1109808097 M * Bertl do we have that as FAQ yet? if not, I guess I have a job for hillct (if he really insists in doing it *hint* see topic *hint* ;) 1109808121 Q * Nik Ping timeout: 480 seconds 1109808210 M * Doener attrs? hm, what could that be? "alternative totally troublesome reiserfs settings"? 1109808236 M * Bertl ;) 1109808294 M * Bertl http://vserver.13thfloor.at/Experimental/FOR-1.9.5/delta-xfs_dinode-fix01.diff 1109808330 M * Doener damn, why is that packet not delivered to the socket?! 1109808352 M * Bertl turn on debugging ... it'll probably tell 'ya 1109808405 M * Doener no, i've already added 10 or 20 additional more or less useful vxdprintks 1109808422 M * Doener s/packet/skb/ 1109808462 M * Bertl I'm going to investigate the x25 issue right now, if we are both still awake after that, we spend some time together on ngnet, okay? 1109808491 M * Doener fine 1109808535 M * Bertl hillct: you did realize that we already fixed the issue? (just checking) 1109808812 M * hillct saw that 1109808824 M * hillct do I need to pull from CVS to get it? 1109808841 A * hillct was working on other things 1109808849 M * Doener 01:04:54 Bertl http://vserver.13thfloor.at/Experimental/FOR-1.9.5/delta-xfs_dinode-fix01.diff 1109808858 M * Bertl hmm, I'm inclined to say yes here .. just to see how you do that ;) 1109808879 M * hillct no CVS? 1109808888 M * hillct come on... 1109808902 M * Doener no cvs, no svn, no arch, no bk... just Bertl ;) 1109808907 M * Bertl what would I use it for? 1109808919 M * hillct granted I should have scrolled back :) 1109808937 M * hillct would you like an SVN repository? 1109808969 M * Bertl if you volunteer to check in all my various patches and different branches, sure ;) 1109808977 M * hillct heh 1109808985 M * hillct initially, perhaps... 1109809009 M * Bertl well, I guess I could live with that once my release scripts are done 1109809012 M * hillct has mikeb or baptiste been in here recently? 1109809036 M * hillct he's got the SVN setup that I was going to volunteer 1109809058 M * hillct if not, I'll be putting one up shortly 1109809059 M * Doener using svn for kernel patches is pretty hard i guess... 1109809065 M * hillct ? 1109809068 M * hillct how so? 1109809083 M * Bertl wouldn't arch be better there? 1109809097 M * hillct svn > cvs 1109809104 M * hillct svn > * 1109809130 M * hillct you can have branches just like any other versioning system 1109809165 M * Bertl yes, the question is mor the joins or merges between different branches 1109809185 M * hillct they work as expected... 1109809217 M * Bertl maybe if you got some time (initially) you could show us how to do that properly? 1109809231 M * hillct sure 1109809245 M * hillct it'll be a month or two before things calm down here 1109809248 M * hillct but sure 1109809249 M * Doener having vanilla stuff in there, too would make the repository huge, having just the patches make merging quite hard (i guess) 1109809271 M * hillct they're text files 1109809278 M * hillct they aren't that huge 1109809285 M * Bertl Doener: let's do a leap of faith here and check it out (for some time in parallel) 1109809311 M * Bertl if hillct shows up in a month or so and helps us with the 0815 issues ... 1109809335 M * Bertl it should be not much more work than right now, right? 1109809369 M * hillct how are you tracking things now? 1109809379 M * Bertl (guess a lot of issues have to be resolved until it would/could work, but I'm willing to try) 1109809408 M * hillct Bertl how are you handling patches sent in by other developers? 1109809414 M * Bertl hillct: it's pretty simple, I have about 1000 kernel trees here, all hardlinked ... 