware might be fun at a pool party, but you wouldn't invite him to your wedding
1142201022 A * mugwump loves anthropomorphising classing flamewars
1142201028 M * mugwump *classic*
1142201048 M * phycho lol
1142201068 M * phycho nothing wrong with it
1142201074 M * phycho give me it over debian with broken crappy packages anyday
1142201316 M * mugwump sure, debian always shows up in those 70s suits and thinks he's hot shit, but some of the girls find his lewd advances irritating.  Ubuntu just goes on to make good friendships with them with his funny mix of cool charm and pleasant manner
1142201358 M * mugwump I mean, Debian was funny, they even based that Austin Powers movie on him, but seriously, retro is SO 90s
1142201363 J * marl ~matt@84.92.193.226
1142201387 M * daniel_hozac lol
1142202444 Q * marl Read error: Connection reset by peer
1142203115 Q * doener_ Quit: leaving
1142204333 M * mugwump Bertl_oO: damn, I can't get to Essen in April after all.  My sister is visiting from the UK then.
1142204693 M * phycho heh
1142204844 M * tokkee Seems like I was programming to much Perl lately ;-) How to I initialize an array in C?
1142204859 M * tokkee s/to I/do I/
1142204864 M * daniel_hozac = { 1, 2, 3, 4, 5 };
1142204870 M * tokkee Ah... thx.
1142204884 M * tokkee I was trying = (1, 2, 3); ;-)
1142205480 M * phycho anyone know where ill find a guide for vserver?
1142205506 M * daniel_hozac guide for what?
1142205515 M * daniel_hozac there are quite a few for various topics on linux-vserver.org
1142205524 M * phycho to build the vps's
1142205529 M * phycho k ill go have a look
1142205530 M * daniel_hozac vserver - build --help
1142205533 M * phycho yeh i did that
1142205534 M * phycho looks confusing
1142205547 M * phycho was straightforward on openvz (which is what i used previously)
1142205553 M * daniel_hozac if you say so.
1142205603 M * phycho :P
1142205633 M * mugwump openvz is a bit of a sore spot, not least because of the attitude of the developers.
1142205706 M * mugwump eg last Friday "After checking proposed yours, Eric and vserver solutions, I must say 
1142205709 M * mugwump that these all are hacks.
1142205726 M * mugwump Smacks of "I am better than all of you"
1142205729 M * phycho lol
1142205743 M * phycho the reason i switched was because openvz + grsec was a pain in the rear :-)
1142205750 M * phycho and i wanted to try something different
1142205863 M * mugwump In case you missed that, dev_, I think your LKML manner is rude.
1142206002 M * phycho haha
1142206022 A * phycho quite likes openvz
1142206229 M * phycho might try moving my configs from another system to a vserver on this one ;-)
1142206267 M * phycho btw
1142206271 M * phycho do you still need the chroot barrier?
1142206275 M * phycho on /vservers?
1142206422 M * daniel_hozac i think no, but what does it hurt?
1142206448 M * phycho was just wondering :-) ill enable it anyway
1142206476 M * phycho reading the blurb it sounds sweet :D (vserver that is)
1142206984 P * meandtheshell 
1142207427 J * michal` ~michal@www.rsbac.org
1142207668 M * phycho im confused 
1142207718 M * tokkee Does anybody in here know collectd (http://verplant.org/collectd/)?
1142207762 M * phycho bugger it.. ill take the private.tar.gz (my vps's from openvz) and migrate them over
1142207799 M * phycho hadnt seen it b4
1142207936 M * tokkee Anyway, I've just written a plugin for it that collectd socket, threads, load, process and memory usage data of each vserver. I guess, it will be included in the next release. Some beta-testers would be appriciated ;-) The plugin is already available from the collects-subversion repository