1119398430 M * micah well we know that ext3 xattr/dquot reports incorrect quota 1119398453 M * micah which you found hadn't been fixed when you were porting the patch 1119398487 Q * dsoul Quit: Lost terminal 1119398503 M * micah so there is *some* xattrs 1119398516 M * Bertl might be, don't really remember, but I'm pretty sure that we do not change 'how' xattrs work ... 1119398568 M * Bertl do you get any messages in the dmesg output? 1119398617 M * micah no, i dont 1119398634 M * micah nor in syslog anywhere 1119398686 M * Bertl okay, could you compile a plain debian kernel with the same .config you used for this one, and see if the xattrs work there? 1119398738 M * micah you mean without vserver support, right? 1119398749 M * Bertl yes, without the vserver patch 1119398806 M * micah ok, it will take a long time 1119398810 M * micah because this machine is very slow 1119398817 M * micah but i will do it and report back :) 1119398826 Q * Givur Quit: 1119398842 M * Bertl hmm, compile it on a different one, doesn't really matter, no? 1119398874 M * Bertl (different being faster :) 1119398895 Q * berni Ping timeout: 480 seconds 1119398906 M * micah heh, true 1119398937 M * micah actually, i am running a kernel that is most likely from that sourcee 1119398943 M * micah without vserver 1119398949 M * micah touch /tmp/x 1119398954 M * micah and then chattr +i /tmp/x gives: 1119398957 M * micah chattr: Inappropriate ioctl for device while reading flags on /tmp/x 1119398976 M * micah slightly different behavior 1119399008 M * Bertl well, you can try on an ext2/3 too 1119399017 M * Bertl it might be disabled for tmpfs in your configs 1119399022 M * micah right 1119399039 M * micah it works on an ext3 1119399048 M * micah rm: cannot remove `x': Operation not permitted 1119399077 M * Bertl good, but your kernel config is likely to be different, no? 1119399087 M * micah no, its exactly the same 1119399091 M * micah except for the vserver pieces 1119399101 M * micah i used the stock debian .config 1119399128 M * Bertl didn't you talk about reducing/removing unnecessary stuff? 1119399146 M * micah yes, but I failed and decided to not do that right now 1119399148 M * Bertl IIRC, about 95% of the default debian config are unneccesary 1119399158 M * micah yes, I agree 1119399176 M * micah thats why iw as going to reduce, and I did and failed to include something important, requiring someone to go to the colo 1119399189 M * micah so then i decided to just use the debian stock one, because I know it works and I had a deadline 1119399205 M * Bertl i.c. 1119399267 M * Bertl so which kernel is that? linux-2.6.8-15-vs1.9.5.x-4 ? 1119399377 M * micah 2.6.8-16 1119399415 M * micah looking at path version 1119399444 M * micah patch-2.6.8-15-vs1.9.5.x-4.diff 1119399459 M * micah but it is against version -16 1119399473 M * Bertl did it apply without fuzz/rejects? 1119399507 M * Bertl do you have a delta between 2.6.8-15 and 2.6.8-16 ? 1119399522 M * micah yes, no fuzz/rejects at all 1119399531 M * micah i can make one 1119399547 M * Bertl please do so, I'll look into it 1119399556 M * micah ok, it'll be just a minute 1119400477 M * hiaslboy good night - :-) 1119400482 M * Bertl night hiaslboy! 1119400559 P * hiaslboy Kopete 0.9.2 : http://kopete.kde.org 1119401201 M * micah trying to track down -15 source 1119401593 M * Bertl you can upload a vanilla -> -16 too 1119401650 M * micah Bertl: have preferred diff args you want me to use? 1119401657 M * Bertl -NurpP 1119401774 J * berni ~berni@svr01.mucip.net 1119401782 M * Bertl wb berni! 1119402172 M * micah those patch options sound like indigestion 1119402210 M * Bertl what about -PurpN ? 1119402240 M * Bertl or maybe -ruNpP ? 1119402559 M * micah I prefer NurpP! 1119402570 M * micah it sounds like something you ate is coming back u p 1119402669 Q * shuri Quit: Leaving 1119402671 M * Bertl btw, I received the FS magazine yesterday ... with a really nice article there :) 1119402685 M * micah oh! 1119402688 M * micah it came out? 1119402695 M * micah I was wondering what happened with that 1119402712 M * micah is it online? 1119402714 M * Bertl obviously, not sure if they send a copy to each of the authors, or just to me ... 1119402742 M * Bertl yeah, it's online too, for some time now, actually 1119402773 M * Bertl http://www.freesoftwaremagazine.com/free_issues/issue_05/focus-linux_vserver/ 1119402785 M * micah Bertl: here is -15 vs. -16: http://riseup.net/~micah/15vs16.diff 1119402803 M * micah oh wow heh 1119402812 M * micah there is my name :) 1119402833 M * Bertl I have to scan the frontpage and the article ... it looks quite nice in the printed version :) 1119402901 M * Bertl http://www.freesoftwaremagazine.com/current_issues/#issue_05 1119403021 M * micah i wonder where I can get a printed copy 1119403056 M * Bertl (it reads: Focus: Virtual Servers and Free Software, and the third article headline is: LINUX-VSERVER ) 1119403155 M * micah very nice! 1119403380 M * Bertl micah: does that sound right to you? 1119403384 M * Bertl With this option enabled, the Token Bucket Scheduler will 1119403384 M * Bertl be able to put a context on hold, when the tokens are 1119403384 M * Bertl depleted and reschedule the tasks once the minimum of 1119403384 M * Bertl required tokens has been reached again. 1119403483 M * micah yes, but you could re-word it to be more "active" which makes things easier to understand: 1119403503 M * Bertl I'm listening ... 1119403536 M * micah With this options enabled, when the tokens are depleted the Tocken bucket Scheduler will be able to put a context on hold, and reschedule tasks once the minimum of required tokens has been reached again. 1119403547 M * micah err, Token not Tocken :) 1119403588 M * micah it just changes it into the "active voice" 1119403640 M * Bertl i.c. but in this case, it would probably be better to make it Scheduler Tokens, because otherwise folks won't know of which tokens I'm talking about, no? 1119403656 M * micah which is a grammar thing, like "The cat ate the mouse" is active, but "The mouse was eaten by the cat" is passive. 