1113005107 M * micah Bertl: here is the uptime virtualization output: 1113005108 M * micah   16:55:20 up 8 min, 8 users, load average: 2.43, 1.77, 1.46 1113005108 M * micah USER PID %CPU %MEM VSZ RSS TTY STAT START TIME COMMAND 1113005108 M * micah root 8243 0.0 0.2 3920 1252 pts/8 S+ 16:47 0:00 bash -c sleep 600 & sleep 500; uptime; ps auxwww 1113005111 M * micah root 8244 0.0 0.1 3180 564 pts/8 S+ 16:47 0:00 sleep 600 1113005114 M * micah root 9173 0.0 0.1 3956 1024 pts/8 R+ 16:47 0:00 ps auxwww 1113005119 M * micah not an hour off this time! :) 1113005240 M * Bertl hmm, could you try once again on the same server? 1113005249 M * Bertl (i.e. is it still active/running?) 1113005258 M * Bertl the host I mean 1113005343 Q * Doener_zZz Read error: Operation timed out 1113005438 M * micah Bertl: sure, but I am confused what you are wanting me to do 1113005458 M * micah you just want me to do the same command? 1113005481 Q * berni Ping timeout: 480 seconds 1113005496 M * Bertl micah: yes, because the 'host' is now up for longer ... 1113005506 M * Bertl add an uptime before and after the chcontext sequence 1113005542 N * kevinp kevinp|gone 1113005545 M * Bertl I would like to know how much it is off (i.e. if it is constant or a multiple of the uptime ... etc) 1113005715 M * micah ok, its running, report back in about 10 minutes 1113005777 J * Doener_zZz doener@193.24.208.125 1113005790 M * Bertl micah: great work! thanks! 1113006283 M * micah Bertl: here's the output: 1113006284 M * micah # uptime; chcontext --ctx 100 --flag ^17 bash -c "sleep 600 & sleep 500; uptime; ps auxwww" ; uptime 1113006287 M * micah 19:15:07 up 8:45, 7 users, load average: 1.18, 1.22, 1.22 1113006290 M * micah 19:23:27 up 9 min, 7 users, load average: 1.01, 1.07, 1.14 1113006293 M * micah USER PID %CPU %MEM VSZ RSS TTY STAT START TIME COMMAND 1113006295 M * micah root 22501 0.0 0.1 3180 564 pts/8 S 19:14 0:00 sleep 600 1113006298 M * micah root 22794 0.0 0.2 3916 1248 pts/8 S+ 19:14 0:00 bash -c sleep 600 & sleep 500; uptime; ps auxwww 1113006301 M * micah root 22804 0.0 0.1 3180 564 pts/8 S+ 19:14 0:00 sleep 600 1113006303 M * micah root 23414 0.0 0.1 3960 1028 pts/8 R+ 19:14 0:00 ps auxwww 1113006306 M * micah 19:23:27 up 8:53, 7 users, load average: 1.01, 1.07, 1.14 1113006380 M * Bertl hmm, so all processes get the context start time, do I see that correctly? 1113006562 J * berni ~berni@svr01.mucip.net 1113006570 M * Bertl welcome berni! 1113006603 M * mikegrb welcome Bertl! 1113006932 M * micah Bertl: its strange because the first uptime is 19:15, but then the processes get 19:14 (the context's start time) 1113007641 M * Bertl yes, the uptime itself is virtualized quite fine 1113007650 M * Bertl the process start time is wrong ... 1113008439 Q * muadda Quit: Leaving 1113009259 Q * Doener_zZz Read error: Operation timed out 1113009560 Q * berni Ping timeout: 480 seconds 1113010215 J * Doener_zZz doener@193.24.208.125 1113010341 J * berni ~berni@svr01.mucip.net 1113013906 Q * micah Server closed connection 1113013907 J * micah micah@micha.hampshire.edu 1113014175 Q * mugwump oxygen.oftc.net nova.oftc.net 1113014175 Q * sith oxygen.oftc.net nova.oftc.net 1113014175 Q * laerec oxygen.oftc.net nova.oftc.net 1113014175 Q * Loki|muh oxygen.oftc.net nova.oftc.net 1113014175 Q * albeiro oxygen.oftc.net nova.oftc.net 1113014175 Q * DaPhreak oxygen.oftc.net nova.oftc.net 1113014175 Q * jd86 oxygen.oftc.net nova.oftc.net 1113014175 Q * bro oxygen.oftc.net nova.oftc.net 1113014175 Q * Zoiah oxygen.oftc.net nova.oftc.net 1113014175 Q * virtuoso oxygen.oftc.net nova.oftc.net 1113014175 Q * rs oxygen.oftc.net nova.oftc.net 1113014175 Q * logger oxygen.oftc.net nova.oftc.net 1113014175 Q * pusling oxygen.oftc.net nova.oftc.net 1113014175 Q * Seraph oxygen.oftc.net nova.oftc.net 1113014209 J * mugwump ~samv@210-54-92-184.ipnets.xtra.co.nz 1113014223 J * rs ~rs@194.98.28.10 1113014226 J * laerec ~cereal@stargate-galaxy.net 1113014232 J * Zoiah Zoiah@matryoshka.zoiah.net 1113014254 J * sith sith@12.152.184.42 1113014258 J * albeiro albeiro@albeiro.usercloak.oftc.net 1113014258 J * DaPhreak ~phreak@lms.rz.uni-greifswald.de 1113014258 J * jd86 ~jim@ip68-9-97-23.ri.ri.cox.net 1113014258 J * bro ~vanity@lanparty.lv 1113014258 J * Seraph kk@projects.verfaction.de 1113014258 J * virtuoso ~s0t0na@80.253.205.251 1113014258 J * pusling ~pusling@195.215.29.124 1113014258 J * logger ~rs@vds.pas-mal.com 1113014259 J * Seraph_ kk@projects.verfaction.de 1113014259 Q * Seraph Read error: Connection reset by peer 1113014259 Q * virtuoso Read error: Connection reset by peer 1113014259 Q * pusling Read error: Connection reset by peer 1113014266 Q * albeiro Server closed connection 1113014272 Q * DaPhreak Server closed connection 1113014272 N * Seraph_ Seraph 1113014272 Q * jd86 Server closed connection 1113014276 Q * bro Server closed connection 1113014277 J * albeiro albeiro@linux.gentoo.pl 1113014283 J * jd86 ~jim@ip68-9-97-23.ri.ri.cox.net 1113014289 J * pusling ~pusling@195.215.29.124 1113014293 J * bro ~vanity@lanparty.lv 1113014304 J * DaPhreak ~phreak@141.53.8.70 1113014308 Q * logger Server closed connection 1113014315 J * logger ~rs@vds.pas-mal.com 1113014505 J * virtuoso ~s0t0na@80.253.205.251 1113014584 J * Loki|muh loki@213.133.125.66 1113023498 M * Bertl night folks! 1113023509 N * Bertl Bertl_zZ 1113039632 J * peet ~chatzilla@217.170.94.110 1113039955 P * peet 1113039966 J * peet ~chatzilla@217.170.94.110 1113039982 P * peet 1113040001 J * peet ~peet@217.170.94.110 1113040063 N * laerec cereal 1113040126 J * kalou ~kalou@AFontenayssB-104-1-3-121.w80-11.abo.wanadoo.fr 1113040143 M * kalou Hi 1113040397 J * erwan_ho ~erwan@lns-vlq-39f-81-56-133-136.adsl.proxad.net 1113040827 M * peet 2all: how I can see, which flags/ccaps/bcaps are set for a security context? 1113042830 M * kalou peet: seems that /proc/virtual//status should provide this information on 2.6 1113042836 M * kalou (http://linux-vserver.org/HowTo+Read+ProcFS) 1113043157 M * peet ok, I see... is there a program that converts these bitmasks to something human readable? 1113043782 M * kalou peet: for this, I really don't know, sorry . 1113043897 M * peet thanks anyway. I'll try to investigate it... It's only bitmasks, I don't think it'll be hard. Thanks 1113046275 Q * flock Quit: Expert, n.: Someone who comes from out of town and shows slides. 