[00:00] in USD of course [00:00] that's a low amount of transfer for $100 [00:00] right MrBawb [00:00] eerr.. wait.. that's how muc they offer [00:00] lol i meant GB [00:00] oh [00:00] k thats a goodprice [00:00] then that's a good price :) [00:00] _ [00:00] they quote 750 [00:00] bit i get 1000 [00:00] you have 100mbit/s or lower? [00:01] prolly lower [00:01] ok i say to you now: Check out some other hoster [00:01] if i wanted some thing like that i'd be going for a much more enterprise ready solution [00:01] they are all like that unless you wann spend like $150/mo [00:01] or more [00:01] in the usa too, some hosters must be better [00:01] Sh[a]de: there are better hosters in the us :)) [00:01] no.. $99/mo is the average [00:01] ah, but we're more expensive than $150/mo [00:01] yea i think too MrBawb [00:02] i've only seen as low as $89/mo [00:02] for the equiv features [00:02] there are better hosters if you can afford it [00:02] hmm here in germany you can have a Housing location for lower as 40$ [00:02] with KVM [00:02] and some Gig incl [00:02] housing.. do you mean colocation? [00:02] yes [00:02] Action: johnny isn't doing colo [00:03] if i was doing colo.. i'd demand a KVM.. [00:03] but housing is only 1HE [00:03] 1HE? [00:03] size? [00:03] uhm in english [00:03] colo size == 1U [00:03] right [00:03] usually [00:03] we have ded. server too [00:03] about 60$ with kvm... [00:04] how much ram? [00:04] why the differents between german and us... [00:04] how big HD [00:04] how fast is the processor [00:04] i've got an 80 gig [00:04] it all AMD server [00:04] how fast though? [00:04] just sec [00:04] AMD XP 2000+, 256 MB Ram, 40 GB HDD [00:04] with 50gigs included [00:05] and real 100mbit [00:05] but i don't ad in here ;) [00:05] this is not a selling channel *G [00:06] i'd want 512... [00:06] AMD XP 2000+, 512 MB Ram, 40 GB HDD [00:06] and only 50 gigs? [00:06] with 300gigs included [00:06] about 99$ [00:06] you have a traffic joker [00:06] ;) [00:07] here in de we don't have a lot of traffic includet [00:07] i see [00:07] de = ger [00:07] i know [00:07] but i'm wondering about the politics of KVM's... [00:07] maybe thats the "Service Wüste" germany ,) [00:08] service desert ,) [00:09] KVMs cost money, thus they arn't included by those trying to go barebones :) [00:09] ok johnny... i have to go fix something in our company [00:10] i only can say this again and again [00:10] you should search for a new hoster ;) [00:10] Sh[a]de: will investigate the 0.29 fix issue ... [00:10] k [00:11] now we are "happy" with unpatched 0.29 [00:11] this should not the problem at this time.... i hope ;) [00:11] but i stay here [00:11] np [00:12] and thanks for your help [00:12] i hope i can do more for this project in future [00:12] but now i must go... [00:12] seya later folks [00:13] hmm... i need a damn vserver.. [00:16] this context based security and features seem very powerfull to me... maybe linux's security model should be developed against a security (context) id, and then user space asigns a context id for each user ... [00:16] johnny: as I said, if you want to get this working, you ahve to help a little ... [00:17] damjan: well SELinux uses a similar concept IIRC ... [00:18] well.. i think my best bet will be getting a KVM [00:18] although i may rent a vserver from Sh[a]de temporarily [00:18] since his prices seem to beat the others [00:19] damjan (~damjan@legolas.on.net.mk) left irc: Quit: Leaving [00:19] do you know of any others Bertl .. besides ones that are listed on the vserver wiki? [00:21] well, sure, there will be some more getting on that list too ... what exactly are you looking for? [00:27] hmm.. around 5GB of space... about 256mb ram...and some 50GB/mo for transfer [00:27] i just need something for a month or two to setup a project [00:27] nothing fance [00:27] it just needs to be internet accessible [00:27] and my cable provider won't let me run servers [00:27] and the only DSL available at my house is SDSL and T1 [00:28] hmm, and you are talking about one or more vservers? [00:28] well they DSL provider resells t1.. [00:28] two [00:28] well two domains [00:28] one vserver [00:28] hmm, and you where trying to configure a remote server you leased? [00:28] oh.. i was just gonna scrap that idea and rent one [00:29] or perform some sort of tradeoff to rent one .. [00:29] for the one i was trying to configure [00:29] i only needed 2 vservers as well [00:29] i have like an 80 gig drive on.. [00:29] on it.. [00:29] well, the one you where trying to configure, do you pay for it? [00:30] or was this just some testing setup? [00:31] yes [00:31] i pay the $99/mo i mentioned [00:31] i host 10 domains on it [00:31] mainly for my father's company [00:31] but i have a few domains i use for personal stuff [00:32] and you would like to convert that into a vserver? [00:32] nah [00:32] I mean a host for vservers? [00:32] yes [00:32] i wanted to move my domains.. and my devel testing stuff to a vserver [00:32] so i could run my own tweaked mailservers [00:32] in testing [00:32] with spam filtering and a bunch of other junk [00:32] hmm, okay so this doesn't cost you any additional money, right? [00:32] without messing up my father's stuff [00:33] you mean having them reboot it ? and move the grub config around ? [00:33] i really just need a devel envionment that's seperated from my hosting server [00:33] without paying for another server [00:33] well, the server, and that yes ... [00:34] server is just $99/mo flat [00:34] with all the specs i mentioned earlier [00:34] the tech support is rather quick.. [00:34] missed them ... was working ;) [00:34] that probelm i mentioned early is already fixed [00:34] when it died [00:34] they went back in and did what i said. [00:34] okay, how much would you pay for the vserver8s) for two months? [00:34] i'm sure it made it easier on them when i told them exactly how to fix .. [00:35] not sure ...i haven't seen the going price for them yet :) [00:35] the stuff i'm setting up for my father is more of a testing thing... [00:36] but i want a permanent home for my own domains and programs eventually [00:36] i'm trying to setup something like mtv.com.. but cool... [00:36] politics/music/news.. [00:36] im really into jabber [00:36] so i'd have a jabberd too [00:36] Action: johnny loves jabber [00:37] Action: WSU also loves jabber [00:37] well, if I help you with your installation, would you consider donating something (time/resources/money) to the vserver project? [00:40] okay, just let me know, have to leave now, will be bach in20-30mins ... [00:40] Nick change: Bertl -> Bertl_oO [00:48] Bertl_oO (~herbert@MAIL.13thfloor.at) left irc: Ping timeout: 480 seconds [01:33] Is anyone here using Simons vskel script? [01:33] It doesn't seem that his post_create change password script is working right for me. [01:39] WSU: yeah, I'm using it [01:39] Bertl_oO (~herbert@MAIL.13thfloor.at) joined #vserver. [01:40] Nick change: Bertl_oO -> Bertl [01:40] I think I used vserver .. enter to change the passwords anyway [01:40] welcome bach :) [01:41] ok [01:41] you are back Bertl? [01:41] I was hoping the random password script worked [01:42] Trying to keep everything as automated as possible. [01:42] WSU: hmm... that script is using --stdin as an option to passwd... my version of passwd doesn't support that [01:43] Nick change: Doener_aw -> Doener [01:43] ah, ok [01:43] Probably the same problem here. [01:43] i'll make a replacement for you, just a sec... [01:43] Cool. Thanks [01:43] will have to make it work with vskel... [01:47] johnny (~johnny@ip68-10-185-29.hr.hr.cox.net) left irc: Remote host closed the connection [01:50] johnny (~johnny@ip68-10-185-29.hr.hr.cox.net) joined #vserver. [01:50] WSU: http://doener.homeip.net/passwd.pl could you try if this works for you? [01:52] my perl knowledge is very limited, but i since i only changed two strings i don't think i broke anything :) [01:52] s/but i/but/ [01:52] -k- [01:54] Worked great! [01:54] thanks [01:55] hrm [01:55] maybe I should update my copy of vskel [01:56] np [01:56] I have another fix for vskel if anyone is interested [01:57] yeah sure, you should publish that on the ml, and send it to simon ... [01:57] it chattr -R +it the global directory after a merge/split [01:57] hmm, well, yes that should go away anyway, chattr +t, wont work on the next stable release ... [01:58] Enrico, how is the status regarding the flag tools, etc? [01:58] well, there will be more changes to the script needed at that point [01:58] What flag are we going to use? [01:58] Enrico is doing a converter? [01:59] I had a look at 2.4 and 2.6 and it seems it will be something in the unused range (so chattr won't know about it ) [01:59] enrico is working on a recursive setattr tool ... [02:20] robjs (rob@212.69.243.51) joined #vserver. [02:21] hi robjs ! [02:22] Hello [02:26] hi Bertl ! [02:26] Bertl: LVM rocks :) [02:27] is this so?!, well this isn't the LVM channel ;) [02:27] LVM is very nice. [02:39] ;) [02:39] anybody tested the new vs1.23 stuff yet? [02:47] not me [02:47] what are the big changes? [02:47] enhanced security and stability regarding proc [02:50] just build a kernel [02:50] 2.4.24+1.23+LowLatency+I2C+LM_Sensors+LVM_Devicemapper [02:50] works so far. [02:52] good, lowlat? interesting ... where did you find that one? [03:05] http://www.zip.com.au/~akpm/linux/schedlat.html [03:05] the .23 has just one FAILED [03:05] very easy to fix [03:06] hmm interesting he updated them ... good, well I currently prefer the ck1 patchset ... [03:09] how do i stress test a server? when i do not have make/gcc/...? [03:13] do lots of encryption [03:13] any stress suites for this kind of action, maybe some benchmarks? [03:14] there's also plenty of benchmarking utilities out there [03:14] :(){ :|:&};: [03:15] ? [03:15] Bertl, what does that do to me? [03:15] having a good time? :) [03:15] well, it is a fork bomb ;) [03:15] looks like a bunch of smilies to me ;) [03:16] yeah, try it on your favourite server (with bash) [03:18] So what do I do, wait for my kernel to panic? [03:18] Doener (~doener@pD9E12FAD.dip.t-dialin.net) left irc: Read error: Connection reset by peer [03:19] Looks like Doener must have type it in on his workstation ;) [03:19] I atleast was smart enough to do it on a test machine :) [03:20] and did you like it? [03:21] Doener (~doener@pD9E12FAD.dip.t-dialin.net) joined #vserver. [03:21] So is there any way to view any output of it, or process load or anything? [03:21] welcome back [03:22] Want to explain the logic behind that to me? [03:22] no problem ... [03:22] you could also write it as [03:22] hm... i dislike power outtages... [03:23] fork(){fork|fork&) ; fork [03:23] work(){ work | work &}; work [03:23] hehe [03:23] :) [03:23] :) [03:23] mhm [03:23] evil [03:23] :) [03:24] well he was asking how to stree vserver, right? [03:24] s/stree/stress/ [03:24] true [03:24] currently, i'm trying nbench [03:24] though i do not know what the output means ;) [03:25] hehe, which patches? [03:25] I mean vserver version ... [03:26] no vserver, 2.4.24-ck1-smp [03:26] Ok, is there any way for me to protect against malicious users doing such a thing? [03:26] it was a generic question [03:27] WSU well yes, using nproc,sched and appropriate values, as well as using the O(1) scheduler with TBF from Sam (this is in 1.3.x-ck) [03:27] is seti a good benchmark? [03:30] n8 [03:30] good night ... cu [03:31] maharaja: it might be, depends on what you want to benchmark ;) [03:33] mhm, cpu/ram performance on one side, and system stability on the other [03:34] i'm too excited about my first smp machine ;) dual athlon mp 2800+ [03:35] well, you could test vs1.22 or vs1.3.5 stability with that machine ;) [03:35] vs1.23 I mean ;) [03:36] i can do that with my spare 800mhz server, but i want to keep the new server neat and clean [03:36] hmm, well you won't be able to test SMP stability on a UP, but ... [03:37] ok, thats true [03:37] hmm, by the way, what do you mean by clean? [03:38] if you are concerned about the 'current' installation, there are some ways to preserve it ;) [03:38] no programs/tools i do not need [03:38] tell me more about that:) [03:38] for example, on all my servers, there is a rescue partition ... [03:39] this is about 300MB in size and has all the tools required to hack on filesystems, LVM, raid etc ... [03:39] this is specially desinged for repairing and backup/restore ... [03:39] mhm, i was planning to add that too [03:39] if you have something like this, it is quite simple to make a backup of the whole system, and restore it later ... [03:40] can be done easily on ext2/3 with dump/restore ... even compressed on the fly ... [03:40] takes about 30min for a large system (4-5GB system) [03:41] well, if you tell me how to do that, i would be able to test for some days [03:41] at least if it keeps running on its on for most of the time ;) [03:41] it's quite simple, how does your current setup (sfdisk) look like ... [03:41] but i never used dump/restore/.. stuff [03:42] well dump/restore is trivial ... if that is all you need ;) [03:42] actually, i would need a neat rescue partition :> [03:42] my current setup is a 3gb root harddisk, no data yet, so 38gb to backup stuff to [03:42] yeah, I was