1119403688 M * micah yes, I was just puzzling about how to resolve that, your suggestion is the answer 1119403734 M * micah its hard to explain, but people reading english are able to understand active voicing more than passive 1119403752 M * micah although both are comprehensible, one is just more easily grasped readily 1119403777 M * micah it has to do with not knowing what the subject of the sentence is until later 1119403781 Q * jdgiguere Quit: I will be back later 1119403796 M * micah which makes it harder to understand what you are referring to and you have to go back and piece it together 1119403809 M * Bertl what about punctuation, shouldn't there be a colon after depleted? 1119403828 M * micah most people use passive voice when writing, but learn to use active over time 1119403841 M * micah blah blah blah grammar blah :) 1119403854 M * Bertl actually I really appreciate this ... 1119403866 M * micah naw, the colon isn't necessary 1119403901 M * micah if there was perhaps one additional item in the list that you enumerate happening after depleted, then a colon could be used more freely 1119403964 M * mugwump hey, nice article in free software magazine 1119403975 M * Bertl yeah! ;) 1119404023 M * Bertl micah: so how's this?: 1119404024 M * Bertl With this option enabled, when the Scheduler Tokens are depleted, the Token Bucket Scheduler will be able to put a context on hold, and reschedule tasks once the minimum of required tokens has been reached again. 1119404115 M * micah one thing could be done to make it perfect, but I dont know if you want to re-order this part 1119404123 M * micah hold on typing 1119404148 M * Bertl mugwump: you are welcome to join the rewording game ... 1119404162 M * micah When the Scheduler Tokens are depleted and this option is enabled, the Token Bucket Scheduler will be able to put a context on hold, and reschedule tasks once the minimum of required tokens has been reached again. 1119404197 M * micah that sounds much better 1119404204 M * mugwump split the sentence I reckon 1119404218 M * micah but I was guessing that you wanted to have "With this option enabled" at the beginning 1119404252 M * Bertl hmm, not necessarily ... but actually I'd like to add some information into it ... 1119404263 M * micah here is vanilla vs. -16: http://riseup.net/~micah/vanillavs16.diff 1119404284 M * Bertl to be precise, the VSERVER_HARDCPU option (which this is the help text for) 1119404308 M * Bertl does enable the scheduler code, but not the actual hard scheduling ... 1119404328 M * Bertl which requires the user to set the hard_sched flag ... 1119404339 M * Bertl (and probably configure the token bucket) 1119404371 M * mugwump With this option set, when the “Token Bucket” is empty, the context will be put on hold, so none of its processes get any CPU time. The context will still have tokens added into its bucket at the prescribed rate, and will resume execution once it reaches the "minimum" level of tokens. 1119404434 M * Bertl hmm, the 'The context will still have tokens added' part sounds confusing to me ... because thats pretty orthogonal, no? 1119404464 M * Bertl I mean the token bucket itself is not changed by this option 1119404543 M * Bertl but the direction sounds good to me, maybe something like: 1119404668 M * Bertl With this option set, when a contexts Token Bucket is empty, the processes will be put on hold, until a required minimum of tokens is available again. 1119404708 M * mugwump wow ... live migration with Xen, that's pretty flash 1119404741 M * Bertl maybe s/hold/hold (no CPU time will be consumed)/ 1119404773 M * Bertl guess we could do that with linux-vserver too, btw ;) 1119404829 A * mugwump wonders how MOSIX is coming along and how well the patches would combine 1119404862 M * Bertl I wouldn't bother with mosix actually ... nobody wants vserver guests with head nodes ... 1119404902 M * mugwump I never looked into it in detail, but just something that lets you treat a cluster like a big NUMA system 1119404902 M * Bertl but recently I spent a little time thinking on guest snapshoting and migration ... 1119404935 M * Bertl and I think the 'real' issues such approaches have to face are actually non-issues with linux-vserver 1119404955 M * Bertl what do we need to freeze? 1119404964 M * Bertl - processes + memory 1119404968 M * mugwump active network sockets are the bitch 1119404970 M * Bertl - filehandles and sockets 1119404992 M * Bertl - context information 1119405019 M * Bertl given that the filesystem problem is solved (GFS, coda ...) 1119405042 M * micah With this option set, when a context's Tocken Bucket is empty the processes will be put on hold until a required minimum of tokens are available again. 1119405043 M * Bertl and the 'network' can be switched in place ... 1119405050 M * micah dang it... TOKEN not tocken 1119405054 M * mugwump for active network sockets, you could simply say that the IPs can't change during migration, to begin with anyway 1119405065 M * Bertl micah: does tocken mean something to you? :) 1119405132 M * micah This option, when set, will put conext specific processes on hold when a context's Token Bucket is empty until a required minimum of tokens are available again. 1119405161 M * mugwump I really think it should be two sentences. too much information for one gulp. 1119405173 M * Bertl micah: yeah, sounds good, except that it is wrong .. ;) 1119405182 M * micah Bertl: I dont know why, but my brain is wired to tock :) 1119405198 M * micah Oh! What is wrong? :) 1119405229 M * Bertl it does not do that stuff, it enables the code, which can be used to do it, when you use the sched_hard flag for that context :) 1119405252 M * micah Split into two sentences: This option, when set, will put context specific processes on hold when a context's Token Bucket is empty. When a required minimum tokens are available again the processes will be rescheduled. 1119405261 M * micah ah! 1119405290 M * Bertl and let's change the 'required minimum' to 'configured minimum' in this case 1119405382 M * micah This option, when set, will enable you to set the sched_hard flag on contexts. When this flag is set and a context's Token Bucket is empty, its processes are put on hold. When the configured minimum tokens are available again, the processes will be rescheduled. 