1113046482 M * peet kalou: I wrote simplest script to list flags, in awk, it seems to work properly... 1113046889 J * flock ~restless@l192-117-111-12.broadband.actcom.net.il 1113047380 Q * flock Remote host closed the connection 1113047458 M * kalou peet: hey, nice .. is there an url to download it ? 1113047481 M * peet kalou: http://radlinux.peet.spb.ru/files/related/lsvattr.awk.gz 1113047511 M * peet I'll add ccaps & bcaps to it tonight 1113047539 M * peet usage -- gunzip & ./lsvattr.awk -v xid=, eg. ./lsvattr.awk -v xid=100 1113047539 M * kalou peet: thank you peet, I'll try this as soon a I've a 2.6 working. 1113047587 J * flock ~restless@l192-117-111-12.broadband.actcom.net.il 1113049440 J * DukeGangsta2 uefgangsta@i5387EDDD.versanet.de 1113049441 Q * DukeGangsta|aw Read error: Connection reset by peer 1113050935 Q * erwan_ho Remote host closed the connection 1113051903 P * peet 1113051999 J * eamacnaghten ~eddy@cpc2-cmbg6-3-0-cust235.cmbg.cable.ntl.com 1113052031 N * Doener_zZz Doener 1113052037 M * Doener morning! 1113052399 J * erwan_ho ~erwan@lns-vlq-39f-81-56-133-136.adsl.proxad.net 1113053717 Q * DukeGangsta2 Quit: ( www.nnscript.de :: NoNameScript 3.81 :: www.XLhost.de ) 1113054655 Q * Pazzo Quit: . 1113056527 N * Doener Doener|gone 1113056873 J * NiCeBoY woom_z@196.192.110.156 1113056880 T * NiCeBoY irc.Locean-indien.com 1113056882 P * NiCeBoY 1113056937 J * erwan_taf ~erwan@lns-vlq-39f-81-56-133-136.adsl.proxad.net 1113057368 T * DaCa http://linux-vserver.org/ | latest stable 1.2.10, devel 1.9.5, ng9.4 -- He who asks a question is a fool for a minute; he who doesn't ask is a fool for a lifetime -- share the gained knowledge on the wiki, and we'll forget about the minute ;) 1113057984 Q * SiD3WiNDR Quit: maintenance 1113058066 Q * kalou Ping timeout: 480 seconds 1113058281 J * Pazzo ~thomas@host130-250.pool8172.interbusiness.it 1113059140 Q * erwan_taf Remote host closed the connection 1113060372 J * muadda ~muadda@gar31-1-82-234-50-167.fbx.proxad.net 1113061297 N * Bertl_zZ Bertl 1113061331 M * Bertl morning folks! 1113061727 M * Bertl any native english speakers around? 1113062257 Q * muadda Quit: Leaving 1113062472 M * Hollow hey Beirdo 1113062474 M * Hollow grr 1113062477 M * Hollow ^ Bertl 1113062492 M * mikegrb Bertl: welcome 1113062492 M * Bertl LOL 1113062499 M * Bertl Hey Hollow! 1113062503 M * Hollow haha ;) 1113062507 M * Bertl hey mikegrb! 1113062548 M * Hollow sometimes i wonder why i can't tab-complete my exams... *g* 1113062567 M * Bertl well, maybe you do ;) 1113062573 M * Hollow this would be so awesome :P 1113062598 M * Hollow Bertl: why do you need a native speaker? 1113062621 M * Bertl well, anybody is welcome to check the LSM paper of course ... 1113062634 M * Bertl http://vserver.13thfloor.at/Linux2.6/index.php?page=LSM_Paper 1113062664 M * Hollow lol.. i already like the first sentence ;) 1113062695 M * Bertl (context: third article, first one is talking about virtualization in general, second is a glorification of xen ;) 1113062858 M * Bertl it has to be finished tomorrow .. so I need some proofreading/final adjustments ... 1113062891 M * Bertl (and I want to add another two paragraphs about community and important urls) 1113062930 M * Hollow sounds quite good though i'm not a native speaker 1113062964 M * Bertl good to hear ;) 1113063049 M * Hollow how's 1.9.6 going? any release plans yet? 1113063067 M * Bertl yes, should be coming soon ... 1113063090 M * Bertl 1.9.5.12 has all new features and all known issues fixed 1113063127 M * micah Bertl: good morning or evening :) 1113063135 M * Bertl morning micah! 1113063138 M * Hollow ok, but i think waiting for 1.9.6 is ok (at least for packaging) 1113063143 J * yarihm ~yarihm@217-162-204-144.dclient.hispeed.ch 1113063180 M * Bertl welcome yarihm! 1113063208 M * micah Bertl: do you think that the -3 debian patch could be released out into the debian-wild? 1113063221 M * yarihm yo Bertl 1113063235 M * Bertl micah: well, I'd prefer to fix the known issues first ;) 1113063238 M * DaCa Bertl: why is it called lsm_paper, are you doing on talk on the libre software meeting in dijon? 1113063317 M * micah Bertl: the reason I ask is because the freeze will be happening soon, so getting something in there before then would be good! 1113063324 M * Bertl no, it's an error on my side ... but I didn't bother to change it 1113063339 M * Bertl should be called FSM_Article ;) 1113063340 M * micah Bertl: apparantly just waiting on the ARM architecture to catch up on building things, and a glibc upload and then thats it 1113063349 M * Bertl (FSM being Free Software Magazine) 1113063415 M * aba well, as long as it resolvs critical issues, it might be accepted even later 1113063416 M * Bertl micah: let's see what we can fix today, you can whatever we have then put into debian tomorrow, okay? 1113063509 M * Bertl micah, aba: is that acceptable? 1113063528 M * Bertl and welcome aba! DaCa! btw ;) 1113063547 M * Bertl s/welcome/morning/ 1113063572 M * micah aba: yes, but only if those issues are what is understood as "Release Critical" issues 1113063596 M * micah Bertl: that sounds good to me! If we can get something in, then we can update it afterwards 1113063606 M * micah Bertl: once it is in there 1113063705 M * micah if we get something in today that is a working patch, but maybe people's process time is off by 1 minute, I think its acceptible :) 1113063733 M * Bertl it's not off by one minute, it's cosntatn at the start time of the context, no? 1113063822 M * micah yes thats right 1113063826 M * micah I was just giving an example :) 1113063828 M * aba well, what is the issue? 1113063841 M * aba .oO(or should I add a block hint to prevent last-minute uploads? :) 1113063868 A * micah scurries to upload ;) 1113063884 M * micah the problem we are working on is uptime virtualization 1113063890 M * aba micah: don't play that game with release team members :P 1113063892 M * micah meaning that if you do something like: 1113063906 M * micah chcontext --ctx 100 --flag ^17 bash -c "sleep 600 & sleep 500; uptime; ps auxwww" ; uptime 1113063924 M * micah the process times that are reported in the ps are not quite correct 1113063931 M * micah aba: was just kidding 1113063951 M * Bertl micah, aba: get working on that FSM article, I'll fix the uptime issues ;) 1113063980 M * aba actually, I don't expect the freeze happening too soon, as the arm buildd is still > 500 needs-build. 