1128558018 N * Bertl_zZ Bertl 1128558035 M * Bertl hmm, morning! couldn't sleep ... 1128558070 J * douglas ~douglas@douglas.user.oftc.net 1128558073 M * douglas hello 1128558090 M * Bertl hey gndmstr! thanks for the good trace! 1128558105 M * Bertl hey douglas! 1128558149 M * douglas heres what I'm trying to do, I want to redirect all traffic from the vservers (a /27) for port 21 to one of the ips in that block 1128558183 M * Bertl what block :) 1128558225 M * douglas so lets say I have vs1-10 192.168.1.170-180 for example. I wanna forward 171.172.173 etc.etc to .170 for only port 21 1128558253 M * mnemoc iptables? 1128558256 M * Bertl from outside or from inside? 1128558269 M * douglas inside traffic going out. 1128558294 M * Bertl so when .171 tries to ssh out, you want it to reach .170, right? 1128558299 M * douglas correct 1128558310 M * mnemoc DNAT 1128558316 M * douglas 22 is ssh I believe, I think 21 is telnet but regardless you understand what I'm trying to do :) 1128558320 M * Bertl so you definitely need a prerouting rule, and it has to change the destination 1128558327 M * douglas right. 1128558329 M * daniel_hozac 23 is telnet ;) 1128558342 M * Bertl and 25 is mail *G* 1128558343 M * mnemoc 21 is ftp 1128558369 M * douglas ok ok nerds, it was just an example lets not get all high and mighty with our /etc/services knowledge :) 1128558382 M * douglas it was just an example. 1128558386 M * douglas heh 1128558435 M * douglas so can you give me an example cuz I'm having problems right the basic rules for that. and I've been working on it for awhile. 1128558607 M * daniel_hozac iptables -t nat -A PREROUTING -d 192.168.1.170/27 -p tcp --dport 21 -j DNAT --to 192.168.1.171 ? 1128558618 M * Bertl tx 1128558648 M * mnemoc -d or -s ? 1128558661 M * daniel_hozac ah, right. 1128558953 Q * jayeola Quit: Zzz 1128559737 M * gndmstr hehe no prob. Bertl. i just hope it helps .. and i was all excited last night... slept like a baby cause of it, but yeah its sitll way improved over previous versions :) 1128559882 M * douglas tx? 1128559924 M * douglas daniel, what if they were public ips? still use the "nat" option? 1128560006 M * Bertl tx = thanks ... 1128560090 M * Aiken depends who you ask, to me tx = transmitt 1128560133 M * daniel_hozac douglas: yes, you're still NATing. 1128560161 M * Bertl Aiken: well I assumed he asked me, and meant the tx to daniel :) 1128560177 M * Aiken :) 1128560272 M * douglas ahh 1128560276 M * douglas whats the diff between -s and -d? 1128560281 M * douglas cuz thats not working 1128560382 M * mnemoc -s source 1128560385 M * mnemoc -d destination 1128560414 M * mnemoc package coming from source, or going to destination 1128560972 M * Bertl gndmstr: you're still around? 1128561211 M * gndmstr yeah 1128561227 M * gndmstr was outside for a few min 1128561240 M * Bertl np, you have the kernel source of that new one? 1128561253 M * gndmstr yeah 1128561294 M * Bertl okay, and the qmail guest did cause the oops/trace? 1128561320 M * gndmstr yes 1128561343 M * gndmstr that was the 2nd one shut down the third is the one that normally shuts down ok but takes a long time 1128561401 M * Bertl hmm, okay, I'll give you a few numbers, and you run them through addr2line for me, yes? 1128561418 M * Bertl (addr2line -e vmlinux 1128561440 M * gndmstr ok 1128561533 M * Bertl c011ae6e c01a1d3b c0189f23 c01872aa 1128561544 M * gndmstr one run per number or all in a row 1128561556 M * Bertl as you like it :) 1128561597 M * gndmstr and here i was expecting long outputs :) 1128561603 M * gndmstr apollo linux # addr2line -e vmlinux c011ae6e c01a1d3b c0189f23 c01872aa 1128561603 M * gndmstr kernel/fork.c:123 1128561604 M * gndmstr fs/proc/inode.c:66 1128561604 M * gndmstr fs/inode.c:1158 1128561605 M * gndmstr fs/dcache.c:378 1128561605 M * gndmstr apollo linux # 1128561649 M * Bertl c011ad04 c011b21a c011b0e0 1128561682 M * gndmstr include/linux/vserver/debug.h:220 1128561682 M * gndmstr include/linux/vserver/debug.h:220 1128561682 M * gndmstr include/linux/vserver/debug.h:219 1128561739 M * Bertl okay, you're around for a while? 1128561751 M * gndmstr at least another 2 or 3 hrs yeah 1128561768 M * Bertl okay, great, will need other numbers too 1128561773 M * gndmstr not much more than 4 tho cause when i go to bed at 1 i only get about 4 hrs sleep 1128561777 M * gndmstr ok 1128561851 M * gndmstr heh boss told me today the serial scsi drives have been shipped.. told him to take his time :) 1128561869 M * daniel_hozac serial SCSI? 1128561869 M * Bertl heh :) 1128561908 M * gndmstr yeah for only a 4 proc p3-500 this thing is gonna be fast. he is putting in everything to push the limits of the internal busses 1128561944 M * gndmstr dont need lots of processor speed in this i dont think.. at least we dont on discreet servers 1128561972 M * daniel_hozac what is serial SCSI? i've never heard of that, and all my SCSI equipment has 50+ pin connectors... 1128561998 M * gndmstr a new interface. 15krpm drives.. he got some of the first ones being shipped 1128562038 M * gndmstr i forget the letterinig on them sas or something like that 1128562042 M * daniel_hozac link? 1128562134 M * gndmstr looking if i can find it.. he showed it to me some time back 1128562167 M * Bertl http://www.interfacebus.com/Design_Connector_Serial_SCSI_Bus.html 1128562197 M * Bertl gndmstr: which gcc are you using? 1128562232 M * gndmstr gcc (GCC) 3.3.6 (Gentoo 3.3.6, ssp-3.3.6-1.0, pie-8.7.8) 1128562292 M * daniel_hozac thanks. 1128562321 M * Bertl hmm, I always thought gcc 3.3.x should honor static inline ... 1128562462 M * Bertl gndmstr: could you upload the output of 'grep -r vxh_ System.map' to pastebin.com, please? 1128562597 M * gndmstr http://pastebin.com/384574 1128562601 M * Bertl tx 1128562970 M * gndmstr daniel_hozac: here is the link he sent me a month ago http://www.maxtor.com/portal/site/Maxtor/menuitem.f9b9cdc332b9a3a0cade4ce191346068/?channelpath=/en_us/Technologies/Serial%20Attached%20SCSI 1128563040 M * gndmstr and an older zdnet article http://news.zdnet.com/2100-9595_22-1019845.html?tag=nl 1128563090 M * gndmstr im quite impressed by their performance capabilities in a raid 5 or 50 array 1128563314 M * gndmstr im really glad he is thinking ahead on this machine. he also bought 2 8-port gigabit nics for it. we will be using only 4 of them for now but it will be super easy to bring up another previously unused network segment this way 1128563347 M * Bertl hmm, 8 port GBit? what bus do you have there? 1128563425 M * gndmstr i have no idea what the machine has, its an older one but it should support them ok i would think... im not worried about gigabit performance im worried about the number of nic ports that can be used.. i believe its a standard pci bus 1128563461 M * gndmstr i think its a 2001 machine 1128563491 M * Bertl ok 1128563525 M * gndmstr i told him to try to put 2 or 3 4 port cards in it like the one he has in it now.. then he went out and got these 1128563610 M * gndmstr my thought was 100mbit would have been sufficient but he is adding an additional gb series of network legs so he wanted to plug this machine into that 1128563658 M * gndmstr all our internet connections are fast_eth so we have the bandwidth for it 1128563747 M * gndmstr says he just ordered an ATM leg.. hehe i would LOVE to plug my machine into that for just one day! 1128563778 M * gndmstr but being 1000 miles away from the noc, im stuck with a t1 1128563831 M * gndmstr which is more trhan i asked for so i cant complain.. when i first started with him i told him a 384k would be sufficient to do what i needed but he got me a t1 so.. :) 1128563874 M * Bertl okay, let's do the following now: 1128563878 M * gndmstr k 1128563902 M * Bertl make ./kernel/vserver/context.s 1128563913 M * Bertl grep -B 5 vxh_ kernel/vserver/context.s 1128563920 M * Bertl please upload the output again ... 1128563946 M * daniel_hozac gndmstr: so 16 gigabit NICs with a standard PCI bus? that's not going to be fun. 1128563972 M * Bertl well, if he uses 100Mbit on that .. no problem ... 1128563984 M * gndmstr i know. i doubt they will all be made to work, but if we can use at least 6-8 of them ill be happy 1128564005 M * daniel_hozac still, 16 100 Mbit connections is a lot for standard PCI. 1128564009 M * gndmstr doesnt matter if it tops out at 100mbit each.. if it can go higher even in bursts is what he is looking for 1128564057 M * eyck 16 x 100 MByte? 1128564062 M * daniel_hozac isn't PCI just 133 MBps? 1128564079 M * Bertl eyck: Mbit 1128564087 M * gndmstr http://www.forestoflives.com/greptest.txt 1128564102 M * eyck Bertl: he said gigabit? 1128564129 M * gndmstr i told him he wasted his money on gigabit for it that i felt with 4 or 6 nics being used that 100mbit would be absolute max 1128564151 M * gndmstr but he thinks it can support it so... hes the boss and he does all the hardware configs so... 1128564178 M * gndmstr he builds, i just make it happen 1128564187 M * gndmstr if i cant i blame his hardware :D 1128564312 M * gndmstr i already have an excuse set up if the vserver setups wind up a bit flaky yet... his hardware is too old for today's software technology :) 1128564441 M * gndmstr then with my luck he will go out and get a sun enterprise 1000 e10k just to spite me 1128564458 M * gndmstr 44 processors, 44gb ram 1128564473 M * gndmstr uses the sun container system 1128564501 M * Bertl *damn* seems I have to reengineer the history tracing ... 