1119405404 M * micah See this document for more information: 1119405406 M * micah :) 1119405418 M * Bertl so what about: This will compile in code, which allows the Token Bucket Scheduler to put processes on hold when a context's tokens are depleted and the sched_hard flag is set. 1119405468 M * Bertl (because the enable to set is wrong too, you can set it anyways, it's just ignored :) 1119405497 M * Bertl maybe the second part could be: 1119405540 M * Bertl (they will not be able to consume CPU resources, until a configured minimum of tokens is reached, and the context is scheduled again) 1119405587 M * micah that sounds good 1119405596 M * micah a couple fixups: 1119405665 M * micah This will compile in code that allows the Token Bucket Scheduler to put processes on hold when a context's tokens are depleted and the sched_hard flag is set. Processes in this context will not be able to consume CPU resources until a configured minimum of tokens is replenished, and the context is scheduled again. 1119405699 M * micah do you think it is worthwhile to say *how* these tokens come back? 1119405708 M * Bertl IMHO no 1119405713 M * micah I could imagine myself reading this and thinking that sounds cool, but i dont know how to get the tokens back :) 1119405741 M * Bertl okay, idea for the token/flag part: 1119405774 M * Bertl This will compile in code that allows the Token Bucket Scheduler to put processes on hold when a context's tokens are depleted (given that the sched_hard flag is set). 1119405790 M * Bertl and for the refill part: 1119405858 M * Bertl Processes in this context will not be able to consume CPU resources until the Token Bucket has reached a configured minimum. 1119405881 M * Bertl (doesn't that somewhat imply that it _is_ refilled?) 1119405933 M * micah a question about the sched_hard part... you seemed to say earlier that you dont actually set the sched_hard flag, it is ignored... is it set automatically when this is compiled in? 1119405960 M * Bertl no, you can set this flag anytime for each context individually 1119405977 M * Bertl but, if this code is not compiled in, it will have no effect at all 1119405986 M * micah it sort of implies that, but thats why i suggested "replenished" because that word really implies it. 1119406053 M * Bertl okay, we can use replenish, no problem with that .. 1119406085 M * micah then instead of saying (given that the sched_hard flag is set), it might be good to make it obvious that it is required for this to work by saying something like (this requires that the sched_hard flag is set on the context) 1119406192 M * Bertl hmm ... 1119406218 M * Bertl requires sounds wrong to me, but I can't explain why ... 1119406267 M * Bertl what about (provided that ...) 1119406453 M * Bertl This will compile in code that allows the Token Bucket Scheduler to put processes on hold when a context's tokens are depleted (provided that the sched_hard flag is set) 1119406455 M * micah yes, that could work as well 1119406464 M * Bertl Processes belonging to that context will not be able to consume CPU resources until a configured minimum of tokens is replenished, and the context is scheduled again. 1119406477 M * micah that is sounding really good now 1119406494 M * Bertl mugwump: your opinion? 1119406528 M * micah I know some kernel config options tell you to reference some other docs, it might be good if it could also do so, because otherwise you are left having to figure out how to set the sched_hard flag 1119406555 M * Bertl yeah, I'll refer to the paper or so ... 1119406570 M * micah or you could say ... provided that command-to-set-sched_hard-flag has been run :) 1119406572 M * mugwump sorry to disagree with you micah but I don't think replenished works there 1119406592 M * micah mugwump: its ok, whats your suggested alternative? 1119406638 M * Bertl s/replenished/has been reached/? 1119406652 M * mugwump that's what I was thinking, sans making the words agree with each other 1119406671 M * Bertl s/is replenished/has been reached/ actually 1119406697 M * micah it doesn't make sense to me how they are reached worded that way 1119406714 M * micah when it is replenished it is implied that it happens "naturally" 1119406731 M * micah ie. as the CPU turns :) 1119406773 M * mugwump well, it makes little difference. it is correct, but I'm always a bit nervous when I have to use "is" after a plural, even when it is correct 1119406787 M * micah but when you say "until a configured minimum of tokens has been reached" you should say how it has been reached like "until a minimum of tokens has been reached through the normal process of CPU scheduling" or whatever 1119406822 M * mugwump This conversation is going on, perhaps wiki-ing the text would be best. 1119406844 M * micah and you are right, it would be "until a configured minimum of tokens are replenished" not "is replenished" 1119406883 M * Bertl minimum of tokens has been reached (natural refill) ... :) 1119406909 M * micah hehe, natural refill = replenished 1119406921 M * Bertl nah, I guess it doesn't really matter who or why it is refilled ... 1119406924 M * micah so you could say minimum of tokens has been replenished instead of the is/are 1119406939 M * micah as long as you know you dont have to do anything, it is auto 1119406956 M * micah woop! gotta go, ciao 1119406974 M * Bertl cya, and thanks! 1119407015 M * Bertl what about: until a configured minimum of tokens has been rached again. ? 1119407023 M * Bertl *reached 1119407068 M * mugwump yeah, that works. Actually tbh lots of these suggestions have been fine :) 1119407107 M * Bertl so would this change anything (meaning) if I make it: 1119407121 M * mugwump I think the golden rule with writing easy to understand English is to keep the sentences short, stay away from putting too much meaning into too few words and not deliberate over them too much 1119407138 M * mugwump and of course not repeat yourself, of course 1119407143 M * Bertl Processes belonging to that context will not be able to consume CPU resources until a configured minimum of tokens has been reached, and the context is scheduled again.? 1119407250 M * Bertl well, that's not that simple with me .. I once wrote an entire essay consisting of just 4 or 5 sentences ... 1119407267 M * Bertl (was about 5-6 pages long :) 1119407314 M * mugwump you're writing documentation for the kernel patch I assume... why not throw all your drafts on a wiki and I'll go through and review for appropriate diction. 