[00:25] Doener_zZz (~doener@pD9E12AE2.dip.t-dialin.net) joined #vserver. [00:32] Doener (~doener@pD9E129DB.dip.t-dialin.net) left irc: Ping timeout: 501 seconds [00:39] Doener_aw (~doener@pD9588B3D.dip.t-dialin.net) joined #vserver. [00:41] Nick change: Doener_aw -> Doener [00:42] seems my isp has some trouble holding my connection... [00:44] seems so ... [00:47] Doener_zZz (~doener@pD9E12AE2.dip.t-dialin.net) left irc: Ping timeout: 501 seconds [00:55] *g* I've got too many balls in the air... [00:56] patching a networks ircd, writing my own ircd, due to that I don't like the others, fixing the damned SAN, and looking for a new gf... [00:58] and on top, admining the fscking network. [00:58] and boxes and users and bleeeh, printers and peripherals [01:05] hmm, writing your own ircd? aren't there enough out there? [01:07] (or is this a religious question ;)? [01:59] suhcoolbro (~Suh@216-161-89-245.ptld.qwest.net) joined #vserver. [01:59] hi suhcoolbro! [02:00] yo. [02:07] ozan (~ozan@dsl81-215-16961.adsl.ttnet.net.tr) joined #vserver. [02:07] hello [02:08] hi ozan! [02:10] suhcoolbro (~Suh@216-161-89-245.ptld.qwest.net) left irc: Quit: NO CARRIER [02:35] ozan (~ozan@dsl81-215-16961.adsl.ttnet.net.tr) left irc: Quit: bzzzt [02:37] nox (~nox@b109150.adsl.hansenet.de) joined #vserver. [02:43] hi nox! [02:44] hi bertl [02:45] how are you? what brings you here? [02:45] did someopne here ever try vserver+ adamnatix patch ? [02:45] adamantix [02:45] what is adamantix? [02:45] oh i am fine and love the vserver idea [02:46] adamantix is a fork from woody [02:47] it has pax and rsbac included and everything is comp with smash-guard gcc [02:47] hmm, means a distro ;) [02:47] a patched kernel [02:48] hmm, okay and you want to use this kernel, not replace it with a vanilla kernel, right? [02:48] like the smash-guard idea against bufferobverflows [02:48] i have no experience witg c and failed to often to combine patches [02:49] well, are those adamantix patches to the base kernel available somewhere? [02:49] yes [02:50] mom [02:53] http://www.szczepanek.de/adamantix/dists/ the german mirror [02:56] the prob is the special gcc. they try to translate more and more pack with it. W^X is asured by pax . So they try come closer to obsd . [02:58] Didn?t want to make a big point of it. i try it when bit more time .... just thought you maybe directly say "no way " [02:58] hmm, well that's nice, but what I need to have a look at it, is some patch-2.4.24-adamantix-27.diff [02:59] or whatever kernel/version they use ... [03:00] u have devbian ? [03:00] deb has this patch in tree [03:01] no I don't have debian, but there is a vserver version for debian if you mean that ... [03:01] i know *g* [03:02] here i meant the "kernel-patch-adamantix - Kernel patches introduced in Adamantix" pack [03:02] look, I have a look at it, if you provide a 'adamantix' patch I can download, which applies cleanly to a vanilla kernel, okay? 8-) [03:03] i search 4 the source http://www.trusteddebian.org/ is the homepage [03:03] thx very kind [03:05] http://doener.homeip.net/adamantix/ [03:06] hmm, not very up to date, but I guess this is debian heritage ;) [03:06] i wonder why the 2.4.21 patch is just 500k but the 2.4.22 patch is 3.3m [03:07] probably it reverts all 2.4.22 stuff *g* [03:08] hehe [03:09] even more i wonder if those will apply to vanilla or just to a debianized kernel :) [03:11] well against vanilla they don't ... [03:12] you still have the debian patches? [03:12] afaik they only did pax and rsbac [03:12] somewhere ... [03:13] the question is now, which one to take ;) [03:15] the 