1128564520 M * gndmstr better to find it now than later on 1128564721 M * gndmstr hehe thats my dream for a workstation and entire network that can do anything 1128564725 M * gndmstr http://cgi.ebay.co.uk/Sun-Enterpise-10000-E10k-Fully-Specd-44-procs-44gb_W0QQitemZ5815415866QQcategoryZ1486QQrdZ1QQcmdZViewItem 1128564838 M * gndmstr heh once i start the machine, as long as i dont try to stop it, everything runs flawlessly :) 1128565349 M * Bertl I can easily fix the shutdown issues, but I'd like to understand _why_ they happen first ... 1128565458 M * gndmstr ok 1128565482 M * gndmstr i have about a week or so to be ready for the big one to start putting its primary host os on it 1128566648 Q * douglas Ping timeout: 480 seconds 1128566915 Q * mef Quit: using sirc version 2.211+KSIRC/1.3.10 1128567270 M * Bertl gndmstr: okay, I (hopefully?) fixed up the history tracing, now the question is, can we easily reproduce the issue/trace? 1128567289 A * Bertl is test compiling the kernel now ... 1128567324 M * gndmstr i dont know. it seems the symptoms are slightly different almost every time but i can try by shutting them down in the same order i did before 1128567336 M * Bertl Aiken: you have a gcc 4.x toolchain lying around? 1128567360 M * gndmstr they ahve had a good 15 hrs or so to dirty up the cache etc for each so ... 1128567394 M * Bertl I don't think that is really relevant, it looks like the triggers are somewhere in: 1128567406 M * Bertl - dentry cache flush (we can trigger that) 1128567421 M * Bertl - proc mount/remount of the single superblock fs 1128567435 M * Bertl - context shutdown (your qmail guest) 1128567460 M * gndmstr are there certain context numbers that may cause problems? 1128567469 M * Bertl no, definitely not ... 1128567472 M * gndmstr ok 1128567502 M * gndmstr the context shutdown and definately the proc mount/remount i see the most of 1128567506 M * Bertl also the debug code has no influence on this issue, and we fixed the other one :) 1128567516 M * gndmstr cool 1128567558 M * Bertl overhead of the history code is detectable but small, the 'normal' debug code is not really detectable 1128567578 M * Bertl (regarding your email questions :) 1128567656 M * gndmstr ahh ok 1128567659 M * gndmstr interesting 1128567686 M * gndmstr back when i was doing dos c programming for a company, debug code created lots of overhead 1128567743 M * Bertl well, I do some tricks to keep it low ... 1128567759 M * gndmstr :) 1128567761 J * Sonarman_ ~cleetus@adsl-67-113-235-92.dsl.snfc21.pacbell.net 1128567797 M * gndmstr one of these days ill have the time to get back into it.. programming under linux for linux is a whole different ballgame and it looks fascinating 1128567817 M * Aiken Bertl I have gcc 4.0.1 installed on the athlon 1128567845 M * Aiken a std 4.0.1 1128567852 M * Bertl Aiken: excellent could you compile test the following patch (ontop of 2.6.13.3-vs2.1.0-rc3 if gcc complains anywhere? 1128567866 M * Bertl gndmstr, Aiken: http://vserver.13thfloor.at/Experimental/delta-2.6.13.3-vs2.1.0-rc3-rc3.1.diff 1128567887 M * Bertl gndmstr: here is an idea of that magic ... 1128567920 M * gndmstr apply this on top of the existing patched code right 1128567925 M * Bertl gndmstr: for you too, ontop of the current kernel, best apply it and call 'make' no need to clean or so 1128567928 M * Aiken ok 1128567934 M * gndmstr ok 1128567937 M * Bertl Aiken: TIA 1128567969 Q * Sonarman Ping timeout: 480 seconds 1128568805 J * dddd44 dhb55@60.49.78.240 1128568870 M * Aiken still compiling 1128568961 M * gndmstr done. how do you want me to shut the existing setup down? just issue a reboot at the host or try to shut each guest down manually first? 1128569117 M * Bertl doesn't matter we need the new kernel first ... 1128569134 M * Bertl Aiken: any suspicious warnings yet? 1128569208 M * Aiken there some early in the compile to do with a str function or 2 1128569220 M * Aiken apart from that a few depreciated warnings 1128569239 M * Aiken using script to record the compile, I'll have a proper look when it has finished 1128569246 M * Bertl okay, excellent! 1128569302 M * gndmstr i just noticed something. it tries to shut vservers down AFTER shutting down the ethernet cards. shouldnt it shut them down first? 1128569337 M * gndmstr some of the applications go nuts when they lose net contact and you try to shut them down 1128569375 M * Bertl hmm, IIRC you can fine tune this via the sysv priorities 1128569386 M * Bertl (but maybe the defaults are not perfect) 1128569420 M * gndmstr ok will have to read up about them. i would be more comfortable if they shut down before ethernet does 1128569475 M * Bertl ls -la /etc/rc*.d/*vservers* 1128569485 M * Bertl (should show you the script levels) 1128569495 M * Aiken a few warnings like drivers/serial/8250.c:2667: warning: `unregister_serial' is deprecated (declared at drivers/serial/8250.c:2664 1128569495 M * Aiken ) 1128569499 M * gndmstr the guests appeared to shut down ok, it blew up again at remounting read only before boot 1128569503 M * Bertl S* is startup, K* is shutdown 1128569510 M * gndmstr this is gentoo 1128569513 M * gndmstr it doesnt use that 1128569519 M * Bertl ah, right ... 1128569522 M * Aiken fs/reiserfs/xattr.c:1410: warning: label `check_groups' defined but not used 1128569531 M * Aiken a few of those kinds of warnings 1128569538 M * Bertl Aiken: okay, anything vxh related? 1128569553 M * Aiken no 1128569564 M * Bertl good ... 1128569584 M * Aiken dam 1128569596 M * Aiken when I started script it used the wrong gcc 1128569597 M * Bertl hmm? 1128569618 M * Bertl ah, well ... :) 1128569728 M * Aiken redoing 1128569729 M * gndmstr ok all up and running properly on 2.6.13.3-vs2.1.0-rc3.1 shoud i just issue a shutdown now? 1128569751 M * Bertl no, let's fill the dentry cache first 1128569767 M * gndmstr ok let it run a bit then 1128569791 M * gndmstr or should it run till tomorrow am 1128569804 M * Bertl 'find / -type f -printf "%p" >/dev/null' 1128569811 M * gndmstr i can still reboot safely without complaints until about 7am 1128569850 M * Bertl check with cat /proc/meminfo while this is going, if the cache doesn't grow anymore, it's fine 1128569953 M * gndmstr it will grow for a few hrs. typically it starts out using about 140mb ram and jumps over time to about 480mb where it seems to settle 1128569992 M * Bertl use the find command, it will grow much faster :) 1128570052 M * gndmstr this should do it then find / -iname '*.txt' 1128570118 M * Bertl well, I'd prefer my line, but that will do something too :) 1128570172 M * gndmstr ok will use yours.. didnt notice it 1128570219 M * gndmstr bash: find / -type f -printf "%p" >/dev/null: No such file or directory 1128570278 M * gndmstr doesnt it need a mask specifier between the last " and > ? like '*' ? 1128570305 M * Bertl I'd say it found everything which was there to find 1128570312 M * Bertl but you can use 1128570320 M * Bertl find / -xdev -type f -printf "%p" >/dev/null 1128570337 M * Bertl (which will not walk over procfs) 1128570363 M * gndmstr ok its doing something 1128570396 Q * meebey Ping timeout: 480 seconds 1128570456 M * gndmstr while its running meminfo cache is hovering just above and just below 60700 1128570479 M * gndmstr never mind still growing 1128570514 J * sebi ~sebi@Fd3e0.f.strato-dslnet.de 1128570521 M * Bertl welcome sebi! 1128570599 J * meebey meebey@booster.qnetp.net 1128570618 Q * sebi_ Ping timeout: 480 seconds 1128570655 M * gndmstr seems to be sticking close to this now and the find routine is done 1128570658 M * gndmstr em: 1034932 860800 174132 0 327856 60832 1128570670 M * gndmstr Cached: 60828 kB 1128570674 M * Bertl okay, let's try with a guest shutdown (your favorite) 1128570687 M * gndmstr ooook 1128570749 M * gndmstr timeout 1128570762 M * Bertl trace? 1128570775 M * gndmstr all the applications inside stopped normally then nothing left 1128570776 M * gndmstr no 1128570790 M * Bertl okay, let's start and stop it a few times 1128570827 M * gndmstr it left a pid in /var/run/vservers.. leave it or delete it 1128570851 M * Bertl what did you before? 1128570857 M * gndmstr delete 1128570864 M * Bertl okay, then do that again 1128570864 M * gndmstr most of the time it would say it was still running 1128570898 M * gndmstr ok deleted it and did the start command and its just sitting there 1128570905 M * gndmstr got a trace 1128570918 M * Bertl excellent! 1128571031 M * gndmstr http://pastebin.com/384656 1128571042 M * gndmstr im gonna power cycle it to get the radius back up and runing 1128571052 M * Bertl okay 1128571117 M * Aiken chown is failing on a file that has had setattr --iunlink applied 1128571119 M * gndmstr heh i like the way they put this line 1128571122 M * gndmstr Badness in send_IPI_mask_bitmask at arch/i386/kernel/smp.c:168 1128571176 M * Aiken the compile finished with http://pastebin.com/384658 1128571215 M * Bertl hum hum ... 1128571242 M * Bertl luckily that is not my code :) 1128571257 M * Aiken a 2nd run of make and it is continuing 1128571346 M * Aiken no errors or warnings to do with vxh 1128571386 M * Bertl excellent! thanks a lot! 1128571413 M * Aiken how do I undo a setattr --iunlink? 1128571420 M * Bertl gndmstr: 3405 is the qmail context? 1128571425 M * Aiken managed to do that to a directory 1128571432 M * Bertl --~iunlink ? 1128571456 M * Aiken cool, could not remember where to put the ~ 1128571497 M * mnemoc the '~' is odd 1128571511 M * gndmstr yes 1128571513 M * Bertl definitely, but it was a good choice ... 