1119407325 M * mugwump s/a wiki/the wiki/ of course :) 1119407349 M * Bertl well, I'm almost done, but I can upload a 'current' version there ... 1119407437 M * mugwump I'm pretty lazy. I won't change anything unless it doesn't make sense. 1119407470 A * mugwump exaggerates a little there 1119407751 M * Bertl http://vserver.13thfloor.at/Linux2.6/index.php?page=Kconfig 1119407762 M * Bertl argl! 1119407813 M * mugwump looks fine in edit mode :) 1119407821 M * Bertl yeah, it's actually a nice bug ... 1119407837 M * Bertl now fixed ... doesn't work as first command :/ 1119407870 M * Bertl as command in the first line/first col 1119407912 M * mugwump Legacy Kernel API enabling "legacy guests"? what do you mean by a "legacy guest", there? 1119407942 M * Bertl guests created with the legacy tools/config 1119407964 M * Bertl probably confusing I agree ... 1119408109 M * mugwump Do you test it much without that setting? 1119408159 M * Bertl well, yes and no, thing is, some remaining tools still requires it 1119408183 M * Bertl *require 1119408213 M * Bertl I was hoping for a tool release which would do without that 1119408232 M * Bertl actually we are close (95% seem to work quite fine without it) 1119408274 M * mugwump Ideally we want to be able to say which version you need to turn that off... 1119408290 M * mugwump Also ideally the tools shouldn't need the legacy API to support the legacy config method 1119408312 M * mugwump Not that this is new to you ;) 1119408354 M * Bertl yeah, the first part should have been 'recent tools' and the second one, well ... you know :) 1119408387 M * Bertl currently I think it's better to leave it on ... 1119408437 M * mugwump ok I've saved that ... first three entries have been reviewed 1119408458 M * Bertl if enrico decides not to continue the tools (and keep them in shape) they'll probably be rewritten anyways ... 1119408569 M * Bertl objection your honor! :) 1119408607 M * Bertl if you do not like Proc Security, the only alternative is ProcFS Security, /proc is just plain wrong ... :) 1119408647 M * mugwump ok, I didn't know that 1119408724 M * mugwump maybe "CPU rationing" would be a better term than "Token Bucket Scheduling" 1119408914 M * Bertl maybe, do we use 'Token Bucket Scheduling' anywhere? 1119408937 M * Bertl and 'CPU Rationalizer' doesn't sound too good to me :) 1119408971 M * mugwump Oh, under the hood it makes sense to call it Token Bucket Scheduler, as that's the algorithm name 1119409020 M * mugwump But I think that terms like "rationing CPU usage", "CPU rations", etc sound good 1119409024 M * mugwump well, it's just an idea 1119409047 M * Bertl k, will think about it ... 1119409111 M * mugwump OK, there you go ... minimal other changes, mostly typographic, things like changing per CPU to per-CPU 1119409160 M * mugwump they probably still want to be re-flowed to conform to a standard width if it matters 1119409201 M * mugwump one thing ... if it's still necessary to run the vprocunhide script to make ProcFS :) security sane, then that should be mentioned there 1119409257 M * mugwump bah ... missed one, “on disk representation” should be “on-disk representation” 1119409282 M * Bertl yep vprocunhide is still required 1119409377 M * mugwump what's some examples of tools that no vprocunhide breaks? 1119409406 M * Bertl it basically breaks everything using proc :) 1119409506 M * Bertl or to be precise, everything depending on non-process proc entries ... 1119409649 M * mugwump btw it would be "CPU rationer", not "CPU Rationalizer" ;) 1119409671 M * Bertl yeah, I know, was just a joke ... 1119409725 M * mugwump kernel panics and oops are different, aren't they? 1119409766 M * mugwump ok, well there we go ... that last question relates to the second to last paragraph 1119409772 M * mugwump s/paragraph/option 1119409841 M * Bertl yep, the panic is when the kernel calls out, the oops is the information it dumps before ... 1119409944 M * Bertl hmm, what about: event of a kernel panic (as part of the oops trace)? 1119410050 M * Bertl 'do not store per-context information in inodes' doesn't that suggest that it is stored elsewhere (e.g. some file or so?) 1119410099 M * Bertl maybe 'do not store per-context information _for_ inodes' ? 1119410260 M * Bertl s/which is the secure default./which is a secure default./ 1119410346 M * Bertl 'You do not need to enable this option for normal operation of the Token Bucket Scheduler' what is the 'normal' oepration? 1119410388 M * Bertl s/improve interactivity and latency/improve interactivity and reduce latency/ ? 1119410421 M * Bertl do you have numbers for the 'marginally' aprt? 1119410606 M * mugwump sorry was in another channel, reading ... 1119410644 M * mugwump 'do not store per-context information in inodes or memory', then 1119410655 M * mugwump or your suggestion 1119410702 M * mugwump normal operation of the token bucket scheduler is to set the proportion of CPU aside and not worry about "minor" deviations from the allocations :) 1119410709 M * mugwump rather than incur overhead 1119410710 M * Bertl ad debug: compared to the debug functions, the history part adds signigicant overhead, compared to the overall use/number of syscalls/whatever both are negligible ... 1119410737 M * Bertl *significant 1119410773 M * mugwump ok. well, I guess it's a balance of not wanting to overly slow things down whilst still collecting enough information to debug problems 1119410809 M * Bertl the thing is, both debug and history only happen when syscalls are executed 1119410822 M * Bertl i.e. vserver syscall switch commands 1119410843 M * Bertl the amount of commands per minute is typically 0.0001 :) 1119410855 M * Bertl (on the average of course :) 1119410892 M * Bertl i.e. overhead will only be noticeable if you call context create in a tight loop ... 1119410902 M * Bertl (or something like that) 1119411004 M * Bertl ad inode (tagging none): they are present in memory, but they are not made persistant (on the filesystem) 1119411069 M * Bertl and the actual 'settings' default to untagged (if that option is selected) so memory xid tags are always 0 (except for some FS which are always tagged, like devpts) 1119411229 M * mugwump well admin commands could do with gratuitous logging I reckon 1119411235 M * mugwump s/commands/syscalls/ 1119411319 M * Bertl hmm, actually I guess I'm wrong here, forgot about the various debug messages we have in the get/put stuff ... 