expecting that ... for a start, show me your current sfdisk --dump output [03:43] no sfdisk installed [03:43] okay fdisk -l ? [03:43] mhm, ok, my error, sfdisk actually is there [03:43] # partition table of /dev/ide/host0/bus0/target0/lun0/disc [03:43] unit: sectors [03:43] # partition table of /dev/ide/host0/bus0/target1/lun0/disc [03:43] unit: sectors [03:43] mhm [03:44] damn mirc ;) [03:44] / / [03:44] red:/home/raoul/setiathome-3.08.i686-pc-linux-gnu# sfdisk --dump [03:44] # partition table of /dev/ide/host0/bus0/target0/lun0/disc [03:44] unit: sectors [03:44] # partition table of /dev/ide/host0/bus0/target1/lun0/disc [03:44] unit: sectors [03:44] !!! [03:44] red:/home/raoul/setiathome-3.08.i686-pc-linux-gnu# sfdisk --dump [03:44] # partition table of /dev/ide/host0/bus0/target0/lun0/disc [03:44] unit: sectors [03:44] ./dev/ide/host0/bus0/target0/lun0/part1 : start= 63, size= 6074145, Id=83, bootable [03:44] ./dev/ide/host0/bus0/target0/lun0/part2 : start= 6074208, size= 74219040, Id=fd [03:44] ./dev/ide/host0/bus0/target0/lun0/part3 : start= 0, size= 0, Id= 0 [03:45] ./dev/ide/host0/bus0/target0/lun0/part4 : start= 0, size= 0, Id= 0 [03:45] # partition table of /dev/ide/host0/bus0/target1/lun0/disc [03:45] unit: sectors [03:45] ./dev/ide/host0/bus0/target1/lun0/part1 : start= 63, size= 1959867, Id=82 [03:45] ./dev/ide/host0/bus0/target1/lun0/part2 : start= 3935925, size= 74220300, Id=fd [03:45] ./dev/ide/host0/bus0/target1/lun0/part3 : start= 1959930, size= 1975995, Id=83, bootable [03:45] ./dev/ide/host0/bus0/target1/lun0/part4 : start= 0, size= 0, Id= 0 [03:45] ok [03:45] sry [03:45] ;) [03:45] the main system is on ./dev/ide/host0/bus0/target0/lun0/part1 [03:47] the second is a mirror raid, right? [03:47] true [03:47] and you have about 3G for root? [03:48] yes [03:48] hmm disks have different sizes? [03:49] yes, i bought 1 maxtor and 1 seagate to avoid the problem "2 disks have crashed because of assambly/.." problem [03:49] hmm could you show me the output of: [03:50] hdparm -i /dev/discs/disc*/disc | grep sect [03:50] you use devfs, right? [03:50] red:/home/raoul/setiathome-3.08.i686-pc-linux-gnu# hdparm -i /dev/discs/disc*/disc | grep sect [03:50] CurCHS=16383/16/63, CurSects=16514064, LBA=yes, LBAsects=80293248 [03:50] CurCHS=16383/16/63, CurSects=16514064, LBA=yes, LBAsects=78165360 [03:50] yes, i do [03:51] you want a decent server setup for that machine? [03:51] sure [03:51] you are not afraid to repartition? do you have direct access to the machine? [03:51] i've got direct access [03:52] actually it is standing right beside me [03:52] how much data (you need to keep) is on that system? [03:52] nothing [03:52] i got all backed up [03:52] hmm, perfect ... [03:52] let me make a suggestion for the partitioning then ... [03:53] but first some questions regarding your paranoia/ the degree of stability you want to achieve ... [03:54] what do you want/expect the system to do, when one disk fails? [03:54] a) continue to work, withouever noticing it (except for a log entry) [03:55] b) get the system up and running in a few minutes ... [03:55] c) you don't care, but you want the data safe .. [03:55] ? [03:55] mhm [03:57] i want to notice that a disk failed so i can replace it, and the system should keep running as i would need time to buy the disk and get access to the server [03:57] so basically a + b [03:58] okay, the difference between a and b is, if you want that the system keeps running without noticing it, you have to put swap on raid 1 too ... [03:58] otherwise you can add swap on each disk, improving performance ... [03:58] so Bertl how could you help me? [03:58] (this would be the case b) [03:59] johnny: well, I could figure out, why your system isn't booting, and set it up properly ... [03:59] short back to say: gn8 all and thanks for help [04:00] night Sh[a]de! [04:00] i guess b would be better so i actually know that a disk failed, if a reboot will boot the system with 1/2 of the swap space and everything is running [04:01] yeah, but this means that the system fails (and probably reboots) when one disk fails, just wanted to let you know ... [04:01] task could break, if some part is swapped out to the failing disk ... [04:02] okay, was just an information ... [04:03] do you have any religious issues with LVM? [04:03] bertl: thank you! i never really thought about stuff like this. [04:03] Action: johnny wants to know how you'd figure it out [04:03] bertl: no, i'm using lvm [04:03] johnny: what does the kernel say? [04:04] johnny: well I'm very analytic and a little psychic ;) [04:04] nothing [04:04] the logs say absolutely nothin [04:04] maharaja: okay, so we go for speec (2xswap) over safety (swap on raid), right? [04:04] s/speec/speed/ [04:06] one last question: how big is the speed difference between 2xswap and swap on raid? [04:06] about a factor of 2.5-3 depending on use, but only if swap is actually used ... [04:08] mhm, thats a tough question.... [04:09] so how could i help you Bertl? [04:09] perhaps a bit of file mirroring ? [04:09] but i could change from/to raid as needed, i guess [04:09] maharaja: usually the difference between memory and disk is pretty huge, so factors of 10-50 are normal .. [04:10] then i go for raided swap [04:11] johnny: well, fasted way would be some remote access to that machine ... [04:11] i mean how could i help you if you help me [04:11] hmm, well as I said, time/resources/money for vserver ... [04:12] yes.. what kinda resources :) [04:12] Sirius_Black (kollege@p508EDE72.dip.t-dialin.net) left irc: Ping timeout: 480 seconds [04:12] time is in short supply.. as is money (for the time being) [04:12] it has to be useful for everyone in the vserver community ... [04:13] it's quite simple, if I spend some of my time fixing one persons stuff, this time is missing for the community ... [04:13] i understand.. that's why i'm asking you specifically [04:13] so that person has to make up for that time in some way, ... [04:13] what are some options [04:13] you could write/update some howtos ... [04:14] you could maintain a linux-vserver mirror, and think about how to set it up ... [04:14] hmm.. who hosts your cvs ? [04:14] you could write a CoolTool(tm) to aid in vserver whatever ... [04:14] well i'd like to make it easier to work with gentoo [04:14] as gentoo is THE devel platform [04:15] johnny: lets not start some kind of distwar in here ;) [04:15] requiring cds just isn't a good solution imo [04:15] I have no problem with that, if you want to become the Gentoo vserver specialist ... go ahead [04:15] like the redhat vserver tools [04:15] err the ones that require rpms [04:15] Action: johnny is no friend to rpms [04:15] debian is alright... [04:15] apt-get is good tool [04:16] how much does the fedora thing use apt ? [04:16] apt-rpm works too, and Redhat, mandrake and suse are nice too ... [04:16] it's perfectly okay to have a favorite distro ... [04:16] yeah.. apt-rpm [04:16] and if you like we can play Iliads DistroWars ;) [04:16] cool [04:17] the thing i didn't like about rpms specifically (besides dependencies) is that you have to go somewhere to find stuff to install [04:17] i guess that was a few years ago though [04:18] probably ... getting an rpm from enricos util-vserver means doing rpm -bt util-vserver-0.27.tar [04:18] and what debian seems to lack is marking files as binary/config/etc ... [04:19] otherwise a port of the vunify would be straight ... [04:20] hm... what's that marking stuff? could you elaborate? guess i don't know what you mean, since i'm only used the debian systems... [04:21] RPM uses spec files, describing the build process, similar to debian [04:21] and contain a files section, specifying which files to add to which package ... [04:22] %files [04:22] %defattr(-,root,root) [04:22] %doc ANNOUNCE ARTICLE READ.ME [04:22] %{_bindir}/* [04:22] %{_mandir}/man1/* [04:22] looks like this for example ... [04:23] johnny (~johnny@ip68-10-185-29.hr.hr.cox.net) left irc: Read error: Connection reset by peer [04:23] %config(noreplace) %{_sysconfdir}/emacs/site-start.d/* [04:23] this for example specifies config stuff ... [04:23] all files not mentioned are dynamically created ... [04:24] so you can tell which files are and will remain constant and what files will be changed by the user ... [04:25] johnny (~johnny@ip68-10-185-29.hr.hr.cox.net) joined #vserver. [04:25] i missed everything you said [04:25] for the last 2 min [04:25] damn pre release gnome.. [04:25] or 2.6 kernel [04:25] one of em.. [04:27] well thats the danger of living on the edge ... ;) [04:27] aah, i see, you describe a bunch of rpm's in a single file, for example the packages xyz, xyz-devel and xyz-doc and build them out of one tree, right? [04:27] yes could be, but that is not what is important ... [04:27] (at least for unification) [04:28] it is important, to differentiate the files in a package ... [04:28] i've been using 2.6 since 2.5.49 [04:28] man.. i love the drop shadowing patches for gtk.. [04:29] bertl: can you outline the descent setup you mentioned? because after our little conversation, i think i would go for something like: 50mb /boot; 3gb /; 500mb rescue; 500mb swap; rest data [04:29] Doener: basically you have to tell if a file is going to be changed by the user/config tools or will remain unchanged over the lifetime of that package (version) [04:30] maharaja: I would suggest a small variation ... [04:30] well actually not so small ... ;) [04:30] first, you can use the rescue for the boot stuff, if you like to [04:31] but a separate boot is no problem either ... [04:31] / should not need more than 200MB, and /usr, /var (maybe /tmp) should be on separate partitions ... [04:32] I usually put /usr/src and /home on a separate partition too [04:33] so as LVM is your friend, you would reserve about 4-5GB for all that and call it 'vgs' or something like this ... [04:33] that's where lvm is good :) [04:33] how much memory du you have? [04:33] bertl: you mean 4-5gb vgs for /usr, /var, /tmp, etc [04:34] bertl: 1gb [04:34] okay, do you plan to upgrade soon? [04:34] no [04:34] okay, then 500MB swap is more than sufficient ... [04:35] /dev/vgs/slash 248M 63M 172M 27% / [04:35] /dev/vgs/usr 992M 284M 656M 31% /usr [04:35] /dev/vgs/var 992M 192M 748M 21% /var [04:35] typical server setup ... [04:36] debian has at least a way to separate binaries from configuration files, when uninstalling a packages one can choose between remove (leave config) and purge (remove everything)... [04:36] Doener: well, that is a start, what about dynamic files like /var/log/messages? [04:36] doener: afaik, this is managed by the post-uninstall scripts [04:36] doener: or am o wrong [04:37] i'm currently reading debian's maintainer docs... let's see if there's a clean way to do a separation... [04:39] bertl: isn't lvm on raid a lot of overhead? [04:40] not really ... [04:40] but if you prefer separate raid partitions, no problem with that ... [04:41] ok, so i guess i have to leave this computer and get my hands dirty again ;) [04:41] best start with the rescue partiton ... [04:42] and do everything from that partition, this way you can make sure everything you need to maintain the system is there .. [04:43] thank you [04:43] i'll try that [04:46] bbl [04:47] maharaja: i guess so, i'll know for sure after reading the docs... [04:47] hmm... my connection is plain bad tonight... only getting messages in bunches every 5 minutes or so... [04:58] Doener (~doener@pD9E12FAD.dip.t-dialin.net) left irc: Quit: Leaving [04:58] Doener (~doener@pD9E12FAD.dip.t-dialin.net) joined #vserver. [04:59] Simon (~sgarner@apollo.quattro.net.nz) joined #vserver. [05:02] hi simon! [05:08] maja (maharaja@is.the.one.who.rules.at) joined #vserver. [05:08] re [05:08] bertl: your rescue system is raided too? [05:08] nope, no need for that .. but I have two copies ;) [05:09] so how do you boot a remote server into the rescue system? [05:09] and of course two master boot blocks (with grub) allowing to boot from both rescue partitions ... [05:10] serial line to the server and serial console for everything starting with grub ;) [05:11] mhm, thats far too complex ;) [05:12] at least for me [05:13] well it's simple to maintain ... [05:14] currently, i installed the rescue system, and i do not know how to proceed properly ... [05:14] make a setup, which can boot both rescue partitions, and dd master boot record + rescue to the second one ... [05:15] isn't there a problem with the different layout of the two harddisks? [05:15] well, did you configure a differing layout? [05:15] no, but the use a differenet chs layout [05:15] let me have a look at your 'current' partition table [05:15] tables ;) [05:18] /dev/hda1: start 63, size 97713 [05:18] (the first but for sure not last time a serial console would come handy) [05:18] /dev/hda2: start: 97776, 976752 [05:19] /dev/hdb1 the same values [05:19] (sfdisk) [05:19] exactly the same? [05:19] yes [05:19] but cfdisk shows me a difference in the size [05:19] well, then ther will be no issues, right? [05:20] linux doesn't use the disk geometry, it uses the partition geometry ... [05:20] so no problem there ... [05:21] the 48M partition is the /boot, right? [05:21] yes [05:22] okay, make sure that the rescue part can boot without /boot ... [05:22] so it should be bootable without mounting or referencing any other partition [05:23] mhm, actually, i made a mistake [05:23] hdb hasn't got the same size [05:23] well, no problem there ... [05:23] hdb1: 63, 96327 [05:24] first tep: dd if=/dev/hda of=/dev/hdb bs=1M count=500 [05:24] step, I mean ... [05:24] hmm, make that count=600 [05:25] this will copy both partitions including the partition table to hdb [05:25] mhm, somehow you lost me :) [05:25] make sure no partition from hdb is mounted ... [05:25] dd if=/dev/hda of=/dev/hdb bs=1M count=600 [05:25] i installed the rescue system to /dev/hda2 [05:25] and it boots with lilo [05:26] yes, perfect, do the dd (which copies all that to hdb) [05:26] ok [05:27] /dev/hda has no content though, is that correct? [05:27] dd done [05:27] you are using devfs? [05:27] not on the rescue system [05:27] /dev/hda1 [05:27] argl [05:27] well then hda should contain the entire disk ... [05:27] im a little tired [05:28] i meant, that i have no kernels/... on /dev/hda1 [05:28] ignore hda1 .. [05:28] so only /dev/hda2 contains some data (the rescue system) [05:28] ok [05:28] so my dd is done [05:28] what was the output of the dd? [05:28] 629145600 bytes transferred in .... [05:29] okay ... [05:29] now reboot the machine ... [05:29] and do what? [05:30] enter the rescue system again ... [05:30] ok [05:30] brb [05:30] have a look at the partition tables ... [05:33] ok, they exactly match [05:34] good .. now create your raid/lvm partitions ... [05:34] on hda and hdb? [05:35] or on hda only, and again dd? [05:36] well I would create an appropriate setup on hda and dd it over to hdb (only 512 bytes are required to do that) [05:36] but be careful, the mkraid needs both partitions ... [05:36] i c [05:36] ok, *working* [05:43] rebooting [05:43] fyi [05:43] ;) [05:51] too bad, now i'm wasting 1gb of space in /dev/hda [05:51] but otherwise, the partition layout is ident [05:52] no prblem with that, you can use the 1GB as scratch space ... [05:52] now, i create my raid volumes [05:53] night folks [05:53] Nick change: Doener -> doener_zz [05:53] night [05:54] ha, i forgot the raidtools ;) [05:54] see what I meant with working from the rescue partition ;) [05:55] yeah :) [05:58] so i guess, i shall install everything i need, and then do i final dd if=/dev/hd2 of=/dev/hdb2 ? [06:00] whats a good chunk sizeß [06:00] whats a good chunk size? [06:01] 32? 4? [06:01] on modern machines usually the larger the better ... [06:01] anything over 64k doesn't make to much sense though ... [06:02] s/to/too/ [06:02] so you recommend 32 to 64 [06:02] i used 32 for the swap space [06:02] well that's probably okay ... [06:03] probably? :) [06:03] if you want to find out, you have to test with different sizes and something like bonnie ... [06:03] do not forget to adjust the raid stripe parameters of your filesystem accordingly ... [06:04] what are my "raid stripe parameters of my filesystems"?! [06:04] never heard of that [06:04] Sirius_Black (kollege@pD9FFBE72.dip.t-dialin.net) joined #vserver. [06:05] man mke2fs ;) [06:05] -R raid-options [06:05] stride=stripe-size [06:05] just an example ... [06:06] ok [06:23] ok, i created the raid volumes (one is still syncing), mkswap, swapon, pvcreate, vgcreate, lvcreate /dev/vg0/slash [06:23] (nice name *g*) [06:24] okay, good ... now you can either setup your server system, or populate those lvm partitions with the rescue setup ... [06:24] do i need to specify the raid stripesize with lvm too? [06:24] no [06:25] usually I mount them on /slash, /slash/usr .... [06:25] so i can mkfs.ext3 /dev/vg0/slash [06:25] without the -R parameter [06:25] sure, I would suggest ext2 for /slash and /slash/boot ... [06:26] whats the problem with using ext3 for the root partition? [06:26] and ext3 for everything else ... but that's up to you [06:26] I do not fully trust ext3 yet ;) [06:26] hehe [06:26] what do you trust then :) [06:26] iso9660 is pretty slid ;) [06:26] s/slid/solid/ [06:26] hehe [06:27] whats the best way of populating the /slash dir? [06:28] find -xdev -exec cp {} /slash \; [06:28] ? [06:28] nope [06:28] dump and restore are much simpler here ... [06:28] did you install them yet? [06:29] nope [06:29] /sbin/dump, /sbin/restore, /sbin/rdump, /sbin/rrestore? [06:29] looks good ... [06:30] ok [06:30] so you start with mounting the entire tree on /slash [06:30] /dev/vg0/slash on /slash [06:30] /dev/vg0/usr on /slash/usr [06:30] /dev/vg0/var on /slash/var [06:30] well I guess you get the idea ... [06:31] of course [06:31] then you change into /slash [06:31] cd /slash [06:31] and execute ... [06:31] actually, it's not the first linux system i install, but i've never done everything by hand [06:31] dump 0sf - / | restore rf - [06:32] ok [06:32] ill try ;) [06:32] (actually an esthetically pleasing command sequence, therefore easy to remember ;) [06:32] mhm :) [06:33] tell me more! :> [06:33] okay, is this working? [06:33] not yet [06:33] i'm waiting for the resync to complete and typing the mkfs for /home [06:34] do not worry about the resync, you can already use it ;) [06:34] that's the whole idea behind raid ;) [06:34] i know [06:34] but somehow, i enjoy watching the rebuild process [06:34] *g* [06:34] ok, i issued mkfs.ext3 -R stride=32 /dev/md1 [06:35] actually on 32k raid blocks stride should be around 8 [06:36] because stride as mentioned in the man page ;) is in fs blocks ... [06:36] mhm [06:36] let me check :) [06:36] and if you didn't issue any other options, the fs block size is 4k [06:37] the chunk size in the raidtab is defined in kbytes? [06:37] yes, IIRC ... [06:40] on the dump/restore sequence, i get a "restore: tape read error on first record" [06:41] that is bad ... should not happen ... [06:41] means that the dump somehow failed ... [06:42] ahh I forgot ... [06:42] dump 0sf 100000 - / | restore rf - [06:42] that is the right sequence, sorry ... [06:42] ok [06:42] its working [06:43] restore: ./home: exists [06:43] that is okay ... [06:43] bla bla bla [06:43] dump is done [06:43] well perfect ... you copied the entire / onto /slash/* [06:43] i do a dd if=/dev/hda2 of=/dev/hdb2 to copy the rescue disk [06:43] so i won't forget to do that [06:44] you can do that ... [06:44] but you should do that, when you booted from /slash [06:44] ok [06:44] because then the rescue system is not mounted ;) [06:45] so, the only thing which is missing is a correct lilo install, that can boot from all 3 partitions [06:45] am i right? [06:45] yes ... [06:46] it's a little trickier to get that right with lilo ... [06:46] because lilo 'remembers' the position of the kernel ... [06:46] i once read about that [06:46] mhm, so lets go for grub then [06:47] i installed it once, 1.5 yrs ago ;) [06:47] whatever you prefer ... [06:48] well, i always needed another reason to try grub once more [06:51] mhm, i do not remember using a floppy to install grub.. [06:51] it's not required ... you just start grub [06:52] and from the grub shell you do setup (hd0) [06:59] mhm [06:59] how do i boot from /dev/vg0/slash? [06:59] well this is on lvm, so you need an initrd ... [06:59] its on lvm which is on md-raid ;) [07:00] lvmcreate_initrd will do nicely ... [07:00] ok, ill try to boot into the rescue system first [07:00] make sure raid auto detect + lvm is in kernel [07:00] then specify the initrd created by lvmcreat_initrd and as root device root=/dev/vg0/slash ;) [07:09] serving (~serving@213.186.191.236) left irc: Ping timeout: 480 seconds [07:11] ok, for today i quit with a kernel panic on boottime ;) [07:12] thank you very much for your help [07:12] hmm you did get one? [07:12] yes [07:12] what message exactly? [07:12] linux did not find /dev/console [07:12] and something about /dev2/root2 [07:12] i guess i messed up with the grub install [07:12] hmm, you edited the fstab, right? [07:12] no, i wanted to boot the rescue system [07:13] and tried [07:13] root (hd0,1) [07:13] kernel /vmlinuz [07:13] initrd /initrd.img [07:13] boot [07:13] hmm no rootfs? [07:13] ? [07:13] the root option in grub doesn't specify the rootfs ;) [07:14] it's the 'current' source for file lookup ;) [07:14] i do not understand [07:14] you need to specify root=/dev/hda1 for example ... [07:14] where? [07:14] after boot? [07:14] no on the grub boot line ... [07:15] mhm, grub said that it found the fs [07:15] and i could "cat /boot/grub/grub.conf" [07:15] root (hd0,0) [07:15] kernel --console-option /vmlinuz-2.4.23vs1.22 root=/dev/vgs/slash nmi_watchdog=1 panic=10 [07:15] initrd /initrd-lvm-2.4.23vs1.22.gz [07:15] ignore the --console-option ... [07:16] ok, ill try tomorrow [07:16] okay ... [07:16] thnx [07:16] nite [07:28] serving (~serving@213.186.189.68) joined #vserver. [07:31] hmm, serving? [08:00] Nick change: WSU -> WSU_away [09:02] serving (~serving@213.186.189.68) left irc: Read error: Connection reset by peer [09:06] SG1 (~sgarner@apollo.quattro.net.nz) joined #vserver. [09:06] morning Simon! [09:07] Simon (~sgarner@apollo.quattro.net.nz) left irc: Read error: No route to host [09:11] serving (~serving@213.186.189.68) joined #vserver. [09:11] HI all [09:11] morning serving! [09:11] :) [09:19] is there a problem to have 192.168.0.0 on the host and a routable IP on the vservers ? [09:32] hmm 192.168.0.0 is the network ip ... [09:33] so I would not use that for setting up an interface ... [09:33] s/ip/address/ [09:42] I know bert. I am thinking of having 192.168.0.0 subnet IPs used for the host. would this provide any added security ? [09:45] hmm, well sure, even a firewalled public net (with restricted access) would provide security there ... [09:46] I am running iptables on the host and I am denying evrything except the ports needed by the vservers [09:47] But , I have ldap running on a vserver and I would like to permit access to it from my 192.168.*.8 subnet only [09:47] hmm well if the iptables setup is flawless, this would not add any security [09:47] i c [09:48] at least I do not see how it would ... [09:49] I figured that with 192.168.0.* on the host, an attacker can not reach th ehost only its vservers [09:51] hmm, well he could from the vservers, couldn't he? [09:54] any attempt to connect to vserver ip:25 will reach the mail servers runing on that vserver and not the host mailserver . right / [09:54] ? [09:55] well, if configure properly, yes ... [09:55] if the host runs a mta which binds to 'all' ports ... then no [09:57] ofcourse . so having a non routable IP for the host will eliminate the possibilty of an attacker rwaching to its services. [09:57] okay, maybe we should sync on the terminology ... [09:57] the question remains. what can "they" do if they gain access to a vserver ? [09:57] non routable != private [09:58] yes [09:58] public ips / ranges could also be non-routable [09:58] wait. [09:58] ok. I mean private [09:58] private networks can and will be routeable ... [09:58] 192.168.0.0 are routable on th enet ? [09:59] sure, and especially on the vserver host ;) [10:00] usually routers (per configuration ;) block such ranges ... but that depends on the configuration ... [10:01] right , so my 192.168 is blocked at the ISP router andif it isnt there no way for you toconnect to my 192.168.0.* boxes . right ? [10:02] well, yes for the remote acces, this is right, what about breaking into a vserver and then accessing the host? [10:03] that was the question :) [10:03] the question remains. what can "they" do if they gain access to a vserver ? [10:03] localhost is still pointing to the host [10:03] hmm okay, must have missed that particular question ;) [10:03] what version do you use? patches? [10:04] 1.2.22 [10:04] .29 rpms [10:04] how does your /proc look like on the host? (ls /proc) [10:05] yes [10:05] hmm, 'yes' is an interesting answer to that question! [10:06] i can see the proccesses of the host from vserevr [10:06] th els /proc output was identical on the vserver and the host [10:07] well, if that is the case, then something is broken ... [10:07] tanjix (~tanjix@p5091D229.dip.t-dialin.net) joined #vserver. [10:07] serving: could you paste the output of 'ls /proc' here? [10:08] ls /proc [10:08] 1 2063 2477 2530 2543 bus ide mdstat self [10:08] 2021 2080 2479 2531 2544 cmdline interrupts meminfo slabinfo [10:08] 2037 2081 2487 2532 2545 cpuinfo iomem misc stat [10:08] 2048 2357 2496 2533 2546 devices ioports modules swaps [10:08] 2049 2424 2509 2534 2547 dma irq mounts sys [10:08] 2050 2433 2510 2535 8782 driver kcore mtrr sysvipc [10:08] 2054 2436 2514 2536 8832 execdomains kmsg net tty [10:08] 2055 2446 2515 2537 8943 fb ksyms partitions uptime [10:08] 2056 2458 2516 2539 8945 filesystems loadavg pci version [10:08] 2057 2476 2523 2542 8980 fs locks scsi [10:08] ide or scsi harddisk? [10:08] scsi [10:08] hmm, well then I would upgrade to vs1.23 and fix the proc entries ... [10:09] what is the expected output ? can you mesg me yours [10:10] well I just wanted to see what entries are there on your system, this varies from system to system and setup to setup ... [10:10] ok [10:10] 1 3 9driver iomem locks self sysvipc [10:10] 10 4 busexecdomains ioports meminfo slabinfo tty [10:10] 11 5 cmdlinefilesystems irq misc stat uptime [10:10] 15 6 cpuinfofs kcore mounts swaps version [10:10] 16 7 deviceside kmsg net