1128571535 M * mnemoc if you say so 1128571558 M * Aiken Bertl http://pastebin.com/384662 1128571566 M * Bertl mnemoc: well, alternatives? 1128571625 M * Bertl Aiken: hmm, sounds reasonable .. the immutable part applies here 1128571662 M * mnemoc Bertl: don't know :p 1128571673 M * Aiken at the time I was trying to change ownship of some files in my master image will all guests rm -rf 1128571685 M * mnemoc {-,+}? 1128571717 M * Bertl ---iunlink? --+iunlink? 1128571731 M * mnemoc .oO 1128571738 M * Bertl long options don't allow for +,- at the beginning 1128571748 M * mnemoc getopt :\ 1128571812 M * mnemoc --iunlink=.. ? 1128571849 M * Bertl yes, that would have been an alternative ... 1128571893 M * mnemoc at least is more intuitive 1128572063 M * Bertl feel free to modify the sources *harhar* 1128572121 M * Bertl Aiken: okay, could you change a single line for me and compile just with 'make ./kernel/vserver/context.s' ? 1128572136 M * Bertl gndmstr: okay, let's see if it was worth the efford ... 1128572148 M * mnemoc modify the sources is easy than telling all people to use a new sintax ;) 1128572163 M * gndmstr k 1128572210 M * Bertl addr2line of c011ad04 c013f0e3 c0138a2e 1128572257 M * Bertl Aiken: in include/linux/vserver/debug.h the definition of VXH_HERE to: 1128572271 M * Bertl #define VXH_HERE \ ({ __label__ __vxh_label; __vxh_label:; &&__vxh_label; }) 1128572298 M * Bertl hmm ... actually there is a linebreak after the '\' 1128572317 M * gndmstr must be in the wrong directory. im in the kernel 'root' addr2line: 'a.out': No such file 1128572333 M * Bertl addr2line -e vmlinux c011ad04 c013f0e3 c0138a2e 1128572344 M * gndmstr oops 1128572345 M * gndmstr sorry 1128572348 M * Bertl np 1128572363 M * gndmstr include/linux/vserver/debug.h:191 1128572364 M * gndmstr include/linux/vserver/debug.h:178 1128572364 M * gndmstr include/linux/vserver/debug.h:178 1128572372 M * Aiken yes 1128572378 M * Bertl gndmstr: *damn* 1128572415 M * gndmstr didnt do it huh 1128572420 M * Bertl gndmstr: you sure you compiled/booted the 'new' kernel? 1128572429 M * gndmstr lemme double check 1128572445 M * Aiken Berlt you want the \ and a line break inserted? 1128572456 M * Aiken Bertl even 1128572459 M * Bertl doesn't matter, you can write it in one line too 1128572474 M * Bertl #define VXH_HERE ({ __label__ __vxh_label; __vxh_label:; &&__vxh_label; }) 1128572481 M * gndmstr the kernel has a date stamp of 23:22 which is about right 1128572505 M * Bertl okay, was the patch applied? 1128572522 M * gndmstr yes 1128572548 M * gndmstr if i prove to myself by applying again it wont hurt anything right? it will just error 1128572560 M * Bertl try with --dry-run 1128572594 Q * serving- Ping timeout: 480 seconds 1128572649 M * Aiken just changing that file has triggered a big recompile 1128572655 M * gndmstr 1 out of 1 hunk FAILED -- saving rejects to file Makefile.rej 1128572655 M * gndmstr patching file include/linux/vserver/debug.h 1128572689 M * Bertl hmm, the debug succeeds? 1128572713 M * gndmstr Hunk #1 FAILED at 142. 1128572713 M * gndmstr Hunk #2 FAILED at 155. 1128572714 M * gndmstr 2 out of 2 hunks FAILED -- saving rejects to file include/linux/vserver/debug.h.rej 1128572721 M * Bertl hmm ... 1128572848 Q * meebey Ping timeout: 480 seconds 1128572854 M * Bertl gndmstr: okay, let's try to substract two from the address 1128572858 M * Bertl so: 1128572872 M * Bertl addr2line -e vmlinux c011ad02 c013f0e1 c0138a2c 1128572888 M * gndmstr include/linux/vs_context.h:99 1128572888 M * gndmstr include/linux/vs_context.h:37 1128572889 M * gndmstr include/linux/vs_context.h:17 1128572928 M * Bertl looks a lot better :) 1128572935 M * gndmstr would it help to have access to the machine? 1128572943 M * Aiken rebuilt fine 1128572949 M * Bertl gndmstr: no, not really 1128572952 M * gndmstr ok 1128573003 M * Bertl Aiken: okay, thanks! 1128573035 M * Bertl gndmstr: you can count in hex? 1128573050 J * meebey meebey@booster.qnetp.net 1128573053 M * gndmstr used to be able to 1128573062 M * Bertl backwards too? :) 1128573072 M * gndmstr 12345678abcdef 1128573087 M * Bertl okay, let's decrement the addresses until we leave vs_context.h 1128573149 M * gndmstr by how much 1128573153 M * gndmstr by 2? 1128573159 M * Bertl yeah, should be fine 1128573172 M * Bertl even 4 might be fine 1128573634 M * gndmstr this is my third one and its still staying on the same context line numbers 1128573662 M * gndmstr first time i decremented by 2 then the last 2 by 4 1128573677 M * gndmstr doesnt seem normal or is it 1128573679 M * Bertl try to increase the step size, but lower it once you're outside 1128573687 M * gndmstr ok ill try 8 1128573773 M * gndmstr first dec by 8 put the first line out into system.h the other 2 are remaining 1128573783 M * gndmstr shall i go up to see where it leaves? 1128573788 M * Bertl yes, please 1128574458 M * gndmstr been dropping by 8 for a while now to catch the last one and it hasnt budged off condext.h:17 1128574487 M * gndmstr the other 2 have gone thru a few files 1128574502 M * gndmstr will keep dropping 1128574513 M * Bertl hmm, I'm working on a better solution ... 1128574596 M * gndmstr finally got it to budge to 14 1128574699 M * Bertl okay, you know the first files/lines outside the .h for each one? 1128574731 M * gndmstr yean 1128574759 M * Bertl ok, let's hear ... 1128574802 M * gndmstr this is the last count they remained in context.h 1128574803 M * gndmstr apollo linux # addr2line -e vmlinux c011acf3 c013f0d8 c0138a14 1128574803 M * gndmstr include/linux/vs_context.h:96 1128574804 M * gndmstr include/linux/vs_context.h:34 1128574804 M * gndmstr include/linux/vs_context.h:17 1128574815 M * gndmstr next 2 moved out 1128574818 M * gndmstr apollo linux # addr2line -e vmlinux c011acf2 c013f0d7 c0138a13 1128574818 M * gndmstr include/asm/system.h:227 1128574818 M * gndmstr kernel/vserver/limit_proc.h:29 1128574819 M * gndmstr include/linux/vs_context.h:17 1128574859 M * gndmstr this is the first one that line 17 went to 14 1128574861 M * gndmstr apollo linux # addr2line -e vmlinux c011acd2 c013f087 c01389f3 1128574861 M * gndmstr kernel/fork.c:105 1128574861 M * gndmstr kernel/vserver/limit_proc.h:29 1128574862 M * gndmstr include/linux/vs_context.h:14 1128575183 M * Bertl ah, that makes sense ... 1128575269 M * gndmstr cool 1128575979 M * gndmstr brb 10 min gotta walk dog 1128575984 M * Bertl k, np 1128576800 M * gndmstr back 1128576809 M * Bertl okay, have another patch ... 1128576833 M * gndmstr k 1128576852 M * Bertl http://vserver.13thfloor.at/Experimental/delta-2.6.13.3-vs2.1.0-rc3.1-rc3.2.diff 1128576869 M * Bertl still no solution, but should give more insight 1128576892 M * gndmstr ok apply this over the other patched code then 1128576895 M * Bertl gndmstr: can we do the following (on next boot)?: 1128576910 M * Bertl - leave the qmail guest unstarted 1128576927 M * Bertl - enable some specific debugging options (I'll tell ya) 1128576935 M * Bertl - then start/stop the qmail guest? 1128576947 M * gndmstr ok fixing guest so it doesnt autostart 1128576947 Q * litage Read error: Connection reset by peer 1128576949 M * Bertl (patch is ontop of your current version) 1128576959 M * gndmstr actually ill turn all autostart off and just manually start them 1128576976 M * Bertl okay .. should work 1128576983 J * litage ~nick@203.220.55.70 1128577018 M * gndmstr will ahve to manually shut them down to reboot.. ok just compile this as is or do i change options in the config first 1128577036 M * Bertl no change to kernel options 1128577042 M * gndmstr k 1128577054 M * Bertl but we will echo some stuff to /proc before the qmail startup 1128577062 M * gndmstr ok 1128577082 M * gndmstr its ok to start the other 2 tho? 1128577091 M * Bertl yep, that's fine 1128577101 M * gndmstr cant leave the radius down long, its auth for all our national cust 1128577117 M * Bertl np, just start it 1128577118 M * gndmstr adn the backup i run here and its down for hardware upgrades 1128577121 M * gndmstr ok 1128577514 M * gndmstr times like this i wish i did have that 44 processor machine :) its doing net/ipv4 now 1128577579 M * eyck why not 128 CPU? 1128577600 M * Bertl too much inter-cpu overhead :P 1128577608 M * gndmstr i should have set up distcc between the gentoo machines.. would have given it 6 processors to work with 1128577635 M * gndmstr that one im talking about is configured for sun containers so you can specify a virtual machine with 2 or 4 processors 1128577669 M * gndmstr hehe eyck... 128 is a bit much 1128577684 M * gndmstr 44 is a bit much too but the machine just impressed me 1128577825 M * eyck Bertl: exactly, that's why it would require some nice programming techniques, not just dumb java-style 1024 threads all doing nothing, 1128577830 M * gndmstr whats a good practical cpu limit before overhead gets to be worrisome... 8? 1128577856 M * eyck gndmstr: depends on application, some say '1' is the limit;) 1128577872 M * eyck while others build 512+ cpu machines and still not reach any limit.. 1128577895 M * gndmstr i mean as a standard server 1128577904 M * eyck define 'stantard server' ;) 1128577918 M * eyck IIS + MS SQL 2000? 1128577918 M * gndmstr internet server... web, email, dns that kind of stuff 1128577930 M * gndmstr mysql or postgresql back ends 1128577946 M * eyck for mysql, I'd say '1' is the limit. 