2.4.22-5 fails too ... [03:15] okay is on hold until vanille-2.4.22 adamantix patches are somewhere ... [03:15] you did apply them in order, didn't you? [03:17] well, I took the applied 2.4.22-5 and made a copy ... [03:17] the one we had for debian ... [03:17] ah ok [03:18] but there is 2.4.22-3,4 and other things too ... [03:28] nathan_ (~nathan@209-6-130-26.c3-0.sbo-ubr1.sbo-ubr.ma.cable.rcn.com) joined #vserver. [03:28] hi nathan_! [03:28] heya [03:28] busy busy busy :/ [03:28] just finished setting up a fbsd jail [03:28] i prefer vserver heh [03:28] Nick change: Doener -> doener_na [03:29] I have your favorite vserver patch on the test ... [03:29] really? [03:29] 1.3.6 ... [03:29] ah, gimme a delta :) [03:29] rock solid including /proc/virtual ... [03:30] ill take it for a ride around the block [03:30] hows 1.2 doing? [03:30] guess 1.24 fixed the issues once and for all ... [03:30] dough, I said it again! 8-) [03:31] what about mah crasing it? [03:32] well that was fixed in 1.24 ... what he tested was 1.23 ... [03:32] but we narrowed the issue down ... and finaly fixed it ... [03:32] oh whoops i guess i missed a version while i was away [03:32] Action: nathan_ takes a look at the delta [03:33] interesting delta [03:33] removal of ctx_ref_lock entirely? [03:33] nope [03:33] well in sched.c it looks [03:33] only in two places, where not necessary, and one of them actually caused the trouble ... [03:34] where does it go wrong? [03:34] in line 170/177 the goodness() stuff ... [03:35] hmm whats the scenerio? [03:35] if I traced the pathes right, this leads to a deadly embrace between the task_list lock and the ctx_ref_lock ... [03:35] and what's especially funny, with interrupts disabled ... [03:35] so no way to get a stack dump ... [03:36] (well nmi opser should have helped ... but) [03:37] hmm interesting [03:38] so both trees sound stable now! [03:39] yeah, only the kernel isn't stable ;) [03:39] well 2.4.25-pre6 hopefully is ... [03:40] hmm should i have been reading lkml? [03:40] well vserver ml would be sufficient ;) http://list.linux-vserver.org/archive/vserver/msg05930.html [03:41] sigh i need to subscribe to that [03:41] lets do that now [03:43] http://vserver.13thfloor.at/Experimental/patch-2.4.25-pre6-vs1.3.5.2.diff [03:43] currently I'm adding XFS support for IUNLINK ... [03:46] hmm so i guess i need to go to 2.4.25-pre6 [03:46] ok all subscribed [03:46] good [03:47] is this the one i want to send me vproc.c diffs to? [03:47] yup ... for example ;) [03:48] in your killer tests, what was your average load? [04:07] loger joined #vserver. [04:10] hehe [04:10] it sounds so...impersonal :) [04:10] okay what about wossname then? [04:11] good wossname to you! [04:11] or 'have a fine wossname!' [04:11] hehe [04:11] well, i think it should end in "ing", just to be conformant with morning and evening [04:11] wossing? [04:12] and what about afternoon? [04:12] wossoon? [04:12] hehe [04:12] or night [04:12] wosst [04:12] hmmm [04:12] wassname may be the best choice after all [04:13] go for something that is non-controversial [04:13] i've been playing with a wiki for the last couple of days, they're good stuff [04:14] that's why we use one ... [04:15] yeah [04:15] the vserver use of it was what made me look at it again [04:15] re all [04:15] hi Sh[a]de! [04:16] hi there [04:16] Bertl: Short question about the php admin interface... [04:16] we have started a PHP Based V-Server Control [04:17] Short PBVSC [04:17] hmm, sounds catchy! [04:17] can we build it for vserver-0.29 or better for enrico's tools? [04:17] I would use util-vserver ... [04:18] for