1113063981 M * micah aba: Bertl has done a lot of work to get a patch that works with debian 2.6.8-15, and we are just ironing out the last pieces 1113063996 M * micah aba: yeah, I noticed that it seems to have gone down, rather than up 1113064018 M * micah Bertl: I'll give the FSM article a pass 1113064047 M * aba well, we need to be < 100 needs-build, or even less. But of course, that's my personal guess, and not any citeable remark :) 1113064099 M * Hollow i'm off, cya later! 1113064102 M * micah we need a good definition for "virtualization" 1113064137 M * Bertl it's the third article, and the first one explains the different kinds of virtualizations ... 1113064159 M * Bertl (written by eamacnaghten (aka. Eddy)) 1113064162 M * micah ah, ok 1113064171 M * aba micah: you know that there are kernel updates going to happen, and we release with 2.4.27 and 2.6.8? 1113064246 M * micah aba: yes, the kernel-patch-vserver package only has patches against 2.6.11 right now, thats why we are getting the 2.6.8 ones worked out 1113064260 M * aba ah, ok. Great. 1113064270 M * aba (that's something that I consider to be RC, BTW :) 1113064273 M * Bertl 2.4.27 should not be a big deal, guess the latest patches work fine there 1113064295 M * Bertl (but somebody should probably try ;) 1113064363 M * aba Bertl: well, would be nice if it works w/o fuzzyness or so. 1113064375 M * aba apart from that, I'm gone now. Have a nice evening ... 1113064391 M * Bertl cya 1113064400 M * micah aba: yeah, the 2.6.8 patches are fuzz free, just small issues remaining :) 1113064559 J * lap1 lap1@glassrom3.mediasat.ro 1113064568 M * Bertl welcome lap1! 1113064617 M * lap1 nice bot :) 1113064639 N * Doener|gone Doener 1113064644 M * Doener evening! 1113064645 M * Bertl evening Doener! 1113064696 M * Doener 18:04:55 Bertl (context: third article, first one is talking about virtualization in general, second is a glorification of xen ;) 1113064716 M * Doener hmm... isn't the first one also pretty xen-ish? ;) 1113064749 M * Bertl well, maybe when eddy comes back (eamacnaghten) we can adjust that a little ;) 1113064751 P * lap1 1113064760 M * Doener and hey, if you can't change the clock, it has to suck... admins tend to use ntp and a clock that's correct is soooo Evil(tm) ;) 1113065004 M * Bertl okay, off for dinner now, back in 20 or so ... 1113065010 N * Bertl Bertl_oO 1113065062 A * micah hacks on the FSM 1113066200 N * Bertl_oO Bertl 1113066222 M * Bertl back now ... 1113067013 J * Art512 ~chatzilla@85.229.dial-up.xter.net 1113067017 M * micah Bertl: working on the article 1113067032 M * micah Bertl: can you describe what happens with disk buffering in a xen/uml environment? 1113067041 M * micah I've seen this detailed before... 1113067062 M * micah something like a file is to be read in the virtualized environment, so it makes a call to its kernel 1113067090 M * micah which then...etc...etc... ultimately that kernel must ask the host kernel for the file...etc 1113067137 M * Art512 Doener: Hi. If you remember yesterdey we had discussion about IF Aliases with same Ip addr, and do not finished it!!! 1113067141 M * Bertl micah: I moved the article, be careful if you are editing it right now 1113067172 M * Bertl micah: (i.e. copy your current version and use it for http://vserver.13thfloor.at/Linux2.6/index.php?page=FSM_Article ) 1113067199 M * Doener Art512: hm, i explained to you what happens, and then had to go... any further questions? 1113067257 M * micah Bertl: I am editing in an emacs window separate :) 1113067280 M * Bertl excellent! 1113067377 M * Art512 Doener: yes, you saw that i cannot assign same IP to more than one aliases on one IF, its ok:) 1113067408 M * Art512 Doener: but my rootdev is not same 1113067478 M * Art512 Doener: fo exmp. lo:vs01=dummy1:vs02 1113067542 M * Art512 Doener: and from vs01 i can see Ip of dummy1:vs02, why??? 1113067589 M * Doener because the virtualisation is based on the ipv4 address, not the interface address alias 1113067706 M * Art512 Doener: I use it for Web cluster, and I need same IP addresses on any VS in Server, how can i solve this problem????? 1113067733 M * Doener Art512: what's the problem anyway? all you see is that the ip address is used in an other vserver, too 1113067847 M * Doener you can use it in both vservers that way... just the interface address naming is a little messed up... 1113067893 M * Doener probably you should switch from 'tools create the address' to 'i pre-create the address'... Bertl (or someone else) can tell you the details, i got to go once again... 1113067897 M * Doener later folks! 1113067900 N * Doener Doener|gone 1113067903 M * Bertl cya 1113067941 M * Bertl Art512: as far as I understand, you want to 'share' one ip between two vservers? 1113068008 M * Art512 Bertl: no 1113068065 M * Art512 Berl: I want have different interfaces wint same IP 1113068066 Q * Pazzo Quit: .. 1113068136 Q * brc Ping timeout: 480 seconds 1113068152 M * Art512 Berl: for use appache , and configure it to listen at that IP 1113068166 M * Art512 Berl: you understand what i need? 1113068177 M * Bertl so eth0 with 192.168.0.1 and eth1 with 192.168.0.1 ? 1113068218 M * Art512 yes, and it will bi virtual IF (dummy, loopback) 1113068233 M * Bertl well, you have to use ngnet to get 'virtual' interfaces 1113068250 M * Bertl the interfaces used by 1.9.x are 'real' host interfaces 1113068281 M * Bertl nevertheless you can assign the same ip to different interfaces ... 1113068295 M * Bertl (not that I think it makes much sense, tough ;) 1113068319 M * Art512 and after, i do not see the IF configuratin from another VS??? 1113068389 M * Bertl without ngnet, all vservers bound to that ip will see both interfaces 1113068433 M * Art512 I need that my every VS have and use only one IF with same IP 1113068450 M * Bertl what's the point in doing so? 1113068532 M * Art512 it's web cluster with load balancing based on LVS and Linux-VServer 1113068592 M * Art512 I solve this problem with ngnet? 1113068602 M * Bertl i.c. well, you will have to use ngnet for that then ... it is basically designed for such tricks 1113068630 M * Bertl but be aware that arp isn't implemented there, so you ahve to make the arp entries statically 1113068662 M * Art512 ok, thanks, there are a ngnet for 2.4.x kernel? 1113068673 M * Bertl nope, sorry, just for 2.6 1113068711 J * Pazzo ~thomas@host130-250.pool8172.interbusiness.it 1113068719 M * Bertl wb Pazzo! 