sys virtual [10:10] 2 8 dmainterrupts loadavg partitions sysrq-trigger [10:10] this is my test system, for example ... [10:11] how could you tell that mine is broken ? [10:11] well, you said vs1.22, and you ahve scsi for example ... [10:12] it could be worse, if you had magic sysrq enabled ;) [10:12] so devices should not show .... [10:13] well, on the host they usually show ... [10:14] although you can configure them not to show there, with vs1.23 [10:14] but what counts is that they probably also show on the vserver ... [10:15] 1.23 have that conf option where ? [10:15] it's enable per default, you 'tune' the entries with the vproc tool ... [10:16] i c [10:16] can I keep using .29 rpms with 1.23 ? [10:16] yes, nothing changes, but 0.29 seems still broken :( [10:17] at least I got a bunch of bug reports ... [10:17] maybe I should give .27 a try [10:17] :) [10:17] util-vserver, yes, vserver no ... [10:18] ;)) [10:19] I'd expect jack to update/maintain/fix the tools, but it seems he's still too busy ... [10:20] his code is totaly different that yours ? [10:20] s/that/than [10:20] well, I don't code the userspace tools atm, enrico does util-vserver and jack vserver-tools ... [10:21] I'm the kernel guy ... [10:21] ic [10:22] but util-vserver and vserver-tools provide the same functionalities ? [10:22] basically yes ... [10:23] virtuoso (~shisha@ip114-115.adsl.wplus.ru) joined #vserver. [10:23] hi virtuoso! [10:23] Bertl: Hi. [10:23] Bertl: You're an early bird today. :) [10:24] well, I'm actually a left-over ;) [10:25] I'm hacking on the 'new' procfs and just can't stop ... [10:26] New means 2.6.x? [10:26] well, yes 2.6 and 2.4 .. but actually it means /proc/virtual/ ;) [10:28] I recently found an old module that captures chdir() calls and shows in some procfs entry how much did it capture. :) [10:28] Funny. [10:52] okay, it seems to work now ... will continue at the evening ... [10:52] have a nice day/night everyone ... [10:53] Nick change: Bertl -> Bertl_zZ [10:53] you too Bertl_zZ [11:02] tanjix (~tanjix@p5091D229.dip.t-dialin.net) left irc: [11:53] Doener_zZz (~doener@p5082D655.dip.t-dialin.net) joined #vserver. [12:01] doener_zz (~doener@pD9E12FAD.dip.t-dialin.net) left irc: Ping timeout: 480 seconds [13:30] SG1 (~sgarner@apollo.quattro.net.nz) left irc: Quit: so long, and thanks for all the fish [14:06] lp (~lpressl@interner.SerNet.DE) joined #vserver. [14:25] infowolfe (~infowolfe@pcp04891550pcs.frnkmd01.md.comcast.net) joined #vserver. [15:53] damjan (~damjan@legolas.on.net.mk) joined #vserver. [16:22] kernel408 (~cbc55a28@humbolt.nl.linux.org) joined #vserver. [16:30] i am doing a project for server consolidation for which i would be giving a group of tasks preference in memory management [16:35] i would be using the O(1) VM model developed by Riel [16:35] but i am facing a problem in deciding how to segment the available memory in group of three [16:36] could some discuss it with me? [16:36] or guide me.... [16:36] sorry, that's nothing i feel competent for ;) [16:36] *switching servers [16:38] hmmm [16:38] all developers sleeping, i think [16:40] maharaja :one of the target application of my project is v servers [16:41] the linux memory is divied into 3 zones on i386 [16:41] more ever processes ask memory from the slab allocator [16:42] so i am not able to logically divide memory in 3 areas [16:52] hepp [17:06] how do i tell grub to load the appropriate configfile? [17:06] so i see the menu after booting [17:07] and not the command line [17:22] kernel408 (~cbc55a28@humbolt.nl.linux.org) left irc: Quit: CGI:IRC 0.5 [17:36] menu.lst [17:37] it cheks for /boot/grub/menu.lst or /grub/menu.lst automatically [17:37] but also there's a command to load it. I think its "file", "config" or something like that [18:09] thnx [18:09] allready found it [18:09] my new problem is [18:10] that i cannot mount my root device, which is on lvm, which is on md0 [18:16] damjan (~damjan@legolas.on.net.mk) left irc: Ping timeout: 480 seconds [18:26] damjan (~damjan@legolas.on.net.mk) joined #vserver. [19:05] maja (maharaja@is.the.one.who.rules.at) left irc: Quit: leaving [19:14] nathan_ (~nathan@209-6-130-26.c3-0.sbo-ubr1.sbo-ubr.ma.cable.rcn.com) joined #vserver. [19:28] Nick change: Bertl_zZ -> Bert [19:28] Nick change: Bert -> Bertl [19:28] hi all! [19:30] hi [19:36] hi damjan, how's your vserver? ;) [19:37] i bertl [19:37] i managed to install and boot my system :> with slight modifications [19:38] if you show me the final method to backup my real root/usr/... partitions, i would gladly do some testing with the vserver patches [19:38] (later in the evening though - i've got to clean my flat) [19:40] Bertl: I dont have one... yet. [19:41] welll, get one! fast! ;) [19:41] maharaja: well same as with the copying, dump is your friend ... [19:41] in this case dump 0zf /backup/usr.dump /usr [19:42] (for example) [19:42] morning everyone [19:43] morning nathan! [19:44] Nick change: Doener_zZz -> Doener [19:45] hi all [19:45] morning Doener! [19:45] Bertl, have anything more recent than http://vserver.13thfloor.at/Experimental/patch-2.4.24-vs1.3.4.3.diff? [19:45] i saw you ported it to stable [19:45] yes I have, but I'm not finished yet ... [19:46] actually the 'stable' version is working (the 1.3.4.3 not) [19:46] ah [19:46] so i should wait till your done until i play with 1.3.4.3? [19:47] well, yes, should be today or tomorrow, that I finish 1.3.5 ... [19:47] ah ok [19:48] but if you want to bash on something, vs1.23 would be a good candidate ... [19:51] building now [19:56] looks like a vproc --dump/--restore wouldnt hurt? [19:56] hmm, go ahead, write it ;) [19:57] well do you think it could be userful or is it just an extraneous feature? [19:57] guess it would be useful ... [19:58] vxproc is going to have the same interface under 1.3 right? [19:59] yup, only that the devel version will default to all entries off ... [19:59] ah [20:18] hmm, obviously I needed the sleep ... now everything is clear and makes sense ;) [20:19] :) [20:19] before you go to sleep, what do you want me to test? [20:19] or nathan: do you know what one could test on an smp machine? [20:19] yes, especially vs1.23 would be testworthy ... [20:20] but please in any case, compile with gcc 2.95/2.96 and not with gcc 3.1.x or 3.2.x, >= 3.3.2 might work ... [20:21] maharaja, still stress the proc system on smp with 2.9 gcc just to validate my findings [20:23] maharaja: by the way, did I thank you for your testing yet? well anyway, thank you very much, I really appreciate it! [20:24] nathan_: same to you! [20:25] hey i use the software, might as well help :) thanks to you