1119411351 M * Bertl (that's probably why I disabled it by default ...) 1119411394 M * Bertl will time it pretty soon ... need to do some profiling anyway 1119411730 J * Doener` ~doener@p54876EA9.dip.t-dialin.net 1119411750 M * Bertl hmm, morning Doener`! 1119411972 M * mugwump hey btw svk now has support for git in the svn version 1119411985 M * mugwump via SVN::Mirror::Git 1119412161 Q * Doener_zZz Ping timeout: 480 seconds 1119412463 M * Bertl cool! 1119412512 M * Bertl guess I'm off to bed now .. but I'll read the irc log tomorrow, and add/change what I consider relevant (on the wiki page) 1119412549 M * Bertl have a good whatever everyone ... and thanks for your help mugwump! 1119412559 N * Bertl Bertl_zZ 1119415549 J * Aiken_ ~james@tooax8-226.dialup.optusnet.com.au 1119415867 Q * Aiken Ping timeout: 480 seconds 1119418060 J * shuri shuri@64.235.209.226 1119418899 Q * Aiken_ Read error: Connection reset by peer 1119418899 J * Aiken ~james@tooax8-226.dialup.optusnet.com.au 1119423857 J * sukria ~sukria@sargon.lncsa.com 1119424732 J * dsoul darksoul@pingu.ii.uj.edu.pl 1119425424 J * rs ~rs@imhotep.rhapsodyk.net 1119425789 Q * Aiken Quit: Leaving 1119428408 Q * rs Quit: rs 1119428628 M * meebey good morning 1119428752 Q * mugwump Ping timeout: 480 seconds 1119429198 Q * Doener` Quit: Leaving 1119430026 J * rs ~rs@staff.lycos.fr 1119430106 M * meebey can I have a general vserver .sh script? 1119430112 M * meebey which is executed for all vservers? 1119430820 Q * sukria Ping timeout: 480 seconds 1119431895 M * id what do you want to accomplish with the script ? 1119433392 Q * case Ping timeout: 480 seconds 1119434474 Q * zimbo Quit: 1119434805 P * berni 1119434819 Q * rs Quit: rs 1119435494 J * zimbo ~zimbo@callisto.dom.bonis.de 1119436015 J * eXplasm explasm@p549F7AC0.dip.t-dialin.net 1119436083 Q * shuri uranium.oftc.net keid.oftc.net 1119436083 Q * BWare uranium.oftc.net keid.oftc.net 1119436083 Q * albeiro uranium.oftc.net keid.oftc.net 1119436083 Q * atsab uranium.oftc.net keid.oftc.net 1119436083 Q * brc uranium.oftc.net keid.oftc.net 1119436083 Q * micah uranium.oftc.net keid.oftc.net 1119436083 Q * Loki|muh uranium.oftc.net keid.oftc.net 1119436083 Q * ArnY[AwaY] uranium.oftc.net keid.oftc.net 1119436083 Q * Pazzo uranium.oftc.net keid.oftc.net 1119436083 Q * SNy uranium.oftc.net keid.oftc.net 1119436083 Q * virtuoso uranium.oftc.net keid.oftc.net 1119436083 Q * meebey uranium.oftc.net keid.oftc.net 1119436083 Q * janra uranium.oftc.net keid.oftc.net 1119436083 Q * lilo uranium.oftc.net keid.oftc.net 1119436083 Q * matti uranium.oftc.net keid.oftc.net 1119436083 Q * mcp uranium.oftc.net keid.oftc.net 1119436148 J * atsab ~as@lotes.vtu.lt 1119436161 J * shuri shuri@64.235.209.226 1119436161 J * BWare ~bware@office.intouch.net 1119436161 J * albeiro ~albeiro@albeiro.usercloak.oftc.net 1119436161 J * matti matti@linux.gentoo.pl 1119436161 J * brc bruce@200141099215.user.veloxzone.com.br 1119436161 J * micah micah@micha.hampshire.edu 1119436161 J * mcp ~hightower@wolk-project.de 1119436161 J * Loki|muh loki@satanix.de 1119436161 J * ArnY[AwaY] ~arny@copernic.lina.sciences.univ-nantes.fr 1119436161 J * Pazzo ~Pazzo@host130-250.pool8172.interbusiness.it 1119436161 J * SNy ~mfr@bmx-chemnitz.de 1119436161 J * virtuoso ~s0t0na@80.253.205.251 1119436161 J * lilo ~lilo@lilo.usercloak.oftc.net 1119436161 J * meebey meebey@booster.qnetp.net 1119436161 J * janra janra@paradox.homeip.net 1119437329 J * Thorsten ~Thorsten@dsl-084-058-141-239.arcor-ip.net 1119437541 Q * eXplasm Ping timeout: 480 seconds 1119437909 J * eXplasm explasm@p549F670A.dip.t-dialin.net 1119437946 Q * Thorsten Quit: Leaving 1119438163 Q * brc Ping timeout: 480 seconds 1119439025 J * Skid ~chris@82-32-192-192.cable.ubr05.newt.blueyonder.co.uk 1119439061 M * Skid hello.. I've been searching around for ages and can't find any info on traffic graphing individual vservers (2.4) - anyone have any pointers pleaes? 1119439185 J * sukria ~sukria@213.223.184.193 1119439960 Q * eXplasm Remote host closed the connection 1119440012 J * eXplasm explasm@p549F670A.dip.t-dialin.net 1119441662 Q * eXplasm Ping timeout: 480 seconds 1119441736 J * erwan_taf ~erwan@81.80.43.67 1119442210 J * eXplasm ~explasm@p549F4A78.dip.t-dialin.net 1119442610 J * brc bruce@201008047032.user.veloxzone.com.br 1119442705 J * rs ~rs@staff.lycos.fr 1119443171 Q * Skid Read error: Connection reset by peer 1119443172 M * alexx hello, i read "32-bit distribution vservers run normally on a 64-bit linux host, but faster, sometime a lot." is it real ? and why it's faster on 64-bit host thant on a 32-bit host (but with a 64-bit CPU) 1119443243 M * maharaja maybe because of the bigger registers (64bit instead of 32) -> twice as much parameters may be passed to subfunctions, etc. 1119443309 N * Bertl_zZ Bertl_oO 1119444217 M * maharaja good morning bertl 1119444335 J * case ~case@donpanic.faveve.uni-stuttgart.de 1119444745 M * alexx hello Bertl_oO 1119444773 M * alexx maharaja, 64bit register is used by optimisation at the compilation level ? 1119445508 M * maharaja alexx: yes 1119445518 M * maharaja afaik 1119445551 M * maharaja the only other thing that might be is, that they optimized the whole x32 architecture while inventing x64 1119445554 P * erwan_taf Leaving 1119445563 M * alexx oki, so with 32-bit vserver and 64-bit host, only kernel thing is optimizes : right ? 1119445584 M * maharaja so that complex operations (like division, etc.) run faster 1119445597 M * maharaja i would guess so 1119445618 M * alexx if it's CPU level, 64bit or 32bit host not change if the CPU is 64bit 1119445775 J * _are_ ~are@213.30.230.63 1119445785 M * _are_ Hi 1119445980 N * _are_ are|linuxtag 1119446115 M * maharaja alexx: i do not understand what you mean 1119446150 M * alexx sorry, i have a poor english :/ 1119446163 M * maharaja no problem - please rephrase your statement 1119446215 J * eXplasm2 ~explasm@p549F61D3.dip.t-dialin.net 1119446381 Q * eXplasm Ping timeout: 480 seconds 1119446435 M * maharaja alexx: so? 1119446446 M * alexx not important :) 1119446461 M * alexx thanks for your help 1119446514 M * maharaja np 1119448952 Q * Hollow Remote host closed the connection 1119448961 J * Hollow ~Hollow@home.xnull.de 1119449782 Q * sladen Ping timeout: 480 seconds 1119450601 Q * rs Quit: rs 1119451198 M * meebey IMHO there is a bug in util-vserver 1119451218 M * aba meebey: tell that you're speaking about the debian version 1119451242 M * meebey I am calling in post-start a script which uses vserver XXX running to check if the vserver is alive 1119451277 M * meebey the vserver which was started that triggered the script, is not running according to that "vserver XXX running" check 1119451302 M * meebey its post-start sso its running... 