several reasons: [04:18] k the enricos one... [04:18] a) it's simpler to use in other languages [04:18] b) it's maintained and you can ask enrico [04:18] c) it's actually working 0.29 is not ... [04:19] ok, these are good reasons :) [04:20] well, I could add, and it follows the kernel development ... [04:20] well, ok i keep you and the Community informed about this project [04:20] make that, make a page in the wiki, or add a link, if you prefer your own pages ... [04:21] and let the folks know, via ml, if there is something to look at, test or you just want to get a decision ... [04:21] sure [04:22] we build a webpage for that and i add a link to the community wiki of vserver's pages [04:23] we hope for good participation and a good cooperation on this opensource project [04:28] well, I'll write the kernel patches for that, okay? [04:29] ok sure [04:32] well this pre6 isnt patching nicely into my tree [04:32] ugh [04:32] hmm ... your tree? [04:33] ah, okay you are trying to patch the current stuff up ... [04:33] my tree just being my directory of 2.4.24+patches [04:33] hmm a lot of this stuff that pre6 is trying to apply seems to already be there, i wonder if ck1 has some of it [04:35] ah got it from uv1 [04:36] hmm my 2.4.24 has been rather stable for me [04:36] i wonder if i should just leave it alone [04:37] hmm, where did you say is the short netconsole howto you wrote? ;) [04:38] hmm i never wrote it :) [04:41] too bad, how am I supposed to use it? [04:42] well i can give you a brief realtime howto until i get time to write something formal up [04:43] ive been spending all my time trying to get my box moved to this smp box because its costing me 2x the cost for having them both active [04:43] please, I would appreciate it ... [04:43] you have netpol for your driver? [04:43] how do I tell? [04:44] the netconsole module loaded without any issues ... [04:44] does your driver define poll_controller? [04:44] or rather assign poll_controller, it is in struct net_device [04:46] /sbin/insmod netconsole dev=eth0 target_ip=0xIPINHEX source_port=8888 target_port=8888 target_eth_byte0=0x00 target_eth_byte1=0xE0 target_eth_byte2=0x52 target_eth_byte3=0x11 target_eth_byte4=0x5E target_eth_byte5=0x56 [04:46] each target_eth_byte is an octet of the hardware address you want to send it to [04:46] and target_ip is just the local ip in hex [04:47] okay, what do I need to send the sysrq? [04:47] ipt_sysrq... [04:47] is this a separate patch? [04:47] http://www.terminus.sk/~marek/prog/ipt_sysrq.shtml [04:47] its an iptables module [04:48] just builds independently [04:48] ah okay ... [04:57] hmm lotsa changes in 1.3.5.2 [04:58] yup ... [04:59] im gonna need to patch a bunch of files by hand [05:06] Bertl, how critical are these updates compared to 1.3.5? [05:06] well, mostly stability ... [05:07] hmm so critical i guess [05:08] ahh, and the memory accounting stuff ... [05:08] you could revert that one ... [05:09] ah memory accounting is the major stuff i was seeing [05:16] hmm took a diff between pre4 and pre6 and got it to apply to get the critical stuff that the email talked about [05:16] i wonder if thats a bad idea [05:16] you'll see 8-) [05:17] that pre5 corruption error concerns me :/ [05:17] well, we had some dumps which just looked like that, right? [05:17] 1.3.5 has been stable for me, what should i take from .2 that is critical? [05:17] the entire /proc/virtual stuff ... [05:18] basically fs/proc/virtual.c [05:20] i was planning on hiding the proc/virtual in all contexts and host [05:20] would that solve the issue? [05:21] Action: nathan_ trying to