1109809428 M * hillct eek 1109809433 M * Bertl when I get some input, i.e. patch 1109809454 M * Bertl then I check what the patch does, and apply it to the appropriate 'tree' 1109809491 M * hillct ah 1109809493 M * Bertl btw, this hardlink forrest is an important part of my development ... so that is something which has to be available via svn too 1109809527 M * Bertl (otherwise all operations done will take too long) 1109809548 M * hillct well, when you migrate to any version control system, there are always changes to development processes to allow you to take full advantage of the system 1109809551 M * Doener svn does some kind of hardlinking internally... but i doubt you get hardlinked trees outside of svn 1109809575 M * hillct it's most likely operate as follows: 1109809580 M * hillct checkout a branch 1109809590 M * hillct apply accepted patch 1109809591 M * hillct build 1109809594 M * hillct test 1109809601 M * hillct merge 1109809604 M * hillct checkin 1109809614 M * Bertl yeah, I know the svn basics ... 1109809642 M * Bertl but for example, when I try to narrow down an issue, via binary search ... 1109809653 M * hillct point is, you'd probably want to change your development process slightly to take full advantage of it 1109809658 M * Bertl I'm compiling a bunch of kernels (partially in parallel) 1109809676 M * Bertl this is quite fast as most files are 'shared' 1109809681 M * hillct I see 1109809713 M * hillct I was involved in the migration at Nortel Networks, from a home grown version control system, to clearcase. 1109809722 M * hillct It was a painful process 1109809729 M * hillct and for soem developers it wasn't worth it 1109809754 M * hillct so svn may be worth a shot and you may find you were more efficient without it 1109809767 M * hillct you never know... 1109809769 M * Bertl yes, I'm willing to try that ... 1109809831 M * Bertl if it turns out to be inefficient for my purpose, I can always keep the svn repository for everybody else around, and continue to use my 'working setup' ... 1109809878 M * Bertl (I guess a central archive of all patches (in active/web) form would be a good thing anyways) 1109809902 M * Bertl and for this svn is probably the best solution ... 1109809922 M * Bertl (folks could build their 'deltas' themselves) 1109809957 M * hillct svn is really cool that way 1109809959 M * Bertl okay, enough on this topic for today, please let me know when you got some time to give it a try ... 1109809971 M * hillct you can generate patches from any version X to any version Y 1109809979 M * hillct yah 1109809980 M * Bertl your help will be appreciated! 1109809984 A * hillct builds kernel 1109810491 Q * mcp Read error: Operation timed out 1109810518 M * Doener hm, i may be getting somewhere slowly 1109810565 A * Doener .oO( i should find someone to clean up the 'debug' mess i've created on the way... ) 1109810605 M * Bertl feel free to submit patches which add useful debug stuff for ngnet 1109810662 M * Doener nothing useful there ;) 1109810690 M * Doener just stupid messages that allow me to find out which way the skb goes through the kernel 1109810885 J * chairuou ~chairuou@210.245.68.196 1109810923 M * jd86 Bertl: is there a util-vserver usage page? meaning all my options and explanations? 1109810945 M * hillct LOL 1109810961 M * Bertl jd86: vserver --help 1109810970 M * jd86 Bertl: yea but i was thinking more eloborate 1109810972 M * Bertl and for a specific topic, do: 1109810974 M * jd86 but i realize theres one of them man pages 1109810979 M * jd86 but i didnt realize that it was there. 1109810981 M * Bertl vserver - --help 1109810982 M * jd86 silly me! 1109810983 M * jd86 ok 1109810990 M * jd86 see that didnt seem to work but i guess i didnt have the - 1109811024 A * hillct watches kernel compile remarkably slowly 1109811027 M * hillct geez 1109811056 M * hillct This is a dual opteron 1.8ghz box with 8GB memory. should be faster than this 1109811071 M * jd86 hillct: patience. 1109811098 A * hillct resists the urge to buy another $4 cup of coffee 1109811162 M * daniel_hozac and i thought the $.4 cups of coffee were a rip off. 1109811248 M * hillct this is why working from starbucks is a bad idea 1109811279 M * jd86 why? 1109811294 M * jd86 you think the verbose output is the problem? 1109811294 M * jd86 lol 1109811810 M * hillct It's the IO ratio that's the problem 1109811898 M * daniel_hozac with that amount of RAM, you could just have the kernel trees on a ramdisk ;) 1109811923 M * Bertl ah, found the x25 issue ;) 1109812160 Q * alexx Read error: Operation timed out 1109812565 M * Bertl Doener: where was/is your latest patch collection for ngnet? 