1128577958 M * gndmstr we dont allow windoze machines on our netoworks at all. 1128577962 M * gndmstr hehe 1128577964 M * gndmstr that good huh 1128578006 M * gndmstr we have a special network segment for windows customer owned servers. they never connect to ours at all 1128578012 M * eyck dns would scale well, web scales very well, SQL needs smart programmers 1128578014 M * gndmstr only via outside routing 1128578026 M * eyck well 1128578042 M * eyck your definition of 'never connect to ours AT ALL' is quaint... 1128578043 M * gndmstr we only use the sql databases for radius and in house customer database 1128578046 M * Bertl gndmstr: okay, let me know when you are ready (rebooted) 1128578052 M * gndmstr i mean direct line 1128578061 M * gndmstr there is no in house network connection 1128578065 M * eyck direct line == line of sight? 1128578094 M * eyck good for you, 1128578128 M * gndmstr they have to route out to the internet and back in to get to our servers 1128578155 M * gndmstr first 2 guests stopped perfectly 1128578163 M * gndmstr qmail one is gonna give me trouble :) 1128578175 M * gndmstr yep timed out 1128578241 M * gndmstr im wondering if i should reinstall the baselayout just in case there is something odd in that one 1128578264 M * Bertl no no, that's fine 1128578274 M * Bertl don't change anything right now 1128578306 M * gndmstr no i was talking about later.. no matter anyway... these are all being rebuilt when they go into the big one. not moving any of them directly 1128578435 M * gndmstr ok its rebooted. should i start the good 2 or do what you want first 1128578457 M * Bertl just start them, let them settle a few seconds 1128578457 Q * litage Read error: Connection reset by peer 1128578469 M * gndmstr leave qmail off right 1128578471 M * Bertl then _before_ you start the qmail, do: 1128578490 M * gndmstr ok they are starting 1128578533 M * Bertl echo 65535 >/proc/sys/vserver/debug_xid 1128578538 M * Bertl echo 65535 >/proc/sys/vserver/debug_switch 1128578582 M * Bertl prepare for some logging in klog/syslog 1128578583 M * gndmstr done 1128578598 M * gndmstr restart qmail guest? 1128578600 M * Bertl okay, then let's start the qmail guest 1128578605 M * gndmstr ok 1128578618 M * Bertl once it's up, repeat the find from last time ... 1128578641 M * Bertl after the find, wait about 30 seconds, then shutdown the guest 1128578809 J * litage ~nick@203.220.55.70 1128579072 M * gndmstr while find is still running.... is the fact that the qmail guest displays all its init to the screen when it never used to mean there is something corrupted in it? 1128579092 M * Bertl no, just means different init style 1128579094 M * gndmstr looks just like a console boot when it starts 1128579106 M * gndmstr wasnt like that before one of the crashes 1128579226 M * Bertl I doubt that is related ... 1128579242 M * gndmstr ok 1128579243 M * Bertl I mean, the init style might be related with the oops/trace 1128579251 M * gndmstr stopping the vserver 1128579257 M * gndmstr the style file still says plain 1128579297 M * gndmstr stopped fine 1128579306 M * gndmstr no errors at all 1128579328 M * Bertl okay, let's restart it and try again ... 1128579334 M * gndmstr ok 1128579349 M * gndmstr starting fine 1128579353 M * gndmstr run find again? 1128579360 M * Bertl can't hurt 1128579416 J * serving serving@86.108.20.30 1128579485 M * gndmstr ok stopping again 1128579508 M * gndmstr its quite busy too 1128579512 M * gndmstr stopped ok 1128579527 M * Bertl maybe the logging is too much ... 1128579538 M * Bertl probably slows down the shutdown 1128579557 M * gndmstr wouldnt doubt it although it does use metalog which caches and writes in bursts 1128579578 M * Bertl but I guess I found and fixed the issue already ... will provide another patch shortly 1128579600 M * gndmstr should i try a reboot letting the init shut all the vservers down 1128579613 M * Bertl no just disable the logging by doing: 1128579623 M * Bertl echo 0 >/proc/sys/vserver/debug_xid 1128579631 M * Bertl echo 0 >/proc/sys/vserver/debug_switch 1128579642 M * Bertl and start/stop the qmail a few times 1128579644 M * gndmstr do that now right 1128579645 M * gndmstr ok 1128579656 M * Bertl I'm pretty confident it will oops again 1128579705 M * gndmstr the qmail vserver killed itself when i did the proc changes 1128579729 M * gndmstr pid file is gone too 1128579735 M * gndmstr start and stop it anyway? 1128579739 M * Bertl yes 1128579923 M * gndmstr without the find it started and stopped perfectly every time.. now its running and im running the find again 1128579953 M * gndmstr find is taking a very short time to run. must be cached 1128579961 M * Bertl yes it is ... 1128579968 M * gndmstr 154504 1128579970 M * gndmstr cache value 1128579973 M * gndmstr kinda 1128579975 M * gndmstr :) 1128579976 M * Bertl you can try with mount -o remount,rw / 1128579988 M * Bertl that should flush it 1128579994 M * gndmstr now while everything is alive? ok 1128580002 M * Bertl doesn't hurt ... 1128580016 M * gndmstr cache is still the same 1128580079 M * gndmstr but find is working now 1128580089 M * gndmstr just finished 1128580138 M * gndmstr stopped and started fine 1128580154 M * gndmstr should i enable autoshutdown and reboot the host 1128580255 M * Bertl maybe ... 1128580288 M * gndmstr ok will try it 1128580307 M * Bertl hmm, wait 1128580327 M * gndmstr oops 1128580329 M * gndmstr too late 1128580333 M * Bertl np 1128580341 M * Bertl http://vserver.13thfloor.at/Experimental/delta-2.6.13.3-vs2.1.0-rc3.2-rc3.3.diff 1128580356 M * gndmstr rebooting fine 1128580361 M * gndmstr shut down everything fine 1128580367 M * Bertl this is the patch, which I think will eliminate the issue for you 1128580380 M * gndmstr ok 1128580400 M * Bertl I'd suggest to compile a kernel with that, leave the history stuff and debug enabled 1128580408 M * Bertl (but not the echo stuff) 1128580408 M * gndmstr ok 1128580420 M * Bertl enable the qmail auto startup as you had it before 1128580433 M * gndmstr ok 1128580439 M * Bertl reboot the machine, and let it run .. today ... 1128580459 M * gndmstr ok so dont reboot it at 7am butr wait till later at night 1128580460 M * Bertl whenever you feel like, stop and start the qmail guest :) 1128580465 M * gndmstr ok 1128580490 M * Bertl I'm pretty confident the patch fixes the issue you observed ... 1128580501 M * gndmstr thankfully its a secondary mx so keeping it up is really a non-issue for a bit. only 2 critical ones are the name server and radius 1128580541 M * gndmstr ok will find out over the next few days... will restart tomorrow night letting the host handle shutdowns, and then if its good will let it run till maybe saturday and reboot it again 1128580552 M * Bertl okay, thanks a lot for the testing so far and for time ... 1128580558 M * gndmstr no prob 1128580562 M * Bertl (and for the excellent traces) 1128580563 M * gndmstr i wanna see this work perfectly 1128580589 M * Bertl when I get up again (now heading to bed) I'll add you to our hall'o'fame ... 1128580630 M * gndmstr hehe dont have to 1128580651 M * Bertl well, seems you found an important bug ... 1128580705 M * gndmstr cool as long as it improves the vserver system... and i suspect it will cause it will stop giving some people headaches like i had :) 1128580746 M * Bertl very likely, I'm somewhat puzzled that this case didn't show up more often yet ... 1128580751 M * gndmstr i can imagine im not the only one with this prob.. the hardware it runs on is too usual. 1128580784 M * Bertl but even in my quad CPU tests, it didn't strike, so it seems it requires certain conditions before it happens 1128580805 M * gndmstr the only unusual, but not really so, thing is we are running all partitions as raid mirrors 1128580862 M * Bertl I guess a special kind of dentry cache pressure and very fast? guest shutdown (with no references) is required to trigger it ... 1128580887 M * Bertl so that unhash_process() is called and the pid is flushed 1128580916 M * Bertl maybe the qmail does something special in this regard (or some other task in that context) 1128580927 M * gndmstr could be.. im not sure what was changed in the 1.11.12 baselayout 1128580969 M * gndmstr seems there is an update i see 1.11.13 is now avail 1128580982 M * gndmstr hmm 1128580985 M * gndmstr hold on a sec 1128580994 M * gndmstr i thought i put 1.11.20 in there 1128581133 M * gndmstr wrong versioning.. the others have 1.12.0-pre8-r2 1128581154 M * gndmstr this does too 1128581291 M * gndmstr tried to stop vserer ns and then i got basename: too many arguments 1128581291 M * gndmstr Try `basename --help' for more information. 1128581305 M * gndmstr however it rebooted fine 1128581319 M * gndmstr going into the new kernel now 1128581368 M * Bertl after complete guest startup, please do the following .. 1128581373 M * Bertl cat /proc/virtual/status 1128581388 M * gndmstr ok 1128581398 M * Bertl and after each qmail start/stop again ... (note the values there) 1128581421 M * gndmstr ok 1128581454 M * Bertl okay, enough for me .. I'm off to bed now ... if something terrible happens (with the new kernel) go back the the previous one ... 1128581470 M * gndmstr ok 1128581478 M * Bertl (don't think so, but you never know ...) 1128581486 M * gndmstr #CTotal: 3 1128581486 M * gndmstr #CActive: 3 1128581489 M * gndmstr looks good 1128581495 M * Bertl night everyone! thanks gndmstr! Aiken! 