1113068738 M * Art512 how can i do it in 2.4.x??? 1113068798 M * Bertl no idea ... I guess you can not 1113068814 M * Art512 where i can find any info about ngnet????? 1113068850 M * Bertl http://linux-vserver.org/ToDo+List+NGNet 1113068856 M * Pazzo hi bertl! 1113068873 M * Art512 ok thanks! 1113068876 M * Bertl Art512: http://linux-vserver.org/NGNET-Testing-HOWTO 1113068886 M * Bertl you're welcome! 1113068900 M * Art512 it is a testing version??? 1113068919 M * Bertl yes, we are at ng9.4 and it's experimental 1113068970 M * Art512 oooo, there are not any stable version??????? 1113068993 M * Bertl no, not for ngnet 1113069057 M * Art512 other thing???????????? 1113069084 M * Bertl parse error: too many question marks ;) 1113069125 M * Art512 other thing? 1113069158 M * Bertl well, you can have a look at UML or Xen, they might be able to do what you want (and might be called stable ;) 1113069216 M * Art512 ok thanks!!! 1113069226 M * Bertl np 1113069519 M * micah Bertl: I am making many changes :) 1113069555 M * Bertl if it gets better ;) 1113069851 Q * Art512 Ping timeout: 480 seconds 1113069879 M * micah I think it is : 1113069880 M * micah :) 1113070077 Q * erwan_ho Remote host closed the connection 1113070461 M * micah Bertl: where should I put my changed document? 1113070477 M * micah I would prefer not to overwrite your original, but put it somewhere separate so you can compare 1113070599 M * Bertl well, put it over that, I can compare with the diff functionality 1113070614 M * Bertl but make sure to include my last paragraph! ;) 1113070825 M * micah Bertl: oops 1113070829 M * micah Bertl: well, I put it here: http://vserver.13thfloor.at/Linux2.6/index.php?page=FSM_Paper 1113070908 M * Bertl hmm, what did you do to the lists? 1113070944 M * micah I integrated them all into paragraphs 1113070954 M * Bertl you sure it has anything in common with the original article? 1113070992 M * micah definately, I took the original article and fixed spelling/grammar errors, tightened up the introduction, put the lists into paragraphs and then added more information 1113071045 M * micah there are a couple places in there where you could add some more information (I think I noted) 1113071070 M * Bertl okay, let me send you something ... 1113071073 M * micah have a read over it to see what you think 1113071185 M * Bertl I think it is good ... and we should add it to the wiki ... 1113071342 M * micah I pulled various points from the Vserver-Paper that had not been covered 1113071363 M * micah the original felt too much like it was bashing on UML/XEN and not saying anything about what vservers can do 1113071373 M * Bertl you've got email 1113071386 M * micah so I wanted to make sure it made that point as well as showed the unique aspects of this project 1113071394 A * micah imagines the AOL voice 1113071868 M * micah hmm, hardware virtualization has been done for a long time in the Sun world 1113071913 A * micah reads and reads 1113072011 M * Bertl (remember we are the third and final article in this focus ;) 1113072027 M * micah yeah he seems pretty Xen biased 1113072041 M * micah especially with the introduction: 1113072041 M * micah In the 1113072041 M * micah world of virtualization a new kid has arrived on the block that is getting 1113072041 M * micah noticed. Virtualization is about to go to the next level. The name of the 1113072042 M * micah kid? Xen. 1113072073 M * Bertl you probably can ask eddy when he comes back (eamacnaghten) 1113072087 M * Bertl btw, read the xen article ;) 1113072474 M * micah ah the xen article is writen by Moshe 1113072478 M * micah of course it is going to promote it :) 1113072498 M * Bertl now does the 'bashing' appear in a different light? 1113072503 M * micah hehe 1113072555 M * Bertl but seriously I like your modifications, maybe we can find some compromise? 1113072612 M * micah yuck, Today’s virtualization 1113072613 M * micah offerings are crippled by poor performance, lack of scalability 1113072618 M * micah ... :P 1113072629 M * Bertl for example the 'seven' sounds weird to me 1113072643 M * Bertl the dozen was much more catchy there, and not really wrong 1113072692 M * Bertl btw, it's Linux-VServer not Linux-Vserver 1113072701 M * micah sounds good, I just made modifications that I felt you could then pick and choose 1113072706 M * Bertl and 'In this way way Linux-Vservers' is way to much 1113072712 M * micah yeah, I thought about the 7/dozen thing 1113072732 M * micah i liked the way dozen sounded, but I wanted to make it smooth with the example, the example is only 7 servers 1113072783 M * Bertl well, I had at least eight ;) 1113072811 M * micah agree about that part being too much, I was trying to find a way to reconcile and conclude by coming back to the original point that opened the piece about the different layers of virtualization 1113072835 M * Bertl okay, you still have some time for this article today? 1113072861 M * Bertl if so, I'd like to try a few versions/combinations 1113072883 M * Bertl (and fix the debian issues in the meantime ;) 1113072925 M * micah I have some time, but not as much as I would like, I'm preparing for a trip and have many things to wrap up today! 1113072941 M * micah but I will help where I can 1113072963 M * Bertl excellent, I'll add you to the 'authors' okay? 1113073211 M * micah sure! 1113073224 M * micah that would be an honor :) 1113073261 M * Bertl okay, so let's try to find a catchy medium for the article ... 1113073488 Q * Pazzo Quit: .. 1113073670 M * Bertl micah: I updated the first part of the FSM_Article (see history) and included some of your changes, elt me know what you think 1113073709 M * micah wow, the xen article is very "corporate" 1113073723 A * micah goes to check out your changes 1113073725 M * Bertl indeed watson, indeed! ;) 1113073805 M * micah hmm, I am not seeing the changes 1113073811 M * micah maybe I do not understand the wiki :) 1113073837 M * Bertl http://vserver.13thfloor.at/Linux2.6/index.php?action=diff&page=FSM_Article&ver2=3&ver1=2 1113073845 M * micah thanks 1113073909 M * Bertl but do not focus on the changes ... 1113073967 M * micah ah, I was trying to find changes you had made to the one I put up (FSM_Paper), no wonder I couldn't see anything 1113073982 M * Bertl hehe, no didn't want to use the formatting 1113073991 M * Bertl s/use/lose/ 1113074212 M * micah changes look good so far 1113074240 M * Bertl okay, then I proceed in this manner ... 