avoid more by-hand patching [05:21] the kernel is already a monster right now [05:21] nope, not really ... [05:21] but you can copy over the fs/proc/virtual.c unmodified ... [05:22] that 'should' contain all changes ... [05:23] ok [05:30] this wont work [05:30] you changed vx_id->xid [05:31] well, this can be done with sed, right? [05:31] sure, will that be sufficient or do i need to pull other changes? [05:32] well, actually, I don't know .. wait until 1.3.6 is released or make a delta yourself ... vanilla 2.4.24 -1.3.5 to 1.3.5.2 [05:33] yea ive got a delta, ok im just gonna hand pach in rejs [05:43] if you have any troubles, just let me know ... [05:44] damn pax has its own handling of the pages [05:44] re Bertl [05:44] Bertl: i have some problems with vproc and 2.4.24/1.24 [05:44] hi rmoriz, another look at the irc? [05:44] yup :) [05:44] well, np, I tried to reproduce your issues, no luck, works here ... [05:45] hm [05:45] is there a kernel config option which could be the reason of it? [05:45] nope [05:46] try to strace the vproc -d /proc/meminfo (for example) [05:46] make that strace available somewhere via web ... [05:46] or send it to me ... [05:46] stat64("/proc/meminfo", {st_mode=S_IFREG|0444, st_size=0, ...}) = 0 [05:46] open("/proc/meminfo", O_RDONLY|O_NONBLOCK) = 3 [05:46] ioctl(3, 0x80047805, 0xbffffc90) = 0 [05:46] ioctl(3, 0x40047806, 0xbffffc94) = -1 EROFS (Read-only file system) [05:46] write(2, "./vproc: read only filesystem.\n", 31./vproc: read only filesystem. [05:46] ups ;) [05:47] hah [05:47] got it! [05:47] and shame on me ;> [05:48] well, it was ro? [05:48] i had still a vserver running with ro mounted /proc where the host /proc was rw [05:48] luckily, that is not possible ... [05:48] but I know what you mean ... [05:49] codernaut.org/procmount.txt [05:50] uhm [05:50] that was alread rw [05:50] damn pasting :) [05:50] well, there is only _one_ procfs, and it's either ro, or rw ... [05:51] you can't have both, because there is only one superblock, and the info is stored there ;) [05:51] "none on /vservers/ns1/proc type proc (ro) [05:51] " [05:51] i had 2 entries of this style [05:52] ok [05:52] if you do mount -o rwmount,ro /proc [05:52] if you do mount -o remount,ro /proc [05:52] then _all_ procfs instantly go ro [05:52] ah okay i got it :) [05:52] same with mount -o remount,rw /proc ... [05:54] do i have to adjust vproc after every mount? [05:56] no, just on the host, and only after the host reboot [05:56] so best done somewhere in rc.local ... [05:56] or you could define a service for that init.d/vproc [05:57] ok [06:02] hm [06:02] i can't hide /proc/mounts [06:03] or /proc/self/mounts [06:03] that is because this is per process, and dynamically generated ... [06:04] hm. it's because all host mounts are visible. [06:04] there are patches for that ... [06:04] and you remember the bunch of lvm mounts i have [06:04] really? you mean grsecurity or sth? [06:04] http://vserver.13thfloor.at/Experimental/no-proc-mounts.diff [06:04] oh :) [06:05] it's a little outdated, so if you want to use it, let me check it first ... [06:05] at least it applies without problems ;) [06:06] but it probably won't compile ;) [06:06] let's see - make-kpgk already started ;) [06:07] kpkg [06:07] lets see if this horrible monster boots [06:10] negative [06:10] sigh [06:10] namespace.c: In function `show_vfsmnt': [06:10] namespace.c:223: error: structure has no member named `s_context' [06:11] well, didn't I say ... ;) [06:11] wait a minute ... I'll update it ... [06:12] no hurry :) [06:25] http://vserver.13thfloor.at/Experimental/no-proc-mounts-vs1.24.diff [06:25] tested ... [06:27] ok fudge ck1 and such, i think im just going to try 1.24 and a little more vanilla until i get local boxes to test with [06:28] okay ... [06:29] hmm i need a patch to fix the HT scheduling though [06:30] I'm impressed how responsive the 2.4.25-pre6 is ... [06:30] hmm i havent booted it yet [06:33] gn8 everybody, sleep well [06:34] seya tomorrow [06:34] night! [06:37] *compiling [06:54] ok rebootet and tested: works fine. again thank you very much Bertl! :) [06:54] you are welcome ... have fun! [06:54] :) [06:54] first i'll sleep. it's 5am already :/ [06:54] n8 everyone [06:54] what does that patch do? hide proc entirely? [06:55] no, only /proc/mounts on vserver ... [06:55] ah [06:57] haha! vserver adds XFS (regarding IUNLINK) [06:57] i hate hand patching :( [06:58] 2.4.25-pre6+vs1.2.24+netconsole+irqbalance+grsec is what i am trying now [06:58] this box still isnt doing anything because everytime i try to put it up some kernel issue comes along :( [06:58] hmm, you know of the 2.4.25-pre6+vs1.2.24 patch, right? [06:58] yea i used that [06:59] its all done, im building it [06:59] okay, XFS isn't working there, but I guess it's no big deal, right? [06:59] hopefully it boots [06:59] na i dont use XFS [06:59] i just need a stable vserver+grsec box at this point [06:59] well, I did my best ;) [06:59] hehe i know [07:00] i appreciate your work :) [07:01] well, and I appreciate your help ... [07:02] guess vs1.24 and 1.3.6 wouldn't be where it stands, without nathan ... [07:02] i help where i can [07:03] btw did you say something about vserver-utils not being ideal? [07:03] sounded like there was another set of tools? [07:04] well, there is vserver-0.29 currently sporadically maintained by jack ... [07:04] and there is Enricos util-vserver-0.27 and the alpha versions 0.27.196 IIRC [07:05] which _is_ actively maintained and follows the kernel development ... [07:05] for example vserver doesn't support the rlimit syscalls ... util-vserver has vlimit ... [07:05] now these are just the userspace utils right? [07:05] jup [07:08] xfs_ioctl.c:901: `S_IMMUTABLE' undeclared (first use in this function) [07:08] hmm [07:08] I said XFS doesn' work ;) [07:08] < > XFS filesystem support [07:08] problem fixed :) [07:08] does nobody listen to me today? [07:08] i figured it would build, just not work [07:09] na i listened, i just thought it was unfinished code, not uncompilable code :) [07:09] well, it was in 1.3.5.2 ;) [07:09] hmm why didnt i use 1.3 [07:10] i guess i could have [07:10] ahwell, i just want this running [07:10] hmm, you remember the small window for races in vs1.23 I mentioned? [07:10] no :( [07:10] im just going to go jump off a bridge [07:11] well, I did some time ago .. I mean mention it to you ... ;) [07:11] yea, i recall something about it. so i guess it still persists? [07:11] well, I wanted to say, I didn't see it while I was looking at vs1.23 code again ... [07:12] ah another thing [07:12] so it might have been removed ... but I'm not sure ... [07:12] i can hide /proc/bus but still cd into it and see the entries [07:13] Bertl, do you remember what triggered it? [07:13] hehe http://www.petting-zoo.net/~deadbeef/archive/571.html [07:13] the Unix Alphabet is wonderful [07:13] rmoriz: well that is a bug then ... [07:13] G is for Grep, a clever detective, while [07:13] H is for Halt, which may seem defective. [07:14] riel: nice! [07:15] nathan_: I remember something about reassigning vx_infos ... [07:16] maybe it was on the migrate_task/user stuff somewhere ... [07:16] the only thing I remember for sure is, that I