1128581501 N * Bertl Bertl_zZ 1128581504 M * gndmstr night 1128581647 Q * gndmstr Remote host closed the connection 1128583065 M * Aiken good night 1128585443 J * Shoragan ~shoragan@d072.apm.etc.tu-bs.de 1128585668 Q * Aiken Quit: Leaving 1128585714 Q * dddd44 Ping timeout: 480 seconds 1128586609 J * sven12 ~sven@80-235-89-98-dsl.prn.estpak.ee 1128587101 J * Aiken ~james@tooax6-100.dialup.optusnet.com.au 1128587102 Q * litage Read error: Connection reset by peer 1128587117 J * dddd44 dhb55@60.49.78.240 1128587143 J * litage ~nick@203.220.55.70 1128587657 Q * virtuoso Ping timeout: 480 seconds 1128588235 N * nokoya nokoya- 1128588236 Q * Aiken Read error: Connection reset by peer 1128588539 J * prae ~prae@ezoffice.mandriva.com 1128588729 Q * dddd44 Ping timeout: 480 seconds 1128589149 J * virtuoso ~s0t0na@shisha.spb.ru 1128589510 J * dddd44 ~dhb55@60.49.78.240 1128589884 J * eXa_bOy ~a@c220-239-68-228.rochd3.qld.optusnet.com.au 1128589906 M * eXa_bOy anyone active 1128589908 M * eXa_bOy ? 1128589935 M * eXa_bOy well here is my problem. im following this tutorial --> http://linux-vserver.org/Step-by-Step+Guide+2.6 1128589936 M * Johnsie Just for a few, then I'm off to bed. 1128589967 M * eXa_bOy the problem im having is when it tells me to "su -" 1128589976 M * eXa_bOy do i need to do this if im already root? 1128590015 M * Johnsie No. 1128590020 M * eXa_bOy ok thanks 1128590023 M * Johnsie You just have to be root to install the kernel, etc. 1128590037 M * eXa_bOy then why did it say that? :S 1128590038 M * Johnsie Might I ask what platform you're doing this on? 1128590042 M * eXa_bOy debian 1128590051 M * Johnsie Well, in case you were doing it from userland. :) 1128590080 M * eXa_bOy http://pastebin.com/384783 1128590082 M * Johnsie And, for example, if you're converting a system on-the-fly, you may very well be logged in as a user to do so. 1128590096 M * eXa_bOy nah im in as root :) 1128590104 M * eXa_bOy thats my current issue. 1128590133 M * Johnsie What exactly is your issue? 1128590160 M * eXa_bOy down the bottom... the script that checks the installation 1128590161 M * eXa_bOy http://pastebin.com/384783 1128590173 M * eXa_bOy chcontext: vc_new_s_context(): Function not implemented 1128590188 M * Johnsie Oh, I think that's permissable. 1128590192 M * Johnsie I'm not too sure. 1128590201 M * Johnsie I run Gentoo VServer and that's a bit different. 1128590210 M * Johnsie I've never run Debian VServer. 1128590212 M * eXa_bOy ok, this is all new to me :) 1128590220 M * Johnsie But, I think you can skate by without that. 1128590226 M * Johnsie Don't quote me on that. 1128590227 M * eXa_bOy thanks mate 1128590235 M * eXa_bOy ill keep going and see how i go 1128590251 M * Johnsie I think it just indicates that certain features are off... and if they're not implimented, they'll inevitably fail. :) 1128590254 M * Johnsie Sure. 1128590259 M * Johnsie Good luck. 1128590267 M * Johnsie Bertl would probably be the guy you want to talk to. 1128590276 J * Aiken ~james@tooax6-100.dialup.optusnet.com.au 1128590287 M * Johnsie He's asleep, but whenever he appears, ask him...he wrote that script up. 1128590308 M * Johnsie He can tell you all about it. 1128590317 M * Johnsie Anyway, I'm off to sleep...nighty night. 1128590349 M * eXa_bOy niht mate 1128590366 M * eXa_bOy yes he wrote the doccos 1128590864 J * francois ~francois@korn.ctrlaltdel.ch 1128591048 Q * Aiken Remote host closed the connection 1128591162 Q * sven12 Quit: Leaving 1128591432 M * eXa_bOy anyone active? 1128591468 M * eXa_bOy http://pastebin.com/384795 1128591476 M * eXa_bOy where can i look to fix this issue? 1128593473 Q * litage Ping timeout: 480 seconds 1128593547 M * BWare eXa_bOy: Try upgrading util-vserver to latest 1128594025 M * eXa_bOy will do :) 1128594033 M * eXa_bOy er how do i do that? 1128594112 J * litage ~nick@203.220.55.70 1128594961 M * eXa_bOy Setting up util-vserver (0.30.204-5sarge2) ... 1128594962 M * eXa_bOy /var/lib/vservers: Function not implemented 1128594977 M * eXa_bOy this is getting frustrating :( 1128595096 M * daniel_hozac are you actually running a vserver kernel? 1128595131 M * eXa_bOy yup 1128595138 M * eXa_bOy hi daniel_hozac :) 1128595184 M * eXa_bOy maybe i should just give up. 1128595196 M * eXa_bOy im not upto this stuff. its all over my head 1128595230 M * daniel_hozac so uname -r returns something containing vs? 1128595275 M * eXa_bOy one sec 1128595288 M * eXa_bOy debian:/usr/local/etc/vservers/room1# uname -r 1128595288 M * eXa_bOy 2.2.20-idepci 1128595318 M * eXa_bOy http://linux-vserver.org/Step-by-Step+Guide+2.6 1128595319 M * daniel_hozac ... that's certainly not a vserver kernel. 1128595327 M * eXa_bOy this is the tutorial that ive been following. 1128595348 M * eXa_bOy and this is the stage where you start the server, its likme the final thing :( 1128595368 M * daniel_hozac you're supposed to reboot after installing the kernel. 1128595375 M * eXa_bOy ooh? 1128595377 M * eXa_bOy shit 1128595381 M * eXa_bOy ok ill do that :) 1128595410 M * eXa_bOy when it comes back should those functions be ok? 1128595418 M * eXa_bOy ill do a uname -r for you again 1128595520 M * eXa_bOy nope its still the same 1128595525 M * eXa_bOy no VS in the uname 1128595597 M * eXa_bOy http://pastebin.com/384795 is what im stuck with still. 1128595823 M * daniel_hozac then you didn't edit your boot loader configuration to boot the new kernel when you installed it. 1128596176 M * eXa_bOy it doesnt say too 1128596257 M * daniel_hozac "Installing your kernel and rebooting" 1128596266 M * daniel_hozac "After you've installed your kernel and setup your bootloader..." 1128596323 J * sven12 ~sven@80-235-89-98-dsl.prn.estpak.ee 1128597343 N * sven12 sven12-away 1128598273 Q * serving Ping timeout: 480 seconds 1128598781 Q * lonewolff Remote host closed the connection 1128598907 M * eXa_bOy ok i give up thanks anyway 1128598965 Q * eXa_bOy Quit: 1128599589 M * Hollow poor boy. giving up is like daying guys! don't do it! 1128599600 M * Hollow s/daying/dieing/ 1128600157 Q * ntrs_ Remote host closed the connection 1128600861 J * lonewolff ~lonewolff@host86-128-128-38.range86-128.btcentralplus.com 1128601261 M * hippo Hollow - quick question - what are the patches you apply to the gentoo kernel sources for vserver? 1128601288 M * Hollow genpatches + vserver patches 1128601862 J * VooDooMaster VooDoo@topas.informatik.uni-ulm.de 1128605002 J * serving serving@86.108.16.227 1128606266 Q * VooDooMaster Quit: There is a byte stuck in my modem!! 1128606464 Q * serving Ping timeout: 480 seconds 1128606615 N * sven12-away sven12 1128609469 J * ntrs ~ntrs@68-188-50-87.dhcp.stls.mo.charter.com 1128610264 Q * Nicoli Read error: Connection reset by peer 1128611420 M * sven12 hi everyone who are here 1128611454 M * sven12 I am still having a problem with per vserver guest quotas 1128611475 M * sven12 I have done things by following this guide: http://linux-vserver.org/Disk+Limits 1128611507 M * sven12 but when I add files / download files inside a vserver guest, available space does not decrease 1128611574 M * sven12 when I try to view it using df 1128611624 M * sven12 chxid doesn't give any errors 1128611654 N * Bertl_zZ Bertl 1128611661 M * Bertl morning folks! 1128611710 M * sven12 morning Bertl! 1128611753 M * sven12 so I got the vserver recompiled with dietlibc 1128611792 M * Bertl excellent! 1128611815 M * Bertl regarding disk limits: here a check list: 1128611816 M * sven12 starting up a vserver does give some errors 1128611839 M * sven12 secure-mount: write(): invalid argument 1128611852 M * Bertl - is the filesystem with the guests mounted with tagxid? (no remount, real mount) 1128611853 M * sven12 Failed to update mtab file 1128611865 M * sven12 no 1128611881 M * Bertl hah! strike on the first gues :) 1128611885 M * Bertl *guess 1128611887 M * sven12 I haven't touched fstab in guests at all :) 1128611897 M * Bertl no, not guest, on the host 1128611917 M * Bertl for the shared disk limits to work, you have to use xid tagging 1128611918 M * sven12 on host the vservers partition is mounted with tagxid 1128611931 M * sven12 yes,guests partition 1128611936 M * sven12 it is reiserfs 1128611956 M * sven12 also noatime is set 1128611960 M * sven12 and defaults 1128611987 M * Bertl okay, so it was a misunderstanding .. can you verify that with 'cat /proc/mounts' on the host? 1128612008 M * Bertl phone *sec* 1128612019 M * sven12 yes 1128612028 M * sven12 it is rw,noatime,tagxid 1128612144 M * sven12 there is also something more - on starting a guest, I can't use df - it doesn't show any information yet, I must always use this: ln -sf /proc/mounts /etc/mtab 1128612176 M * sven12 aftre that df seems to work again 1128612504 M * eyck anyone heard about min fanout filesystem? 1128612970 Q * yungyuc Quit: Changing server 1128613031 M * Greek0 Bertl: is the strange BUG() fixed already? I was offline for the last few days :-/ 1128613102 J * yungyuc ~yungyuc@220-135-53-220.HINET-IP.hinet.net 1128613120 Q * yungyuc Quit: 1128613168 J * yungyuc ~yungyuc@220-135-53-220.HINET-IP.hinet.net 1128613192 Q * yungyuc Quit: 1128613208 J * yungyuc ~yungyuc@220-135-53-220.HINET-IP.hinet.net 1128613307 M * hippo "genpatches + vserver patches" 1128613327 M * hippo Hollow: I guessed that - I meant what was included in the vserver patches - just the main patch, or the "split" patch too? 1128613362 M * mnemoc Bertl: can you mknod -m 666 foo c 1 3; ls > foo there? please 1128613395 M * Hollow hippo: only the main patch 1128613400 M * hippo OK, thanks :) 1128613460 M * Bertl back now .. just a second 1128613481 M * sven12 ok 1128613543 M * Bertl Greek0: I'd say so, but gndmstr(chuck) will have to confirm ... 