1113074284 M * micah it might be difficult now because I re-ordered things a little in order to better construct the weight of the argument 1113074585 M * micah it is hard to read the xen article :) 1113074619 M * micah I wonder if their live relocation is reloction between hardware level servers, or between servers locally on the host machine 1113075481 A * DaCa just uploaded a new patch-2.4.30-grsec2.1.5-vs1.2.10-dm.gz 1113075872 M * Bertl micah: okay, another update 1113076344 M * micah Bertl: ok, I'm looking 1113076364 M * Bertl please feel free to correct spelling and grammar there ... 1113076492 M * micah DaCa: I wonder if you are good at patch fixing, the debian kernel-patch-vserver has 2.4.27-8-vs1.2.10-rc1 and patch-2.4.27-8-vs1.2.10+q0.14.1-rc1 for 2.4.27. Would be nice to get patches in for -9 and the actual released 1.2.10 1113076903 M * DaCa micah: I am certainly not good at it, just started and getting experience while doing it :) 1113076926 M * micah these probably would be very easy :) 1113076942 M * Bertl DaCa: just try it, if you encounter any issues, I'll come to rescue ;) 1113077067 J * Nik ~Nik@cable-153-130.online.bg 1113077077 M * Bertl evening Nik! 1113077077 M * Nik hi all 1113077088 M * micah Bertl: changes are looking really good... I think that the first paragraph should be changed, while its nice to not be so "formal" it seems to undercut the legitimacy of the rest of the article 1113077105 M * Nik Bertl :-) 1113077111 M * micah it seems like we don't know what we are saying, just fumbling around :) 1113077117 M * Bertl micah: okay, make a proposal ;) 1113077165 M * micah Bertl: well, I had in my version, just cut that first paragraph, lead-in with this sentence, "Everyone is eager to virtualize their working environment to take 1113077182 M * micah advantage of the abstraction layer that it provides." and then continue with the next... "Some may require..." 1113077206 M * DaCa well, if its not super urgent, its one of those things I can look into on my daily train commute 1113077243 M * Bertl micah: okay, any changes to the current version so far? 1113077244 M * micah DaCa: trying to get these in before "freeze" which seems tied to the ARM buildd queue :) 1113077292 M * micah Bertl: no, I just finished reviewing your changes, and was going to take a whack at some pieces, but go ahead 1113077327 M * micah btw. before the list starts, there should be a space before "Basically you can virtualize" 1113077373 M * Bertl okay, added the first par change and the irc channel at the end ... 1113077382 M * Bertl didn't deal with space or layout yet ... 1113077420 M * Bertl (won't change it now) 1113077427 M * micah I have my eye on three paragraphs 1113077433 M * Bertl go ahead 1113077436 M * micah that I'd like to tighten :) 1113077463 M * Bertl well, if you don't lose any information/flow ... 1113077683 M * micah of coruse 1113077945 M * DaCa micah: atm I have some other things I'd like to finish first, but I'm willing to give it a try next monday or so 1113077961 M * Bertl np, we'll handle it ... 1113078187 M * micah Bertl: ok, have a look at the changes 1113078224 M * micah version 7 vs. 6 1113078272 M * micah I dont fully understand what this part means (maybe it should be explained more?): The para-virtualization techniques assume you have a single kernel running inside your virtual environment and this is the one and only virtual unit that exists. This means that to achieve 100% native performance, then everything runs as if the virtualization did not happen at all. 1113078276 M * Bertl don't agree with: The para-virtualization techniques assume you have a single kernel running inside your virtual environment and this is the one and only virtual unit that exists. 1113078282 M * Bertl hehe 1113078357 M * Bertl the original article had: 1113078359 M * Bertl Assuming that there is a single kernel running inside a 1113078359 M * Bertl virtual environment, which is, and that's the important part, 1113078359 M * Bertl the one and only virtual unit there, then 100% native 1113078359 M * Bertl performance means that everything runs as if the 1113078361 M * Bertl virtualization did not happen at all. 1113078364 M * Bertl But what if you run two virtual units, or five, or maybe ten? 1113078486 M * micah yeah, I dont understand that 1113078516 M * micah mostly my brain is giving a parse error 1113078581 M * Bertl okay, it basically means 1113078603 M * Bertl when xen speaks about almost 100% native performance 1113078622 M * Bertl then they mean a single unit running as guest 1113078647 M * Bertl (all xen testing is done with one guest) 1113078684 M * Bertl now they also do tests with more than one guest, but in this case they 'adjust' the percentage 1113078708 M * Bertl (i.e. they correct for the kernel) 1113078716 M * yarihm Bertl: what do you think, how good are my chances to have vserver work with selinux-patched kernels? 1113078733 M * Bertl 2.4 or 2.6? and what selinux parts? 1113078751 M * Bertl micah: does it make sense now? 1113078767 M * yarihm Bertl: 2.6, having no clue about selinux i can't tell ... but AFAIK the sources are patched with pax too 1113078778 M * yarihm is selinux in vanilla now? 1113078781 M * Bertl yep 1113078786 M * yarihm ah, improves my chances 1113078823 M * yarihm Bertl: in fact i'm talking about the gentoo-hardened kernel-sources ... i'll check whether i find a patch-list 1113078859 M * Bertl hehe, IIRC the hardened compiler wasn't even able to compile the tools ;) 1113078859 M * micah Bertl: I think so, my problem is I dont quite understand xen :) I think what you are trying to say there is that Xen's near-native performance claim is based on a single unit running. 1113078879 M * Bertl yes, and they have near native performance with that 1113078900 M * micah and that when they run tests with more than one unit running they adjust the percentage to correct for the kernel overhead, so that the statistic of 100% native is not a true statement. 1113078911 M * Bertl if you would adjust linux vserver performance the way xen does it, we would have 200% native performance with 10 units ;) 1113078922 M * micah ahaha 1113078934 M * yarihm Bertl: so there won't be much chances i get vserver on gentoo-hardened? 