1109812589 M * Doener http://doener.homeip.net/doener/vserver/diff-2.6.11-rc5-vs1.9.4.9-ng9.1-various.diff 1109812592 M * Doener and the one i mailed 1109812595 M * Bertl thanks a lot! 1109812607 M * Doener everything else is not interesting/usable 1109812623 M * Doener i'm still trying to figure out how it works 1109812837 M * Doener not sure if any of the changes in the 'various' patch are actually good ;) 1109813136 P * hillct 1109813304 M * Bertl Doener: what is the rationale behind the change from 0 -> VNET_UNTAGGED ? 1109813306 M * Bertl + skb->nfvnet = VNET_UNTAGGED; 1109813312 M * Bertl - if ((*pskb)->nfvnet) 1109813312 M * Bertl + if ((*pskb)->nfvnet != VNET_UNTAGGED) 1109813520 M * Doener that was because it a) seemed more logical to me and b) there's been some problem with packets not getting tagged if the traffic is local 1109813553 M * Bertl hmm, local as in 'lo' local? 1109813565 M * Doener local as in host<->guest 1109813605 M * Bertl hmm, so host <-> guest traffic is already working to some extend? 1109813623 M * Doener in my current setup, a ping from host to guest shows up as: request and reply on guest's en0, reply on host's eth0 1109813644 M * Doener but the ping process does not receive the reply... 1109813661 M * Doener probably because i'm doing some strange kind of loopback action on eth0 :) 1109813665 M * Bertl okay, let's delay that, I basically agree with the change ... 1109813677 M * Bertl regarding rtnetlink_rcv() 1109813700 M * Bertl would it be way better to use the sock's xid instead of current? 1109813793 M * Doener from looking at the code (and feeling a little lost) i assumed that the sk get's its xid from the dev 1109813797 M * Bertl maybe something like the vx_tag_output_skb() just for input? 1109813799 M * Doener and the dev has xid=0 1109813821 M * Doener now that you mention that function, i actually tried just using that one :) 1109813830 M * Doener so using the sk's xid is no good 1109813839 M * Bertl hmm .. 1109813857 M * Bertl shouldn't every user have his own socket to the netlink? 1109813888 M * Bertl I mean we do many netlink related checks based on the xid ... (current xid that is) 1109813902 M * Bertl could that be based on the sock's xid? 1109813919 J * ntrs ~ntrs@Dardeene-68.188.50.87.charter-stl.com 1109813922 M * Bertl (assumed that we assign the right xid to the socket that is) 1109813978 M * Bertl wb ntrs! 1109813992 M * Bertl Doener: or is there just 'one' socket for netlink? 1109814020 M * Doener the sk has xid=0, i checked that 1109814052 M * Bertl hmm hmm ... okay, I'll add some debug stuff there and check if that is the same socket and where it comes from 1109814100 M * Doener global rtnl 1109814140 M * Doener i guess that's the socket we're after ;) 1109814153 Q * ntrs_ Ping timeout: 480 seconds 1109814198 M * Bertl why is it passed via sock *sk then? 1109814212 M * Bertl (the socket I mean) 1109814293 M * Doener because the networking framework requires that? 1109814311 M * Bertl hmm, good point ... 1109814440 J * alexx ~alexx@84.4.10.48 1109814447 M * Bertl welcome alexx! 1109814504 Q * ntrs Read error: Operation timed out 1109814608 M * Doener look at rtnetlink_rcv...do { ... while ((skb = skb_dequeue(&sk->sk_receive_queue)) ... } while (rtnl && rtnl->sk_receive_queue.qlen); 1109814624 M * Doener so i guess rtnl == sk 1109814734 Q * chairuou Read error: Connection reset by peer 1109814876 J * Nik ~Nik@cable-153-130.online.bg 1109814892 M * Bertl right it's the one and only socket for netlink ... 1109814905 M * Bertl so using current_xid() is perfectly fine there ... 1109814959 M * Doener ok 1109814988 J * ntrs ntrs@Dardeene-68.188.50.87.charter-stl.com 1109815068 M * Bertl hmm, one issue I see here is that the 'destroy devs' stuff strikes even when the context is still active, did you observe similar? 1109815174 M * Bertl hmm, no actually my bad ;) 1109815195 M * Doener only had a dev that wasn't freed once 1109815198 M * alexx hi Bertl 1109815251 M * Bertl yeah, my fancy init script kind of backfired (doing it's job), and I thought if I can see en0, it has to be the 'right' context ;) 1109815543 M * Bertl okay, I'm looking into the 'packet from #65535 on vnet ...' issue right now ... 