1119452590 J * sladen paul@starsky.19inch.net 1119452803 J * Jani ~mail@G9d73.g.pppool.de 1119452805 M * Jani Hi all 1119452923 J * rs ~rs@staff.lycos.fr 1119453247 Q * sannes Ping timeout: 480 seconds 1119454482 Q * are|linuxtag Quit: bbl 1119454699 M * Jani Bertl: Could you poke me please when you are avaible? 1119454749 A * SiD3WiNDR hands Bertl_oO a nice sharp pointy object 1119454814 M * Jani Hey, that is mean 1119454830 A * Jani looks around and searches something to dodge. 1119455406 M * Jani Have someone Patch up the 2.6.12 Kernel successfully? 1119455413 M * Jani 2.6.11.12 1119455831 M * DaPhreak which one Jani ? 1119455837 M * DaPhreak 11.12 or .12 ? 1119455906 M * Jani Oh 1119455911 M * Jani *grummels* I see the problem. 1119455921 M * Jani I have thought to use .12 but have download .11.12 1119455927 M * DaPhreak heh 1119455972 M * DaPhreak for 11.12 use patch-2.6.11.11-vs2.0-rc4.diff.bz2, which should apply nearly/absolutly clean 1119456027 M * DaPhreak for .12 use Doeners patch for .12 1119456029 M * DaPhreak (http://www.13thfloor.at/~doener/vserver/patches/patch-2.6.12-vs2.0-rc4.diff) 1119456107 M * Jani Yeah, I just need to download the right kernel version ;) 1119456113 M * Jani HEy, .11.12 and .12 sounds similar 1119456114 M * DaPhreak heh 1119456134 M * DaPhreak and so does .11 and 11.11 ;) 1119456937 J * Doener ~doener@p54876EA9.dip.t-dialin.net 1119456952 M * Doener evening folks! 1119456956 M * Jani Heya 1119456975 M * Jani Yeah, with the right kernel this works ;) 1119456993 A * Doener finally got a silent box... :) 1119457081 M * Doener now using athlon halt disconnect stuff and nforce2 fsb dynamic fsb lowering :) 1119457097 M * Doener s/fsb dynamic/dynamic/ 1119457914 M * Pazzo hi all! i know that it is possible to add caps while a vserver is running - is it also possible to add ip addresses without rebooting the vserver? 1119457963 M * Hollow Doener: what is a network context? i'm just looking at network_cmd.h and grepped the utils, but only the syscal is implemented in the lib... is this the successor to ngnet? 1119457963 Q * eXplasm2 Quit: Verlassend 1119457970 J * eXplasm ~explasm@p549F61D3.dip.t-dialin.net 1119457992 M * Doener Hollow: hm, it's the precedessor (is that the right spelling?) 1119458001 M * Pazzo hi Doener :) 1119458002 M * Hollow vorgnger? 1119458007 M * Doener yep 1119458009 M * Hollow ah 1119458017 M * Hollow but never used? 1119458018 M * Doener hi Pazzo 1119458024 M * Hollow or is it even functional atm? 1119458025 M * Doener of course it is used ;) 1119458033 M * Hollow hm 1119458042 M * Doener the whole chbind thing works with network contexts 1119458042 M * Hollow the utils dont use it ;) 1119458048 M * Hollow hm 1119458054 M * Hollow seems like i have missed sth 1119458152 M * Hollow hm.. chbind only uses the legacy api (set_ipv4root) 1119458173 M * Doener yep, but that works with network contexts as well 1119458185 M * Doener dinner time, back later 1119458189 N * Doener Doener|gone 1119459627 Q * duckx Ping timeout: 480 seconds 1119460143 M * Pazzo any idea how to extract the interface number part (2) with shell commands from /etc/vservers/test35/interfaces/2/ip ? 1119460541 J * _maharaja maharaja@ipax.at 1119460549 Q * maharaja Read error: Connection reset by peer 1119460675 N * Bertl_oO Bertl 1119460685 M * Bertl evening folks! 1119460689 M * Bertl Pazzo: huh? 1119460793 M * FaUl hey Bertl 1119460856 M * Pazzo hi Bertl! 1119460872 M * Jani Heya Bertl 1119460887 M * Pazzo Bertl: I have /etc/vservers/test35/interfaces/2/ip - could also be /somewhere/etc/vservers/test35/interfaces/2/ip 1119460900 M * Pazzo now I need to get the number of the interface (2 in this case) 1119460973 M * Pazzo with perl it would be: s/(\d+)\/ip$/$1/ 1119461044 M * Pazzo (can't use "cut" as it is not sure how many slashes will be there - and I'm too stupid to tell "sed" to do this regexp) 1119461130 M * Bertl ah, you can do that with bash 1119461210 M * Pazzo ? 1119461221 M * Bertl x=/somewhere/etc/vservers/test35/interfaces/2/ip 1119461231 M * Bertl y=${x%/*} 1119461252 M * Bertl z=${y##*/} 1119461279 M * Pazzo hehe - great, thank you! 1119461295 M * Bertl yw, btw with sed it would be something like this: 1119461371 M * Bertl echo $x | sed 's|^.*/\([^/]*\)/[^/]*$|\1|' 1119461576 M * Pazzo thnx bertl - i got it (more or less) now :-) 1119461585 M * Pazzo echo x | sed 's|^.*\([0-9]\)/ip$|\1|' 1119461593 M * Pazzo s/x/$x/ 1119461647 M * Bertl yeah, probably easier :) 1119461698 M * Pazzo or ${##x%/*} 1119461712 M * Pazzo but no idea what this is doing :-) 1119461726 M * Pazzo ehm ... no 1119461730 M * Pazzo sorry, doesn't work 1119461838 M * Pazzo what exactly does y=${x%/*} do? 1119461956 M * Pazzo and - btw: why do I need to add NET_BROADCAST bcap AND an interface entry for the broadcast address? 1119462120 Q * sukria Quit: Leaving 1119462181 M * Bertl %, %% as well as # and ## remove patterns at the beginning or end 1119462204 M * Bertl okay, off again, back later ... 1119462210 N * Bertl Bertl_oO 1119462211 M * Pazzo thnx bertl! cya! 1119462609 J * duckx ~Duck@mna75-1-81-57-39-234.fbx.proxad.net 1119462681 Q * rs Quit: rs 1119462804 J * erwan_ho ~erwan@konilope.dyndns.org 1119464298 J * flock ~restless@l192-117-111-12.broadband.actcom.net.il 1119464810 Q * erwan_ho Remote host closed the connection 1119464973 J * erwan_ho ~erwan@konilope.dyndns.org 1119465944 J * rs ~rs@212.43.230.5 1119466191 Q * flock Remote host closed the connection 1119466249 J * flock ~restless@l192-117-111-12.broadband.actcom.net.il 1119467110 Q * rs Read error: Connection reset by peer 1119467281 M * id is there a way to fix gethostbyname() iside a vserver ? 