1113078938 M * micah ok, I understand the point of that paragraph now, I thought it was meaning something else 1113078949 M * Bertl yarihm: maybe, please try ... and let the ml know 1113078951 M * micah I can re-word it so that is more clear 1113078959 M * yarihm Bertl: ok, i'll do so 1113078959 M * Bertl please do so 1113078964 M * Bertl (both ;) 1113079417 Q * Nik Ping timeout: 480 seconds 1113079661 M * micah Bertl: have a look at the re-wording, I really liked your 200% native performance comment :) 1113079689 M * Bertl yeah, well, but we cannot make that 1113079709 M * Bertl it would bring up endless discussions how to calculate it properly ... 1113079729 M * yarihm and engineer-wars often suck :) 1113079762 M * Bertl precisely, maybe we should just try to point out that more units mean more overhead (as I tried in my first attempt) 1113079794 M * Bertl (while more units mean less overhead, even negative one sometimes with linux-vserver) 1113080134 M * micah yes, I see thats true 1113080182 N * Doener|gone Doener 1113080189 M * Doener evening! 1113080224 M * Bertl wb Doener! 1113080232 M * Bertl http://vserver.13thfloor.at/Linux2.6/index.php?page=FSM_Article 1113080239 M * Bertl (work in progress ;) 1113080282 M * Doener yeah, just noticed that it moved 1113080455 M * micah Bertl: maybe the re-wording I just did? It could perhaps have a little more added 1113080460 M * Doener btw, what kind of virtualization do you have in mind when talking at virtualization at the application level? 1113080480 A * Doener never thought of virtualizing his word processor... 1113080503 A * Doener .oO( hmm... i don't even have any installed on my box... ) 1113080551 M * Bertl micah: hmm, yeah, looks much better ... 1113080558 J * Nik ~Nik@cable-153-130.online.bg 1113080563 M * Bertl Doener: well, think gmail ;) 1113080594 M * Doener sorry, all i know about gmail is that some folks don't like it for various reasons ;) 1113080813 M * Bertl okay, some time ago, M$ announced a network based word application 1113080829 M * Doener yep, i remember that... 1113080832 M * Bertl (it was just PR, but the idea was to provide application level services via internet) 1113080849 M * Bertl that's what I mean by application virtualization 1113080860 M * micah oh they are developing that very hard now 1113080870 M * micah its like Xul in mozilla 1113080877 M * micah an application delivery framework 1113080981 J * trulux ~lorenzo@tuxedo-es.project.oftc.net 1113080991 M * Bertl welcome trulux! ;) 1113081008 M * trulux hey Bertl ! thanks ;) 1113081009 M * eamacnaghten Hi all 1113081014 M * trulux ok, here we go 1113081028 M * Bertl short moderation info: 1113081035 M * eamacnaghten Regarding Bertl's FSW article..... 1113081051 M * Bertl eamacnaghten is the one writing the first article for FSM 1113081065 M * trulux I was thinking on the status of kernelnewbies wiki, among the problems regarding the very limited capabilities of moin and also that the current releases don't support latex parsers and other cool stuff, even not in a really safe way, I was setting up wiki.tuxedo-es.org 1113081072 Q * Nik Ping timeout: 480 seconds 1113081094 M * trulux adding tex, ams maths, gnuplot, graphviz support... : http://wiki.tuxedo-es.org/Sandbox 1113081107 M * Bertl and trulux is a kernel 'hacker' working on virtualization ;) 1113081133 M * trulux and just created a a few pages for explaining and documenting common data structures 1113081143 M * trulux http://wiki.tuxedo-es.org/Super_block 1113081147 M * trulux http://wiki.tuxedo-es.org/Notes_on_Linux_kernel_hacking 1113081150 M * trulux etc 1113081172 M * trulux It was more for my own usage, but I think that It might be great if more people can contribute to it 1113081178 M * eamacnaghten Bertl: I believe the article would benefit from you declaring your involvment with the project early on. This will show the reader that it is written by someone who knows what he is talking about (not always the case for Magazine Article Writers). 1113081208 M * Bertl hmm, okay, we can add some lines about authorship 1113081220 M * trulux Bertl: what article? just curiosity :) 1113081230 M * Bertl http://vserver.13thfloor.at/Linux2.6/index.php?page=FSM_Article 1113081230 J * SiD3WiNDR luser@bastard-operator.from-hell.be 1113081235 M * Bertl welcome SiD3WiNDR! 1113081262 A * Bertl .o( this is truly an unusual saturday night ;) 1113081274 M * trulux haha 1113081293 M * eamacnaghten Bertl: Also I think you do not show enough that Linux-VServer can match the "virtuality" of UML and Xen feature for feature, like the fact you can install your own software on your virtual machine independantly of others, that you can run your own services and so on. 1113081333 M * Bertl hmm, okay, please go on ... 1113081333 M * trulux eamacnaghten: I think that a comparison table could rock for showing what you have with vserver and the other projects 1113081390 M * eamacnaghten trulux: That I think would be good, but be objective. Where there is an advantage with Xen and UML say it. Flame wars are pointless and counter-productive. 1113081393 M * trulux btw, doxygen can document the kernel source but you need to patch it first 1113081402 M * Bertl well, actually I think (but this would be something for your article, eamacnaghten) that Xen (or UML) and Linux-VServer are orthogonal 1113081417 M * trulux eamacnaghten: right 1113081448 M * trulux eamacnaghten: anyways, there're always people who want to mess others work rather than contributing to it, that's life (sadly) 1113081468 M * Doener Bertl, eamacnaghten: "we can build units very similar to real machines which can work side by side on the same hardware", i guess extending that a little, would be a way to tell about the virtualization level one can achieve 1113081498 M * Bertl Doener: feel free to add something there (as proposal) 1113081504 M * eamacnaghten Bertl: Also show if there are workarounds for where features are not supported. You mentioned (to me in private) about time zones for each virtual machine. I do not know if Linux-VServer have sorted out the ability for each virtual server to configure there own firewall yet, if it has, say so. 1113081541 M * Bertl no, this is something which will be done with ngnet, not possible right now 1113081549 M * Doener Bertl: comments/proposals in curly braces as usual? 