1128613554 M * Bertl Greek0: http://vserver.13thfloor.at/Experimental/delta-2.6.13.3-vs2.1.0-rc3.2-rc3.3.diff 1128613572 J * mrec ~revenger@p54B02CB6.dip0.t-ipconnect.de 1128613579 M * Bertl welcome mrec! 1128613587 M * Bertl mnemoc: where? 1128613602 M * mnemoc Bertl: on a guest with CAP_MKNOD 1128613623 M * Bertl sven12: ad the df, that's a non set limit (i.e. limits are 0 -> no filesystem is shown) 1128613628 M * Greek0 Bertl: ic 1128613634 M * mnemoc Bertl: to see if i am the problem 1128613665 M * Bertl mnemoc: okay, will check! 1128613726 M * sven12 Bertl: so setting limit did not succeed - that's why df is not working as default? 1128613745 M * Bertl could be, how did you set it? 1128613752 M * daniel_hozac sven12: your mtab is empty, and so df doesn't know of any filesystems to check. 1128613769 M * Bertl that's another good explanation 1128613787 M * sven12 ok,thats the reason then 1128613788 M * Bertl df also looks at the mtab, no mentioning there, no filesystem is shown 1128613821 A * Bertl wishes he could make a copy of daniel's brain :) 1128613835 M * sven12 :) 1128613856 M * sven12 but then this other problem - why disk space does not automatically decrase when I downlaod a file inside a guest 1128613913 M * Bertl the funny part is, daniel was sitting around not saying much in the beginning, and it seems he absorbed (and categorized) all useful information since .. now he knows all the answers .. we are very glad to have you here, daniel! 1128613916 M * mnemoc :) 1128613925 M * daniel_hozac hehehe. 1128613933 M * sven12 :) 1128613937 M * daniel_hozac i do my best ;) 1128613951 M * Bertl can't say it too often, excellent work! 1128613975 M * Bertl sven12: what xid tagging method have you selected for the kernel? 1128613992 Q * mrec_ Ping timeout: 480 seconds 1128613997 M * sven12 let me check,in a sec 1128614021 M * Bertl testme.sh output might help ... 1128614059 Q * BWare Read error: Operation timed out 1128614131 J * ionsphere ~ionsphere@S01060050dab69981.no.shawcable.net 1128614192 M * Bertl welcome ionsphere! 1128614250 M * ionsphere hello Bertl! 1128614314 M * sven12 it is UID24/GID24 1128614331 M * Bertl mnemoc: http://pastebin.com/385075 1128614364 M * Bertl sven12: okay, that should be fine, do you ahve the testme.sh output at hand too? 1128614377 M * sven12 where could I find testme.sh? 1128614381 M * Bertl sec 1128614401 M * Bertl http://vserver.13thfloor.at/Stuff/SCRIPT/ 1128614444 M * sven12 ok 1128614454 M * mnemoc Bertl: yes, but ls > /tmp/foo ? 1128614474 M * Bertl works fine ... 1128614480 M * sven12 all tests succeded 1128614485 M * sven12 chbind works 1128614490 M * mnemoc :'( 1128614496 M * sven12 chcontext is working 1128614502 M * Bertl okay, great, VXI/kernel info? 1128614506 M * Bertl *VCI 1128614525 M * sven12 kernel 2.6.12.4 1128614531 M * mnemoc what is vxc? 1128614597 M * Bertl a test script, sec 1128614599 M * sven12 but what does VXI mean? 1128614607 M * sven12 or VCI 1128614656 M * Bertl mnemoc: (I just <3 google) http://vserver.13thfloor.at/Experimental/TOOLS/vxc 1128614682 M * mnemoc :) 1128614685 M * mnemoc thanks 1128614858 J * gndmstr ~gndmstr@ip1.pathworx.sbbsnet.net 1128614864 M * Bertl welcome gndmstr! 1128614872 M * gndmstr hi 1128614882 M * gndmstr question... kinda an emergency 1128614887 M * daniel_hozac sven12: could you do grep -Hr mtab /etc/init.d (or whatever the equivalent directory is in your distribution)? 1128614892 M * Bertl gndmstr: hmm? 1128614892 M * gndmstr i created a guest and it doesnt show on ifconfig 1128614900 M * gndmstr inside it it does not show the ip with ifconfig 1128614904 M * gndmstr yet it can be pinged 1128614913 M * gndmstr ssh cant find route to host tho from remote 1128614918 M * gndmstr but i can ssh into it from the host 1128614924 M * gndmstr i notced another thing 1128614925 M * Bertl sven12: VCI is the Vserver Configuration Information ... 1128614940 M * gndmstr i used the same template as i used on prometheus 1128614947 M * Bertl gndmstr: that's quite normal ... no emergency :) 1128614949 M * sven12 ok :) 1128614953 M * gndmstr and prometheus the qmail guest also does not show it 1128614956 M * gndmstr yet all the others do 1128614960 M * gndmstr hmm 1128614972 M * gndmstr all the others show eth0:guest ip etc 1128614978 M * Bertl ifconfig was replaced by 'ip' from iproute2 a few years ago 1128614986 M * gndmstr ahh 1128614992 M * Bertl most didstros didn't catch up though, so ifconfig is still around 1128614998 N * nokoya- nokoya 1128615012 M * Bertl if you use aliases (like the :guest) then ifconfig will show them 1128615028 M * gndmstr ahh 1128615028 M * Bertl you can do that by adding a 'name' entry to the interfaces config 1128615042 M * Bertl otherwise 'ip addr ls' will show everything 1128615120 M * gndmstr ahh ok 1128615138 M * gndmstr also odd this is the first guest that ssh isnt working. i have it fixed on the ip addy and port 22 1128615147 M * Bertl sven12: I'm not 100% sure that your kernel version supports reiserfs dlimits 1128615162 M * sven12 I see 1128615205 M * Bertl so I'd like to get some more info about the kernel ... 1128615220 M * gndmstr in abit 1128615224 M * Bertl gndmstr: btw, how is the qmail kernel doing? 1128615235 M * gndmstr the emergency is my router tech is bugging me about getting a guest up and running for her yesterday 1128615241 M * sven12 would the .config file be OK 1128615243 M * gndmstr it blew up this morning 1128615247 M * sven12 I could send it to pastebin 1128615258 M * Bertl sven12: yes please ... 1128615261 M * Bertl gndmstr: :( 1128615262 M * sven12 ok 1128615267 M * gndmstr but i forgot to log into the serial console so there was no trace 1128615278 M * mnemoc Bertl: http://pastebin.com/385085 <--- any idea about what can be the difference? 1128615295 M * gndmstr i documented what i could in the list.. be back in a bit after i get this sorted out for her 1128615322 M * Bertl mnemoc: nodev? 1128615346 M * mnemoc none on /tmp type tmpfs (size=16m,mode=1777) 1128615346 M * Bertl mnemoc: at least here, tmpfs is mounted with nodev 1128615381 M * Bertl mnemoc: could you try somewhere else than /tmp? 1128615382 M * mnemoc none on /tmp type tmpfs (rw) <--- host 1128615409 M * mnemoc my normal problem is on /dev/hdv1 1128615421 M * mnemoc i just repeated /tmp 1128615482 M * sven12 Bertl: here it is: http://pastebin.com/385093 1128615573 M * mnemoc aaaaaarg! nodev :\ 1128615580 M * mnemoc very indirect, but it's there 1128615748 M * Bertl sven12: as I thought: http://pastebin.com/385099 1128615782 M * Bertl jfs and reiserfs are missing (dlimit wise) in your patch 1128615805 M * Bertl so the only supported filesystems for 2.0 are ext2 and ext3 1128615840 M * sven12 ok... 1128615865 M * sven12 so I need a different patch? 1128615872 M * Bertl the next stable release (2.0.1) will probably include it ... 1128615879 M * sven12 ok 1128615882 M * eyck no xfs? 1128615912 M * Bertl eyck: don't understand xfs .. didn't manage to contact somebody there ... 1128615929 M * sven12 so I'll convert the vservers partition to ext3 and try again 1128615947 M * Bertl or, alternatively, you can try: 1128615952 M * Bertl patch-2.6.13-vs2.0.1-pre2.diff 1128615987 M * Bertl http://vserver.13thfloor.at/Experimental/patch-2.6.13.2-vs2.0.1-pre2.diff.bz2 (or that one) 1128616000 M * sven12 well,I also planned to use grsecurity with vserver 1128616012 M * sven12 I guess I have to stick with older versions then 1128616030 M * Bertl ah, k, you could also try to add the missing patches ... 1128616057 M * Bertl http://vserver.13thfloor.at/Experimental/FOR-2.0.1/ 1128616089 M * sven12 ok 1128616095 M * Bertl a few folks around here do heavy backporting :) 1128616111 M * mnemoc :) 1128616111 M * sven12 now I got something more to try at last :) 1128616121 M * Bertl eyck: do you know somebody who understands the xfs code? 1128616141 A * Bertl would be happy to add xfs to dlimit ... 1128616148 M * sven12 thanks for checking things,Bertl 1128616156 M * mnemoc i have grsec 2.1.6 merged with 2.0.1-pre2-rkill for 2.6.11.12.... and builds... but i'm not sure now if i'll use grsec 1128616188 J * tuxcapoeira ~capoeira@85-124-167-78.work.xdsl-line.inode.at 1128616227 M * Bertl sven12: you're welcome! 1128616256 M * mnemoc --barrier supose to be persistant? 1128616262 Q * liquid_ Ping timeout: 480 seconds 1128616278 M * tuxcapoeira hello sven12! 1128616404 M * sven12 hello tuxcapoeira 1128616449 M * Bertl mnemoc: yes, that's something the script (testfs.sh) checks for 1128616578 M * mnemoc # CONFIG_REISERFS_FS_XATTR is not set <--- may this be the reason why it's not preserved from boot to boot here? 1128616773 N * sven12 sven12-away 1128616911 M * Bertl mnemoc: either that or a missing 'attrs' option? 1128616945 M * Bertl mnemoc: don't ask me why reiserfs requires it :) 1128617012 M * mnemoc thanks for the tip ;) 1128617368 M * gndmstr to save me time is here something special that needs to be done for links? 1128617377 M * eyck Bertl: nope, sorry. 1128617378 M * gndmstr getting segmentation fault when i try it 1128617385 M * gndmstr so it isnt compatible 1128617397 M * gndmstr i hate links anyway but our router tech loves it 1128617403 M * gndmstr ill just tell her no 1128617478 M * Bertl ahem, what? 1128617558 M * gndmstr links isnt working 1128617559 M * SiD3WiNDR "compatible" ? :) 1128617568 M * gndmstr i just installed it and it is segfaulting 1128617580 M * Bertl links as the web browser? 