1119467597 N * Doener|gone Doener_zZz 1119467776 J * rs ~rs@212.43.230.5 1119467800 Q * rs Quit: 1119468305 Q * eXplasm Remote host closed the connection 1119469810 Q * DaCa Ping timeout: 480 seconds 1119469999 J * newz2000 ~newz2000@12-226-91-204.client.mchsi.com 1119470024 M * newz2000 hey guys, quick bash question... 1119470053 M * newz2000 I'm searching html|php files in my web directory for pages that contain the term "CYCU" using this command: 1119470064 M * newz2000 for i in `find ./ | egrep '(htm|php)$'`; do grep -li CYCU "$i"; done 1119470089 M * newz2000 However, sometimes the file name I'm searching has a space in it so $i might be something like, 'test pages' 1119470136 M * newz2000 bash thinks I'm searching two files, test and pages, neither of which exist. Is there some way to keep spaces in $i from messing things up? 1119470949 M * Pazzo id: fix gethostbyname() ?? 1119471008 M * Pazzo newz2000: find ./ -name "*.php" -exec grep -H CYCU {} \; 1119471091 M * Pazzo ehm... if you need only the filenames do grep -l 1119471095 M * newz2000 Thanks! I need to learn all of the options for find. 1119471109 M * Pazzo it has a lot of them ;) 1119471111 M * newz2000 yeah, that's what I'm doing. 1119471117 M * newz2000 Yeah 1119471151 M * newz2000 Can you use a regex for -name? 1119471203 M * newz2000 oh, -path is what I need 1119471216 M * newz2000 hm. 1119471279 J * bbkenshin ~bbkenshin@d213-103-216-151.cust.tele2.fr 1119471285 M * bbkenshin hi ! 1119471293 M * bbkenshin is there anybody here 1119471303 M * newz2000 Pazzo: I got it. Thanks for your help! 1119471307 M * Pazzo np 1119471315 A * Pazzo hides 1119471330 M * bbkenshin i'm configuring samba 1119471360 M * bbkenshin and i would like to find a websit to fint every option of samba 1119471381 M * newz2000 bbkenshin, have you used swat? 1119471394 M * bbkenshin yeah a litte partition swat 1119471415 M * newz2000 It gives you a web interface to samba config with just about every option imaginable configurable through a web browser. 1119471455 M * bbkenshin mmmmmmmm okok , but iwould like to config it withh console 1119471470 M * bbkenshin swat is a module ? 1119471493 M * Pazzo newz2000: you could use swat as a reference - it explains all the possible configuration switches 1119471508 M * bbkenshin kikou toi 1119471508 M * Pazzo http://us1.samba.org/samba/docs/man/Samba-HOWTO-Collection/ is also a good start point 1119471514 M * newz2000 good point. Swat is useful 1119471526 M * bbkenshin ok thnaks i'll see it now 1119471540 M * newz2000 bbkenshin, what distro are you using? 1119471554 M * bbkenshin mdk 10.1 1119471590 M * newz2000 hm. never used mdk, but I'll bet there's a package on the install disk for swat. Redhat, debian and most others offer it 1119471621 M * bbkenshin redhat is similar as msk dor packtages 1119471676 J * rs ~rs@212.43.230.5 1119471722 J * sannes ~ace@cm-84.118.217.070.chello.no 1119471749 Q * rs Quit: 1119472447 M * newz2000 ins't this ugly: find ./ -regex '.*[ph][ht][mp]' For some reason, find ./ -regex '.*(php|htm)' simply will not work. To make it more frustrating, have you ever tried to google for help with find -regex? 1119472455 M * newz2000 it works though. ;-) 1119472496 Q * Hollow Remote host closed the connection 1119472525 J * rs ~rs@212.43.230.5 1119472905 M * bbkenshin newz2000: what distro have u ? 1119472909 M * newz2000 ubuntu 1119472922 Q * rs Quit: rs 1119472928 M * newz2000 4.1, not the newer one, whatever its called (hoary? Warty?) 1119472941 M * newz2000 does it work on yours? 1119472960 M * bbkenshin ubuntu it's a distro ??? 1119472989 M * newz2000 Si! Been number one on distro watch for a while now. 1119473001 M * newz2000 Debian based. 1119473039 M * bbkenshin ok ok 1119473048 M * bbkenshin i don't know it 1119473071 M * newz2000 its worth a check out. 1119473107 M * bbkenshin for the moment i use MDK , maybe later redhat or debian , i don't know 1119473122 J * rs ~rs@212.43.230.5 1119473126 M * bbkenshin but mdk is very good to do servers 1119473149 M * newz2000 I tried out mdk when version 5 came out. Then it was just a custom install of redhat. 1119473186 Q * rs Quit: 1119473198 M * newz2000 I was a diehard redhat fan until about a year ago when I couldn't stand how fast fedora was becoming unsupported. 1119473199 M * bbkenshin yeah maybe , now it's a very easy install 1119473236 M * newz2000 I know a lot of people are die-hard fans of it. 1119473251 M * bbkenshin I think i'll work on MDK and when i'll know it good, maybe i'll chose another distro 1119473307 M * bbkenshin i don't know what die-hard means 1119473369 M * newz2000 it means, people who are big fans of it. 1119473385 M * bbkenshin okok 1119473411 M * bbkenshin thanks a lot 1119473437 M * bbkenshin every days i use linux and every day a love more and more linux 1119473455 M * newz2000 Did you recently switch from Windows? 1119473501 M * bbkenshin no i've installed it 2 years ago 1119473526 M * bbkenshin and last weed i've switch on windows and i was lost ............... lol 1119473584 M * newz2000 I use both equally. I find each has their benefits. 1119473627 M * newz2000 I still need to use MS Access and VB for some of my clients, and photoshop doesn't work that great in wine yet, so it will be a while before I switch completely. 1119473672 M * bbkenshin i've found windows XP pro LSD III 1119473689 M * newz2000 LSD III? 1119473707 M * bbkenshin if i stay on windows it's jsut for spcific progs 1119473713 M * bbkenshin yeah LSD III 1119473728 M * newz2000 never heard of it 1119473750 M * bbkenshin it's a windows XP configured 1119473759 M * bbkenshin and it takes less RAM 1119473778 M * bbkenshin spyware have been deleted 1119473782 J * Aiken ~james@tooax6-158.dialup.optusnet.com.au 1119473789 M * bbkenshin and other programs, 1119473810 M * newz2000 I see. 1119473812 M * bbkenshin when u've installed it, it works with 70Mo memory 1119473837 M * bbkenshin i've installed it on a Pentium 2 300Mhz 1119473845 M * newz2000 Interesting. 1119473848 M * bbkenshin and it works so good, 1119473863 M * bbkenshin my pc has got just 96Mo in memory 1119474444 Q * erwan_ho Remote host closed the connection 1119474506 J * rs ~rs@212.43.230.5 1119475240 J * mugwump ~samv@210-54-92-184.ipnets.xtra.co.nz 1119475756 Q * bbkenshin Remote host closed the connection 1119476350 Q * rs Quit: rs 1119476642 J * DaCa ~danny@mail.limehouse.org 1119476770 J * rs ~rs@212.43.230.5 1119476861 Q * rs Quit: 1119477500 J * DigitallyStoned ~ds@64.178.241.26 1119477520 M * brc anyone usnig memory limit ? 