1113081563 M * Bertl Doener: just normal text, we can check the history 1113081678 M * Bertl eamacnaghten: another thing which I was missing in your article, but which didn#t really fit into mine (IMHO) was: 1113081724 M * Bertl there is often the argumentation for UML or XEN: that if a kernel crashes, you just restart the virtual unit and be done ... 1113081731 M * Doener Bertl: "per context procfs permissions/visibility" - did i miss something? ;) 1113081749 M * Doener (just asking, no offense intended) 1113081756 M * Bertl hmm, yeah, please remove the per 1113081830 M * eamacnaghten Bertl: Yes - I suppose so - though I was concentrating on the Web/Service Hosting aspect of virtualization rather than the more volitile development side - though you are right - that is an advantage.... 1113081836 M * Bertl eamacnaghten: but nobody mentions that you have 10 kernel running there, and if the one kernel on the real machin crashes, you're screwed ... 1113081843 J * Nik ~Nik@cable-153-130.online.bg 1113081869 M * Bertl so it's basically the same as with Linux-VServer (except for the 9 additional kernels ;) 1113081889 M * Doener and Murphy says that only the host's kernel will ever crash ;) 1113081933 M * eamacnaghten Bertl: True to that as well.... However, I was about to say I was not trying to push the article in that direction. 1113082001 M * Bertl yeah, no problem with that, as I said, just commenting on that ... 1113082039 M * eamacnaghten Bertl: What I was doing with my article was to try and show how virtualization allows guys to have a the functionality and versility of a colo machine for the price of a virtual host, thus seriously effecting both those markets. 1113082044 M * micah eamacnaghten: have you seen the new drafts thus far? 1113082095 M * eamacnaghten Bertl: As people associate configuring there own kernel when running colo machines my article naturally bends towards Xen (and UML slightly), however... 1113082171 M * eamacnaghten Bertl: I think you are right that Linux-VServer can fill in that roll, to all due intents and practical purposes, just as well as Xen/UML at less (hardware resource) cost, but I think that really is one for your article.... 1113082235 M * Bertl yes, but you ahve to admit that your article (probably because you do not know linux-vserver that much) is a little Xen centric, no? 1113082253 M * eamacnaghten micah: I think so. 1113082282 M * Doener article updated... 1113082444 M * eamacnaghten Bertl: Yes - Guilty your honour! But it needs to be something-centric or else it does not flow right (IMHO). Please let me mitigate by saying I do not knock VServer anywhere - in fact I state it's assets by saying that it does not use any resource in the virtualization process itself while providing what normall is enough virtualization. 1113082449 M * Bertl eamacnaghten: ah, and I would really appreciate to see QEMU or Bochs or one of the other free emulators wthere instead of VMware ;) 1113082515 M * eamacnaghten Bertl: OK - I will look at them and update my article accordingly (if Tony will let me). I believe the deadline is extended a week so I will quickly get the next version in. 1113082539 M * Bertl if you need some QEMU examples, we have a lot QEMU experience here ... 1113082557 M * Bertl it's a really interesting emulation project ... 1113082576 M * eamacnaghten Bertl: Good - Does it offer the same functionality as VMWare? (Windows/Linux running side by side on different virtual machines)? 1113082612 M * Bertl yes, it is a complete PC (or PPC or arm) emulation 1113082614 M * eamacnaghten Bertl: Is it stable? Is it a "Emulated Hardware Virtual Machine" model? 1113082645 M * Bertl http://fabrice.bellard.free.fr/qemu/ 1113082653 M * Bertl http://fabrice.bellard.free.fr/qemu/ossupport.html 1113082656 M * eamacnaghten Will look at it now... 1113082672 M * Bertl http://fabrice.bellard.free.fr/qemu/status.html 1113083114 M * Bertl micah: guess I fixed the process uptime ;) 1113083137 Q * Nik Ping timeout: 480 seconds 1113083230 M * Doener Bertl: 1.9.5.12 is latest? going to do a port to 2.6.12-rc2... 1113083243 M * Bertl good idea, many changes? 1113083270 M * Doener don't know yet, wanted to know what i got to port, first ;) 1113083290 M * micah Bertl: oh nice :) 1113083302 A * Bertl is testing it now ... 1113083399 M * Doener Bertl: btw, is there a way to use a vim window to contain a bash or at least show the output of command one start from within vim (like :!ls) 1113083406 M * Doener ? 1113083434 M * Bertl sure, that's easy 1113083447 M * Bertl vi /tmp/x.txt 1113083462 M * Bertl :.!/bin/ls 1113083531 M * Doener hm, yeah that would do as a workaround 1113083548 M * Doener thanks 1113083550 M * Bertl ah, you mean split and goto shell? 1113083629 M * Doener i guess that's it ;) 1113083644 M * Bertl I don't think that vim can do that ... that's emacs stuff, but you could use screen to do this 1113083667 M * Bertl well, here mgt does it for me ... 1113083855 M * Bertl micah: yup, works fine now ... 1113083870 M * Bertl micah: did you get around testing some of the other stuff I mentioned? 1113083940 M * Bertl (probably not .. as I kept you busy ;) 1113084598 M * micah Bertl: great to hear it works fine now 1113084603 M * micah Bertl: you are right, I have been busy :) 1113084629 M * micah I can do some tests though, which one should we do? 1113084649 M * Bertl np, I guess except for the initpid stuff, which I'm going to check now, all other things should be harmless 1113084657 J * Nik ~Nik@cable-153-130.online.bg 1113085137 Q * Nik Ping timeout: 480 seconds 1113085714 M * Bertl Doener: do you remember the command you used to test the initpid session and such (the ps sequence, I mean) 1113085729 M * Doener ps axo stat,euid,ruid,tty,tpgid,sess,pgrp,ppid,pid,pcpu,comm 1113085739 M * Bertl thanks a lot! 1113085745 M * Doener you're welcome 1113086140 M * Bertl micah: http://vserver.13thfloor.at/Experimental/patch-2.6.8-15-vs1.9.5.x-4.diff 1113086158 M * Bertl should be fine for a release ... 1113086181 M * Bertl (i.e. might have the one or other issue, but nothing obvious atm) 1113086186 M * micah Bertl: excellent! 1113086196 M * micah Bertl: what should be done about the .x part? 1113086209 M * Bertl well, it is between 1.9.5 and 1.9.6 ... 1113086247 M * Bertl no idea how to handle that properly ... 