1128617597 M * gndmstr yes 1128617609 M * Bertl ah, probably bad configuration 1128617629 M * gndmstr maybe. ill try reinstalling it. never had any config files to set up before on it 1128617646 M * Bertl check with strace. IIRC it had some issues with termcaps or so 1128617722 M * gndmstr ok have to install strace.. im giving her a bare bones system which is what she wanted :) 1128618952 M * gndmstr found out they are offering lynx now as well so i installed that instead and it works 1128619268 M * gndmstr gotta take the dog out. back in 10... im not gonna be able to do anything intrusive on that system till later at night since its a production system and too many people are relying on the radius for it to go down for even 2 min during the day 1128619286 M * mnemoc there is any reason to mount rootfs after chroot and not before? 1128619566 M * mnemoc inside chroot i wont find the locations i want to use for unionfs :\ 1128619734 M * Bertl hum? 1128619806 M * mnemoc i want to use unionfs at / 1128619828 M * mnemoc but vserver start chroots before mounting rootfs 1128619850 M * Bertl rootfs is not mounted, it's an --rbind, no? 1128619910 M * mnemoc ++ /usr/lib/util-vserver/secure-mount -a --chroot --fstab /etc/vservers/bugatti/fstab --rootfs only -n 1128619910 M * Bertl so mounting your unionfs on the guest root, is probably what you want 1128619944 M * mnemoc ^---- vserver --debug bugatti start 1128619980 M * mnemoc where should i call mount/umount then? 1128620008 M * mnemoc ++ test -e /etc/vservers/bugatti/fstab.local .oO 1128620010 M * Bertl pre start/post stop? 1128620032 M * mnemoc thanks :) 1128620096 M * Bertl btw, we did some tests regarding the scripts and I still have to evaluate them .. 1128620106 M * Bertl (regarding context and such) 1128620156 J * Nicoli ask@208.53.159.170 1128620181 M * Bertl welcome Nicoli! 1128620320 M * mnemoc pre start/post stop scripts recieve some argument or they are just executed? 1128620584 M * Bertl sec 1128620685 M * Bertl did I say I love google? 1128620691 M * Bertl http://vserver.13thfloor.at/Experimental/UTIL-VSERVER/SCRIPTS 1128620732 M * Bertl this is a test script run as each helper script, you see what is defined and what arguments are passed 1128620756 Q * prae Quit: Execute Order 69 ! 1128620801 M * mnemoc thanks :) 1128620980 M * Bertl you're welcome! 1128620987 J * traffic ~bob@206.127.78.199 1128621051 M * mnemoc no script there :p 1128621097 M * mnemoc but the output also work :p 1128621192 M * eyck nice, http://www.scaramanga.co.uk/stuff/qemu-usb/ 1128621212 M * Bertl ah, good to know ... 1128621250 M * eyck yeah, still not working though :( 1128621278 M * Bertl well, mrec? might start something 'similar' for pci soon? 1128621301 M * mnemoc Bertl: prestart is loaded 400 set -x lines _after_ secure-mount --chroot --rootfs only 1128621303 M * Bertl (unless I got all wrong again :) 1128621327 M * Bertl mnemoc: and prepre-start? 1128621350 M * mnemoc also after 1128621382 M * Bertl really, hmm ... 1128621403 M * Bertl but the rbind/chroot is supposed to happen after that ... 1128621416 J * Doener ~doener@i5387EB9F.versanet.de 1128621426 M * Bertl well, at least it seems to work in most cases 1128621432 M * Bertl welcome Doener! LTNS! 1128621449 M * Doener evening folks, Bertl! 1128621542 M * Bertl Doener: got a moment for me? (not necessarily right now :) 1128621556 M * Doener hm, i'd actually prefer right now ;) 1128621565 M * Bertl k :) 1128621608 M * Bertl http://pastebin.com/pastebin.php?dl=384656 1128621650 M * Bertl Doener: remember, some time back, we decided to drop the reference on hash/unhash 1128621685 M * Bertl now gndmstr has a setup, with a special guest, which seems to trigger this oops 1128621711 M * Bertl BUG_ON(!vx_info_state(vxi, VXS_SHUTDOWN)); 1128621720 M * Bertl /* context shutdown is mandatory */ 1128621729 M * Doener hm, let me get a recent source tree 1128621758 M * Bertl the backtrace, when evaluated, shows that the ref count reaches zero, while the context is still hashed 1128621780 M * Bertl http://vserver.13thfloor.at/Experimental/delta-2.6.13.3-vs2.1.0-rc3.2-rc3.3.diff 1128621796 M * Bertl (this is a 'fix?' I tried yestarday ... 1128621803 M * Bertl *yesterday 1128621817 M * Bertl but we are not sure yet, that this actually fixed it ... 1128621834 M * Bertl the important change is the first hunk, although that might not be obvious at first glance ... 1128621863 M * Doener hm, procfs holding a ref? 1128621912 J * liquid3649 ~inet@p549748BC.dip.t-dialin.net 1128621978 M * Bertl indirectly 1128621987 M * Bertl what happens is this (as far as I know) 1128622000 M * Bertl - procfs keeps entry _and_ reference to the task struct 1128622083 M * Bertl - the dcache shrink removes the inode 1128622092 M * Bertl - the task gets released 1128622102 M * Bertl - the vxi reference is dropped *bang* 1128622269 M * Bertl it would be easy to paper over this by adding a hash reference .. but I do not understand why/how this can happen ... 1128622275 J * mef ~mef@targe.CS.Princeton.EDU 1128622284 M * mef howdy 1128622286 M * Bertl welcome mef! 1128622306 M * mef My head is still underwater... plugging ahead with the xen / vserver comparison paper. 1128622312 M * mef Hope to get back to you after that is done. 1128622341 M * mef Maybe you can read a draft and give us some feedback. 1128622357 M * mef But we don't have a draft yet. 1128622385 M * Bertl hmm, then it will be hard :) 1128622413 M * mef Just use your imagination... ;) 1128622429 M * mef about using CFQ as a disk limit controller... 1128622436 M * mef I have not done that yet. 1128622448 M * mef I remember that the CKRM folks were putting together such a controller. 1128622477 M * mef However, as best as I can remember, it was purely priority based. 1128622495 M * mef For processes within the same priority it essentially is best effort. 1128622508 M * Bertl the cfq is part of the kernel 1128622517 M * Bertl (as io scheduler) 1128622522 M * mef I know... 1128622531 M * Bertl so the only thing to do is assign the priorities 1128622551 M * mef the CKRM folks just wrote a controller to group processes into a cfq priority. 1128622551 M * Bertl (IIRC it also inherits priorities across tasks) 1128622587 M * mef How does CFQ behave when two competing processes have the same priority? 1128622589 M * Bertl but to assign those priorities on a context basis is not a big deal 1128622615 M * Bertl they are sheduled in a non starving 'equal io count' way 1128622640 M * mef what does "equal io count" mean? 1128622646 M * mef fair share? 1128622661 M * Bertl IMHO that's basically as good as it will ever get, except from penalizing I/O heavy contexts 1128622698 M * mef I agree... just trying to wrap my head around what you mean by "equal io count". 1128622733 M * Bertl the io operations are accounted, and the cfq (by design) keeps the counts equal at the same priority 1128622741 M * mef ok I think that's what it is. 1128623231 M * mef stupid question: how do I assign a process a cfq priority? 1128623434 M * Bertl http://lwn.net/Articles/143474/ 1128623885 M * tuxcapoeira Bertl: vserver-copy only works with legacy is there a new version for default? 1128623973 M * tuxcapoeira sorry, my mistake 1128624428 M * Bertl hmm? 1128624937 M * mef We need to upgrade to the latest version of the FC4 kernel, which is based on 2.6.13. 1128624947 M * mef What version of vserver should I grab with that? 1128624959 M * mef We currently use 2.0-rc7 w/ 2.6.12. 1128624983 M * tuxcapoeira when i use vser-copy the script is searching for a config-file wich i dont have 1128625117 M * Bertl tuxcapoeira: you are trying to 'copy' a new-style guest, yes? 1128625142 M * Bertl mef: probably the 2.0.1-pre2 is the best choice 1128625159 M * tuxcapoeira Bertl: jap! 1128625165 M * Bertl mef: (there will be a 2.0.1 release soon, I guess= 1128625178 M * Bertl tuxcapoeira: the best way right now is this: 1128625193 M * Bertl - create a new skeleton (-m skeleton) with the 'new' options 1128625220 M * Bertl - copy over the guest contents (with cp or dump/tar) 1128625237 M * Bertl that's it ... 1128625260 M * tuxcapoeira just did so ;) but tx for the answer ;) 1128625285 M * Bertl np, next tool release will hopefully contain something like that 1128625314 M * tuxcapoeira apart from flowers-page, is there a description on 1128625319 M * mef bertl: thanks. 1128625338 M * tuxcapoeira the net about the confdir layout? 1128625396 M * Bertl http://www.nongnu.org/util-vserver/doc/conf/configuration.html? 1128625430 M * mnemoc " apart from flowers-page" 1128625465 M * tuxcapoeira Bertl: exactly the link i was talking about :( 1128625483 M * Bertl well, why don't you like it? 1128625485 A * mnemoc loves 'disable all CSS' from firefox 1128625504 M * Bertl the 'weedpage' style is pretty nice ... 1128625533 M * tuxcapoeira NOT because of colors - CSS is great!, i copied it to a texteditor anyway for understanding;) 1128625554 M * mnemoc :) 1128625564 A * Bertl wonders .o( are most browsers broken today? ) 1128625566 M * tuxcapoeira just missed some explenations yesterday but cant remember them now 1128625609 M * SiD3WiNDR mnemoc: why disable css when you can have sucha nice one on that page :) 1128625642 M * mnemoc SiD3WiNDR: printing purposes :) 1128625707 M * tuxcapoeira copy & paste :) 1128625820 M * eyck lynx works just fine, thankyouverymuch, 1128625836 M * Bertl indeed ... 