1119478725 Q * newz2000 Quit: Chatzilla 0.9.68a [Firefox 1.0.4/20050511] 1119478956 Q * duckx Quit: Client exiting 1119480926 M * pusling Hi! I do not actually know if this is vserver related, but my kernel with vserver writes some funny stuff on boot multiple times: http://pastebin.com/302713 1119481919 N * Bertl_oO Bertl 1119481932 M * Bertl evening folks! 1119481970 M * Jani Heya 1119482001 M * Bertl pusling: nope, interesting but unrelated ... 1119482027 M * Bertl pusling: sounds like swsupsend or something (hot)plug related to me ... 1119482039 M * Bertl Hey Jani! 1119482547 M * Aiken I have narrowed my segfault problem with dietlibc + /usr/lib/util-vserver/lockfile to http://pastebin.com/302723 1119482560 M * Aiken it looks like it is a dietlibc problem 1119482585 M * Bertl yeah, looked like it last time too ... 1119482601 M * Bertl but I guess with this simple test, we can fix it too :) 1119482622 M * Bertl I assume this provokes the issue? 1119482668 M * Aiken lockfile gets to the while(getppid()=ppid) line fine then segfaults when it is sent the SIGHUP 1119482699 M * Aiken with the test code if I trap SIGQUIT ro SIGTERM everything is fine, only segfaults with SIGHUP 1119482727 M * Aiken everything is fine when compiled against glibc 1119482737 M * Bertl okay, so what do I do with this code and/or how do you test? 1119482780 M * Aiken diet gcc tsig.c && ./a.out then I hit a.out with kill -HUP 1119482895 M * Bertl Segmentation fault 1119482906 M * Bertl okay, looks good ... 1119482945 M * Aiken I had put a printf just before that while loop and another in quitFunc. The printf in quitfunc is never executed. That is when I get the segfault 1119482957 M * Aiken then did that testcase 1119483069 M * brc bertl 1119483083 M * brc how does memory limitting work ? 1119483085 M * brc does it apply to swap ? 1119483102 M * Bertl yes and no ... you limit RSS and VM 1119483115 M * Bertl the host system will take care about swapping ... 1119483133 M * Bertl (in the future, over soft limit will penalize the context) 1119483189 M * brc hmm cool 1119483192 M * brc and if the user exceeds his limit 1119483197 M * brc the processes are killed, is that it ? 1119483212 M * Bertl either that or they get out of memory replies 1119483241 M * Bertl basically the context becomes unable to fork/allocate more memory 1119483320 M * brc itneresting 1119483326 M * brc i just dont get the swapping thing 1119483328 M * brc how does it work 1119483333 M * brc if he uses all his real memory limit 1119483336 M * brc then it starts swapping ? 1119483357 M * Bertl no, it starts returning out of memory 1119483378 M * Bertl as I said, the 'swaping' is handled on the host ... 1119483392 M * Bertl so you can not control 'when' the host starts to swap out 1119483411 M * Bertl (which is actually a good thing, as you do not want to swap if there is still memory left) 1119483553 M * brc good 1119483556 M * brc so it means 1119483561 M * brc if i limite a user to 64MB ram 1119483569 M * brc he can use his 64 of real memory plus some swap space 1119483582 M * brc is that it? 1119483586 M * Bertl he can use up to 64M of resident memory 1119483608 M * Bertl the swap might or might not be available, depends on the host system 1119483672 M * brc cool 1119483685 M * Bertl well, it's not perfect yet ... 1119483708 M * brc do i limit that on rlimits/rss.hard ? 1119483744 M * Bertl in the future we might evict some pages for the guest (or even better, mark them swapped out :) so that it doesn't get OOM if pages can be removed ... 1119483758 M * Bertl yes, rlimit/rss is the one ... 1119483781 M * Bertl Aiken: it seems that the dummy while loop causes the segfault ... 1119483861 M * brc Bertl: liked that a lot 1119483861 M * brc btw 1119483868 M * Aiken but it works as expected when SIGQUIT and SIGTERM are used 1119483869 M * brc thanks god vdlimit is working fine on / 1119483869 M * brc :) 1119483889 M * brc on 3 servers 1119483899 M * Bertl Aiken: yes, it's related to the signal handler 1119483930 M * Aiken what about the while that lockfile stops at? 1119484019 M * Aiken lockfile stops at a while loop so I did the my while for somthing simillar 1119484030 M * Bertl yeah, that's fine ... 1119484048 M * Bertl I'm not sure I actually 'see' what happens (with gdb) 1119484194 M * Bertl brc: good! 1119484243 M * Aiken that same test case does what it should on x86 1119484258 M * Bertl in gdb, I see the following: 1119484269 M * Bertl Starting program: a.out 1119484273 M * Bertl Program received signal SIGHUP, Hangup. 1119484277 M * Bertl main () at tsig.c:18 1119484281 M * Bertl 18 while (4 < 5); 1119484288 M * Bertl (gdb) disp/i $pc 1119484288 M * Bertl 1: x/i $pc 0x1200001a0 :br0x1200001a0 1119484304 M * Bertl if I single step from here ... 1119484310 M * Bertl (gdb) s 1119484313 M * Bertl Program received signal SIGSEGV, Segmentation fault. 1119484316 M * Bertl main () at tsig.c:18 1119484319 M * Bertl 18 while (4 < 5); 1119484322 M * Bertl 1: x/i $pc 0x1200001a0 :br0x1200001a0 1119484333 M * Bertl which is not what actually happens ... 1119484577 M * Aiken so why does it work if the signal is changed to SIGQUIT or SIGTERM? 1119484584 M * Aiken that is what I don't understand 1119484623 M * brc bertl, last question, is there any file on /proc that would show me how much bandwith traffic the user has used ? 1119484643 M * Aiken as far as the tools go with starting/stoping a vserver this segfaulting is my only problem 1119484660 M * Bertl brc: bandwidth traffic being? 1119484669 M * brc Bertl: bandwith monitor 1119484693 M * Bertl something showing 'my band is 2 inches wide'? 1119484695 M * brc i dunno, read somewhere there was a file on /proc which would show how much the user has sent/received 1119484702 M * brc something showing 1119484715 M * brc this vserver has uploaded 1000 bytes and download 30000 bytes 1119484737 M * Bertl you have a send/receive accounting for sockets 1119484768 M * Bertl if you want to have packet accounting for the networking, you ahve to do it on the host 1119484778 M * Bertl (i.e. with iptables) 1119484787 M * brc i am actually monitoring the network traffic with iptables 1119484799 M * brc logging packets that are going in/out