1113086279 M * Bertl (it doesn't compare to any mainline version) 1113086303 M * Bertl probably naming it 1.9.5-4 would be the debian way ;) 1113087017 A * micah works on packaging this patch 1113087361 M * micah Bertl: looks like 2.4.29 and 2.4.30 are the only patches available on the website for 1.2 now 1113087402 M * Bertl yep, correct ... isn#t 2.4.27 patched up to something similar? 1113087410 M * Bertl (the debian 2.4.27-x 1113087548 M * micah there are patches for 2.4.27-8 with vs1.2.10-rc1 and vs1.2.10+q0.14.1-rc1 1113087568 M * micah maybe these 2.4.29/30 patches will apply, going to try that first 1113087592 M * micah debian 2.4.27 is -9 now 1113087645 M * DaCa micah: I am looking into patching debian 2.4.27-9 with patch-2.4.29-vs1.2.10.diff 1113087728 M * micah oh! great 1113087761 M * DaCa they dont apply, there is some work to do, but I think I can do it 1113087795 M * micah DaCa: awesome, any idea how much time it will take you (estimate?) 1113087890 M * Bertl DaCa: what about taking an older release and patching up with the deltas? 1113088054 M * DaCa micah: I can have a diff ready in an hour so I guess 1113088082 M * Bertl micah: can you provide a 2.4.27-9 patch (all in one?) 1113088101 M * micah DaCa: thats fine, dont rush... wasn't sure if you were thinking of still doing it on the train :) 1113088118 M * micah Bertl: you mean a patch against vanilla, like I did for 2.6.8-15? 1113088148 M * Bertl yep, but with diff -NurpP linux-2.4.27 linux-2.4.27-9 1113088166 M * Bertl (the last one required -p4 or -p5 IIRC) 1113088177 M * DaCa micah: well, now that I am in to it :) but I won't have the time to really test it tonight 1113088213 M * Bertl DaCa: you're adapting 1.2.10 to it, right? 1113088216 M * micah Bertl: it'll take a few minutes to download etc. 1113088251 M * DaCa Bertl: yes 1113088272 M * Bertl I#ll do it the other way, we then compare the results an have some verification 1113088510 M * Doener Bertl: do you remember why you removed the sock_init_data() call from x25_alloc_sock? 1113088528 M * Doener x25_alloc_socket even 1113088531 M * Bertl the first time? 1113088546 M * Bertl because it is done twice ... but I removed the wrong one ;) 1113088558 M * Doener well, the way it is done in 1.9.5.12 ;) 1113088573 M * Bertl yeah, there it is the correct one, probably in mainline now? 1113088597 M * Bertl (or did they remove the wrong one too *G*) 1113088631 M * Doener which is the wrong one? 1113088650 M * Bertl http://vserver.13thfloor.at/Experimental/FOR-1.9.6/delta-x25-fix02.diff 1113088702 M * Bertl so the one in alloc is the one to be removed 1113088704 M * Doener they removed neither... but the function got changed in some other way... x25 == sk now... 1113088722 M * Doener thus i ask what's the intention behind the removal :) 1113088754 M * Bertl well, it gets initialized twice, which will bail out on set/init happening a second time 1113088777 M * Bertl doesn#t really matter in mainline (except that it is not necessary) 1113088805 M * Bertl linux-vserver has a context reference in the socket, this would get assigned twice (which is bad) 1113088815 M * Doener ah, i c 1113088929 M * Doener otoh it won't get initialized in x25_make_new now... 1113089017 M * Bertl well, the only thing that matters is that it is only initialized once 1113089028 M * maharaja mhm 1113089030 M * Bertl i.e. sock_init_data(sock, sk); 1113089036 M * maharaja i again forgot how to build the vserver - damn 1113089039 M * maharaja i should write that down 1113089043 M * Doener x25_make_new calls x25_alloc_socket 1113089044 M * Bertl Doener: (with both sock and sk) 1113089058 M * Doener and x25_make_new doesn't call sock_init_date 1113089064 M * Doener s/date/data 1113089079 M * Bertl show me the x25_alloc_socket line 1113089123 M * Bertl *please* 1113089137 M * micah Bertl: http://people.debian.org/~micah/linux-2.4.27-vs-linux-2.4.27-9 1113089145 M * Bertl thanks! 1113089258 M * Doener Bertl: net/x25/af_x25.c:527 (in vanilla 2.6.11.6, looks the same in .12-rc2) 1113089273 M * Doener or what do you mean? 1113089280 M * Bertl so it has sock_init_data(NULL, sk); ? 1113089280 A * Doener is a little confused :) 1113089292 M * Bertl if so, then this is plain wrong ... 1113089295 M * Doener that is in x25_alloc_socket 1113089310 M * Bertl doesn't work with NULL there, has to be a socket 1113089313 M * Doener and x25_make_new calls x25_alloc_socket, but not sock_init_data 1113089339 M * Bertl yeah, I suspected they get it wrong ;) 1113089355 M * Bertl x25 is soooo broken ... 1113089367 M * Doener why doesn't it work with NULL? it has a if(sock) check... 1113089395 M * Bertl sec 1113089413 M * Doener with the vanilla source, the two calls to sock_init_data are bloat, but not broken AFAICT 1113089501 M * Bertl do they assign the socket later? 1113089513 M * Bertl or do they have the second init call? 1113089535 M * Bertl in .12-rc2 1113089725 M * Doener second init call 1113089739 M * Bertl okay, this 'second' initcall assigns the sock properly 1113089753 M * Bertl without it, ifconfig -a gives a kernel panic 1113089754 M * Doener yes 1113089771 M * Doener with vserver patches i assume, right? (the panic) 1113089786 M * Bertl yes, but probably without too 1113089800 M * Bertl you can try, just remove the second init from vanilla 1113089830 M * Doener ah, sorry, misread that... 1113089932 M * Doener to clarify some things: we agree, that the second call is needed. that fixes the x25_create_socket->x25_alloc_socket path. but there's also the x25_make_new->x25_alloc_socket path. and if we just remove the call to sock_init_data in x25_alloc_sock, then that path doesn't call sock_init_data at all 1113089958 M * maharaja mhm - broke my vserver... 1113089960 M * maharaja damn 1113089968 M * Bertl Doener: yup, that's how I see it 1113090028 M * Doener and it's fine not to call sock_init_data at all in the x25_make_new->x25_alloc_socket path? 1113090073 M * Bertl well, I guess so .. at least I didn't see any bad effects, but I haven't tested with x25 either ;) 1113090539 M * Bertl micah, DaCa: http://vserver.13thfloor.at/Stuff/Debian/ 1113090562 M * Bertl (moved all the debian stuff there) 1113090917 M * Doener Bertl: what about that one? http://www.13thfloor.at/~doener/vserver/patches/diff-2.6.11.6-vs1.9.5.12-x25-fix.diff 1113090987 M * micah Bertl: great, now we just need you and Daca to compare your patches and then I can upload and it will be complete :) 1113091037 M * Bertl Doener: should work fine too ... 1113091049 M * Bertl micah: yeah, mine is basically untested (not even compile tested)