decided that fixing it would result in another 1.3.x ... [07:16] so I left that for 1.4 which will have that fixed ... [07:17] ok works now. hm probably because i was still in that directory on the host [07:17] hmm, yes, that could be ... [07:18] im considering running my 2.4.24+1.3.5 [07:18] ok finally good night ;) [07:18] the procs also share the dentries and inodes ;) [07:18] i stressed it for two days it and was fine [07:18] maybe i just got lucky [07:18] rmoriz: have a good night! [07:18] nathan_: well, I guess for 'normal' use, it should be pretty stable ... [07:19] are you defining normal as low load or what? [07:19] no as 'not 4 killers killing around' [07:19] ah [07:19] remember that our tests are doing 1000 context changes per minute? [07:20] the 'normal' vserver host will do about 10 per day, I guess ... [07:21] yea [07:27] hmm, any easy way to test the netconsole? [07:29] insert a bad module [07:29] just do something that will trigger a printk [07:29] okay, what do you use on the client side? I mean options? [07:30] ./netconsole-client -server serverip --cient myip -p 8888 [07:31] hmm, yeah, same here except for -client and -port ... [07:31] not getting anything? [07:32] no, messages in dmesg, but nothing on the wire ... [07:32] did you get the hardware address and ip address correct? [07:33] I'm currently trying with a crossover cable ... [07:33] so I should see any packets [07:34] what did netconsole say when you insert it [07:35] ah now we are getting somewhere ... [07:35] [...network console startup...] [07:35] there you go [07:35] by the way, I fixed the broken ip display, and I'll add the missing GPL definitions ... [07:36] ah nice [08:57] nox (~nox@b109150.adsl.hansenet.de) left irc: Ping timeout: 492 seconds [09:16] kungfuftr (~scott@jadis.narnia.org.uk) got netsplit. [09:45] loger joined #vserver. [10:20] noel- (~noel@pD9FFA206.dip.t-dialin.net) joined #vserver. [10:20] hi noel-! [10:28] noel (~noel@p50859C2B.dip.t-dialin.net) left irc: Ping timeout: 504 seconds [11:42] Nick change: Bertl -> Bertl_zZ [13:15] Simon (~sgarner@apollo.quattro.net.nz) joined #vserver. [13:31] Nick change: noel- -> noel [13:41] Simon (~sgarner@apollo.quattro.net.nz) left irc: Quit: so long, and thanks for all the fish [15:10] serving (~serving@213.186.190.24) left irc: Ping timeout: 492 seconds [15:35] ccooke (~ccooke@spc1-walt1-4-0-cust238.lond.broadband.ntl.com) left irc: Read error: Connection reset by peer [15:42] ccooke (~ccooke@spc1-walt1-4-0-cust238.lond.broadband.ntl.com) joined #vserver. [16:27] ccooke (~ccooke@spc1-walt1-4-0-cust238.lond.broadband.ntl.com) left irc: Read error: Connection reset by peer [17:02] serving (~serving@213.186.190.24) joined #vserver. [17:22] WSUatWORK (~Josh@ny.webpipe.net) left irc: Ping timeout: 480 seconds [18:14] Topic changed on #vserver by ChanServ!services@services.oftc.net: http://linux-vserver.org/ || latest stable 1.24, devel 1.3.5, exp 0.04 [19:42] Sh[a]de (shade@cpe109.bb101.cablesurf.de) left irc: Quit: Excursion (On IRC.BONGSTER.DE [#wwip, #german-elite and #lov]) [20:55] Nick change: Bertl_zZ -> Bertl [20:55] hi all! [21:04] hey bert [21:10] _shuri (~shushushu@3ffe:bc0:8000::5bb) joined #vserver. [21:10] hi nathan_! [21:10] http://www.13thfloor.at/patches/patch-2.4.24-nc0.01.diff.bz2 [21:11] (netconsole) [21:12] ah [21:12] it uses now the 'simpler' command line options also used in 2.6-netconsole ;) [21:12] and it supports about 30 nics or so ... [21:15] nice work [21:15] well, the buggy output and stupid hex interface annoyed me ... ;) [21:15] i dont mind, one time setup :) [21:16] jesus [21:16]