1128625837 M * SiD3WiNDR youreverywelcome 1128626034 M * wibble hello! 1128626041 M * wibble I am back! 1128626046 M * eyck aloha, 1128626048 M * eyck good for you, 1128626056 M * eyck so, I'll make you a deal, 1128626062 M * eyck you'll fix 2.6, 1128626068 M * wibble ok 1128626071 M * eyck and I'll cut down on my jokes. 1128626077 M * eyck sounds fair, right? 1128626082 M * wibble hmm, a very tempting deal 1128626088 M * wibble how about 1128626089 M * Bertl extremely fair! I'd take it :) 1128626101 M * wibble you tell me how to fix bind9 :) 1128626109 M * wibble hello Bertl ! 1128626116 M * wibble what's wrong with 2.6 ? 1128626121 M * Bertl hey wibble! 1128626133 M * eyck well, for one, it doesn't work. 1128626135 M * mnemoc install djbdns and problem solved 1128626147 M * eyck djbdns is non-free, 1128626174 M * eyck although bind9 does suck :(, 1128626177 M * mnemoc its "i don't care what you do with it, just don't bother me" 1128626191 M * wibble lol 1128626201 M * eyck mnemoc: yeah, just like M$ Windows used to be, 1128626239 M * Bertl hmm, wasn't M$W more like: "I don't care what it does, as long as you payed for it?" 1128626271 M * tuxcapoeira why do u say was and used to? did i miss something? 1128626274 M * eyck Bertl: that was later, 1128626292 M * eyck well, whole Poland used to run on 'pirated' M$ Software, 1128626303 M * eyck except for those strange BSD and Linux folks, 1128626308 M * Bertl tuxcapoeira: now it is more like: "we know it's not working, but that's your fault, so pay for it!" 1128626350 M * wibble omg, I didnt know that bind9 would cause all this! 1128626361 M * mnemoc don't blame djb becuase MS used to not worry about poland people using pirated software 1128626376 M * eyck there was a lot of head scratching when they created 'M$ Poland', everyone was wondering - what are they going to sell, we all already 've got windoze and office from this guy ... 1128626378 M * Bertl wibble: see, one wrong comment .. and that's what you get ... goes off like a bomb :) 1128626401 M * tuxcapoeira Bertl: dont get me wrong, please but in many occasions they are right the problems were the users 1128626405 M * eyck mnemoc: you can't blame djb for anything, including shipping broken software. 1128626411 M * eyck mnemoc: that's the problem. 1128626434 M * mnemoc eyck: you have the source, you can fix it 1128626459 M * eyck mnemoc: what do you mean by 'have' ? I can't do anything with it. 1128626471 M * mnemoc only under glibc 1128626475 M * Bertl tuxcapoeira: sure, agreed, but that's the problem with tools, you have to make them people proof ... 1128626476 J * Blissex pcg@82-69-39-138.dsl.in-addr.zen.co.uk 1128626489 M * tuxcapoeira Bertl: *GGG* 1128626504 M * wibble didnt djb write qmail too? or am I mistaken? 1128626512 M * mnemoc yes, he did 1128626518 M * wibble just cause if we are taking about naff software... there is a prime example :) 1128626535 M * Bertl wibble: omg so he is responsible for gndmstr's issues as well! 1128626539 A * wibble hides 1128626566 M * mnemoc :D 1128626579 M * wibble so what does CAP_SYS_RESOURCE do in S_CAPS then? 1128626622 M * Bertl in what? 1128626639 M * wibble in this particular case: http://list.linux-vserver.org/archive/vserver/msg06158.html 1128626644 M * eyck you gotta hand it to djb though, he's great in attracting newbies. 1128626679 M * wibble oh Bertl - did you get anywhere with testing your reiser patch? 1128626704 M * wibble and has vs2.0 got support for maintaining disk limits upon reboot (had to create a dirty script to do it) 1128626709 M * Bertl wibble: well, back in the old days ... where we used to have S_CAPS entries in our configs, it was a real security issue ... 1128626731 M * Bertl wibble: still with ctx17a? then don't do it :) 1128626746 M * wibble ic, it would appear the only way to get bind9 to work correctly (or use djbdns) 1128626759 M * wibble well this is still running 1.2 1128626769 M * mnemoc wibble: you can build bind9 properly ;) 1128626772 M * eyck wibble: nooo, bind9 needs no SYS_RESOURCE, 1128626789 M * mnemoc --disable-linux-caps use to work 1128626791 M * eyck wibble: you just need to recompile your bind9 with no caps 1128626797 M * wibble i c 1128626804 M * eyck hmm, damn. 1128626977 M * Bertl Doener: if you have any clues/ideas, please let me know ... 1128626989 M * Bertl okay, I guess I'm off for now ... back later ... 1128626997 N * Bertl Bertl_oO 1128626997 M * Doener for sure i will ;) 1128627034 Q * tuxcapoeira Quit: Ex-Chat 1128627150 Q * Johnsie Remote host closed the connection 1128627192 J * Johnsie ~john@acs-24-154-53-217.zoominternet.net 1128628280 Q * Johnsie Quit: G'bye! 1128628770 J * Aiken ~james@tooax6-039.dialup.optusnet.com.au 1128629244 Q * sven12-away Quit: Leaving 1128630069 J * prae ~benjamin@sherpadown.net 1128630439 J * Johnsie ~john@acs-24-154-53-217.zoominternet.net 1128631111 M * mnemoc how much time testfs should normaly take? 1128631444 N * Bertl_oO Bertl 1128631457 M * Bertl mnemoc: you need to specify a few things to use it 1128631480 M * Bertl first, you need a partition/loopback of at least 128MB of size 1128631490 M * Bertl (better make it 256MB) 1128631510 M * Bertl then you need a mount point and the tools for xfs,jfs,reiserfs and ext2/3 1128631532 M * Bertl testfs.sh -h (for details), but -D for the device is mandatory 1128631547 M * mnemoc :D 1128631554 Q * ionsphere Quit: ChatZilla 0.9.61 [Mozilla rv:1.7.12/20050915] 1128631689 Q * liquid3649 Quit: 1128631814 M * mnemoc sweet.... destructive test 1128632015 M * mnemoc Bertl: how bad is this: http://pastebin.com/385495 ? 1128632598 Q * Johnsie Read error: Connection reset by peer 1128633046 J * IEF ~ief@cp115413-b.landg1.lb.home.nl 1128633046 J * Johnsie ~john@acs-24-154-53-217.zoominternet.net 1128633403 Q * IEF Quit: zzz 1128633428 M * Bertl mnemoc: mainly strange ... 1128633474 M * mnemoc uhm 1128633567 M * mnemoc what is the strange part? 1128633798 M * Bertl the 'i' parts in the flags? 1128633832 M * Bertl seems you have either very old tools or something is broken 1128633862 M * mnemoc what kind of something broken? 1128633908 M * Bertl well, the immutable flag seems not present ... 1128634163 M * mnemoc does it help to try on an ext3 loopback instead of reiser phisical device? 1128634193 M * Bertl ahem, the device is (re)formatted anyway 1128634201 M * mnemoc :) 1128634211 M * Bertl so no, no difference, and loopback file works fine 1128634230 M * Bertl but I'd test on ext2/ext3 too, so you get a feeling for _what_ is broken or not 1128634253 M * Bertl (btw, there is a -t (terse) and -n (nocolor) option too) 1128634266 M * Bertl okay, off again ... back later ... 1128634273 N * Bertl Bertl_oO 1128634276 M * mnemoc cu 1128634378 Q * francois Ping timeout: 480 seconds 1128634846 M * mnemoc Bertl_oO: this is even weirder http://pastebin.com/pastebin.php?dl=385556 .... -F ext2 and reiserfs crashes 1128634956 M * Aiken that is like for me loading the xfs module and accessing /dev/sda then have xfs crash 1128635031 M * mnemoc .oO 1128635065 Q * Johnsie Remote host closed the connection 1128635165 Q * Doener Remote host closed the connection 1128635639 J * Johnsie ~john@acs-24-154-53-217.zoominternet.net 1128635668 J * ntrs_ ~ntrs@68-188-50-87.dhcp.stls.mo.charter.com 1128635673 Q * ntrs Read error: Connection reset by peer 1128635689 Q * Johnsie Quit: 1128635947 M * Aiken mnemoc for what is worth reiserfs on 2.6.13.3-vs2.1.0-rc3 http://pastebin.com/385570 1128635961 M * Aiken so testfs can pass on reiser 1128637498 Q * traffic Quit: 1128639510 Q * Blissex Remote host closed the connection 1128639560 J * Johnsie ~john@acs-24-154-53-217.zoominternet.net 1128639568 Q * Johnsie Quit: 1128639849 J * Johnsie ~john@acs-24-154-53-217.zoominternet.net 1128639890 Q * prae Quit: Pwet 1128640014 Q * Johnsie Quit: 1128640162 J * Johnsie ~john@acs-24-154-53-217.zoominternet.net 1128641223 Q * ntrs_ Ping timeout: 480 seconds 1128642275 M * mnemoc my machine didn't come back after reboot :\ 1128642374 Q * litage Ping timeout: 480 seconds 1128642466 M * Aiken I don't like it when that happens, usually means I get to hold the camera very steady while I take a shot of the oops :( 1128642479 J * ntrs ~ntrs@68-188-50-87.dhcp.stls.mo.charter.com 1128642564 M * mnemoc having the machine at 420km is also a problem 1128642588 J * ntrs_ ~ntrs@68-188-50-87.dhcp.stls.mo.charter.com 1128642604 M * mnemoc i'll have to phone someone tomorrow to help me to start it again 1128642795 Q * yungyuc Ping timeout: 480 seconds 1128642890 M * gndmstr mnemoc: it may pay to check for some used equipment and get a serial interface box and an apc remote power management box. we have them on ours and its so nice. telnet into the cms and select the machine to get onto via serial console, and if necessary the apc lets you power cycle the machine, or just plain turn it off .. used they aren't very expensive 1128642942 M * gndmstr has saved the boss literally hundreds of miles in trips to the noc over time 1128643011 J * litage ~nick@203.220.55.70 1128643026 Q * ntrs Ping timeout: 480 seconds 1128643059 M * mnemoc around how much? 1128643148 M * gndmstr we have a 16 port cms unit, they are avail from 4 ports up.. and 8 outlet apc boxes.. i think we paid around $100usd for the cms and i think $75 for each apc.. found them on ebay... new they are quite expensive. 1128643185 M * gndmstr only thing i have to do is keep a chart here which tells me what machines are plugged into what apc and what cms:)