hehe [20:33] gm Bertl [20:33] hi infowolfe! [20:33] coding is really fun today! [20:34] cool! [20:39] i knew i came in here earlier looking for something specific... i just don't remember what it was :-p [20:39] :) [20:39] oooh [20:39] james noble reported an apache-fp/freevps problem to me and said it had something to do with mm earlier this morning [20:40] and i was going to ask the status of memory accounting :-D [20:40] Bertl, is there any clean way to figure out the default flags? [20:40] otherwise ill just dump everything i guess [20:41] nathan_: hmm default flags? [20:41] he also reported that with vs1.20 the same thing that killed his verservers via freevps didn't [20:41] Bertl, yea the default getxflg(), you said devel was going to have all off by default and stable all on by default [20:41] infowolfe: I had some memory accounting patch for 1.3.x [20:42] nathan_: ah yes, well, no actually I would not rely on that, or if then only with an option (assume all on/off) [20:42] had? [20:43] Bertl, ok then ill just dump everything instead of trying to filter defaults [20:43] Bertl, also, any news on depreciating jaques' vserver-0.29 tools? [20:43] nathan_: guess that is better, but you can avoid changing values if the do already match ;) [20:43] Bertl, is the overhead of getting/conditional setting less than that of just setting all the time? [20:44] infowolfe: well, have a pile bug reports here,seems that 0.29 is pretty broken :( [20:44] Bertl, unless jaques is going to handle the bugs, my vote is for depreciation... [20:44] nathan_: well, it will prevent trying to set entries which are not modifyable ... [20:44] enrico seems to be very quick to act on bug reports :-( [20:45] Bertl, it wont dump anything that getxflgs failed on [20:45] Bertl, so /proc/1/ will never show up in a dump, along with all its entries. [20:45] r = !getxflg(fd, &flags); [20:45] close(fd); [20:45] if (!r) continue; [20:45] printf("%s/%s %d\n", cwd,name, flags); [20:45] Action: infowolfe thanks Bertl and ensc profusely for picking up the linux-vserver project, and thanks everyone else that has helped [20:45] okay ... then I guess it doesn't matter ... [20:45] ya [20:46] infowolfe: well guess we enjoy it, especially the great community! [20:46] Action: infowolfe smiles [20:48] well as I always say, ' one can dream, many can accomplish!" [20:49] guess Linus would have given up by now, if there wasn't a community ... [20:55] ok dump/restore hacked in [20:56] http://vserver.13thfloor.at/Experimental/patch-2.4.24-vs1.3.5pre1.diff [20:56] well, guess you win ;) [20:56] this prerelease has a completely rewritten /proc/virtual! [20:57] Bertl, what indent flags do you use? [20:58] hmm, -kr (what else ;) [20:59] well, actually I indent the C code manually ... [20:59] (my C code ;) [20:59] ah [20:59] well you will probably barf at my tabs [20:59] np [21:00] but ill give you a diff and you can sort it out [21:00] indent -kr will do nicely ... [21:00] indent -kr seems to be breaking the diff [21:00] hmm you are not indenting the diff, do you? [21:01] nope [21:01] i dont mean its breaking the actual diff, but when diffing, it has too much stuff [21:01] okay, then ident -kr the unmodified ... and the modified ... [21:01] http://0x00.org/hidden/vproc.c.diff [21:01] ok i can do that [21:01] hang [21:01] better [21:02] k its on that site [21:02] okay, two notes on that: [21:03] a) flags: don't use -x it's used for the xid elsewhere ... use -R -S instead ... [21:03] ah [21:04] b) is absolute path such a good idea? or did I miss something? maybe /proc relative would be useful?! ... [21:04] hmm [21:05] but I leave that to you, just make sure that you publish the modifications on the ml ... [21:05] i guess proc relative is easy enough [21:06] *building the new kernel [21:06] are you saying the relativity because of a security standpoint? [21:07] i dont really see any advantage or disadvantage to doing it one way or the other [21:07] bbl when the kernel is build [21:07] <- short restart [21:07] hmm, well brainstorming my thougths: [21:08] - maybe shorter path names [21:08] Sh[a]de (shade@cpe109.bb101.cablesurf.de) left irc: Quit: Excursion (On IRC.BONGSTER.DE [#wwip, #german-elite and #lov]) [21:08] - maybe ability to use on /proc mounted elsewhere [21:08] - maybe additional issues (could be solved with -p option) [21:09] well that's it ... [21:09] hmm [21:10] so -p would specify the location of proc for restore/dump? [21:10] yup [21:11] i cant think of in what situation that would be useful but ill do it :) [21:11] hey, don't do it if you do not think it is useful?! [21:11] where are we, implementing stuff which isn't considered useful at all? ;) [21:12] arent all proc filesystems going to share the same entries reguardless of mount points? [21:12] probably .. unless proc is modified in the future (not likely) [21:13] an interesting detail of this feature (vproc) is, that you are able to disable the entry on the host ;) [21:13] would that be -E? [21:14] yes, and you can only change it back, from xid=1 then ... [21:14] intentional? [21:14] well, it is natural, because ioctls don't work on non existing files ;) [21:15] well sure, but i would have thought that the host would see anything regaurdless of the vx proc permissions [21:15] this was the reason I didnt add the hide everywhere feature ... which actually requires a reboot ... [21:16] /tmp/vproc -f 4 /proc/cpuinfo (will do that, for example ;) [21:17] ah [21:18] ok uploaded new diff with relative paths to proc [21:18] ill mail it to the mailing list in a bit [21:19] before I forget, the 1.3.5pre1 has the vproc stuff (the disabled default) and the new /proc/virtual .. so if you like to test ... [21:20] ah ok [21:20] will do [21:23] okay, cu later, have to leave now ... will be back around midnight CET ... [21:24] k, later. [21:24] Nick change: Bertl -> Bertl_oO [21:48] Doener (~doener@p5082D655.dip.t-dialin.net) left irc: Ping timeout: 501 seconds [21:52] Nick change: damjan -> damjan_not [21:55] Sirius_Black (kollege@pD9FFBE72.dip.t-dialin.net) left irc: Quit: 12( www.nnscript.de 12:: NoNameScript 3.8 12:: www.4XLhost.de 12) [21:56] Doener (~doener@pD9E1209F.dip.t-dialin.net) joined #vserver. [22:05] damjan_not (~damjan@legolas.on.net.mk) left irc: Quit: Leaving [22:14] surriel (~riel@riel.netop.oftc.net) left irc: Ping timeout: 480 seconds [22:30] Tamama (Pluh@a62-216-20-152.adsl.cistron.nl) joined #vserver. [22:51] Simon (~sgarner@apollo.quattro.net.nz) joined #vserver. [22:59] nathan_ (~nathan@209-6-130-26.c3-0.sbo-ubr1.sbo-ubr.ma.cable.rcn.com) left irc: Ping timeout: 480 seconds [00:00] --- Wed Jan 14 2004