1109815611 Q * Nik Ping timeout: 480 seconds 1109815908 J * Nik ~Nik@cable-153-130.online.bg 1109815917 Q * Nik Quit: 1109817105 J * hillct ~hillct@client200-5.dsl.intrex.net 1109817933 M * Doener Bertl: any success yet? i didn't come up with much in that direction (nothing at all actually ;) 1109817993 M * Bertl it seems that they are caused by some strange 1109817994 M * Bertl Ethernet portion of AoE driver 1109818014 M * Bertl which has a timer and starts to broadcast stuff every now and then? 1109818036 M * Doener AoE? 1109818128 Q * monrad Quit: Leaving 1109818339 M * Bertl yeah, no idea where it actually comes from ... 1109818353 A * Bertl doesn't remember to have selected it ;) 1109818361 M * Bertl *having 1109818407 M * Doener what is AoE? i remember a game from MS called AoE, but that's it ;) 1109818414 M * Bertl added a dump_stack() at the debug location ... 1109818425 M * Bertl 80237975: aoenet_xmit drivers/block/aoe/aoenet.c:100 1109818425 M * Bertl 80236ef4: aoecmd_cfg drivers/block/aoe/aoecmd.c:520 1109818425 M * Bertl 80237600: discover_timer drivers/block/aoe/aoemain.c:21 1109818425 M * Bertl 80237713: discover_timer drivers/block/aoe/aoemain.c:44 1109818425 M * Bertl 80121093: run_timer_softirq kernel/timer.c:472 1109818436 M * Doener ata over ether? 1109818450 M * Bertl maybe, yep that could be it 1109818519 M * Bertl CONFIG_ATA_OVER_ETH=y *bingo* 1109818568 M * Bertl now let's see if we can catch other broadcasters ... 1109818687 M * hillct Bertlthe patch is looking good 1109818706 M * hillct thanks for your help 1109818707 M * Bertl excellent! so everything working now as expected? 1109818719 M * hillct so far so good 1109818729 M * Bertl was a pleasure to be of assistance! 1109818759 M * hillct what happened to the v_* init scripts between tool 0.29 ant toold 0.30 ? 1109818769 M * hillct er tools 1109818778 M * Bertl they should be still there ... 1109818783 M * hillct hmm 1109818814 M * hillct I now have only: 1109818816 M * hillct vprocunhide 1109818825 M * hillct vservers-default 1109818826 M * Bertl rpm -ql util-vserver-legacy 1109818826 M * Bertl /etc/rc.d/init.d/rebootmgr 1109818826 M * Bertl /etc/rc.d/init.d/v_gated 1109818826 M * Bertl /etc/rc.d/init.d/v_httpd 1109818826 M * Bertl /etc/rc.d/init.d/v_named 1109818830 M * Bertl ... 1109818835 M * hillct none of those... 1109818841 M * hillct ah 1109818846 M * hillct that's in legacy? 1109818850 M * hillct hmm 1109818850 M * Bertl yep 1109818875 M * hillct how is it handled going forward? 1109818898 A * hillct tries to avoid legacy implementations for the moment 1109818910 M * Bertl the thing is the future networking (and those scripts only affected networking) will be ngnet 1109818919 M * Bertl (the stuff we are currently hacking on ;) 1109818936 M * hillct ah 1109818941 M * Bertl and with that, the services on the host will not require restraining ... 1109818955 M * hillct I see 1109818967 M * hillct what's the status of ngnet at this point? 1109818977 M * Bertl so basically the entire networking stuff is considered legacy atm ... ;) 1109818985 M * hillct yah 1109819008 M * Bertl ngnet allows to have separate interfaces, routing and most (maybe all now, thanks doener) netlink stuff 1109819023 M * Bertl looks like this: 1109819041 M * Bertl chcontext --xid 10 ip addr ls 1109819045 M * Bertl 4: lo: mtu 16436 qdisc pfifo_fast qlen 10 1109819046 M * Bertl link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00 1109819046 M * Bertl inet 127.0.0.1/8 scope host lo 1109819049 M * Bertl 5: en0: mtu 1500 qdisc pfifo_fast qlen 10 1109819050 M * Bertl link/ether 52:54:00:12:34:56 brd 00:00:00:00:00:00 1109819050 M * Bertl inet 172.16.0.2/16 brd 172.16.255.255 scope global en0 1109819081 M * Bertl while on the host: 1109819087 M * Bertl 1: eth0: mtu 1500 qdisc pfifo_fast qlen 1000 1109819087 M * Bertl link/ether 52:54:00:12:34:56 brd ff:ff:ff:ff:ff:ff 1109819087 M * Bertl inet 10.0.0.2/8 brd 10.255.255.255 scope global eth0 1109819087 M * Bertl 2: lo: mtu 16436 qdisc noqueue 1109819087 M * Bertl link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00 1109819090 M * Bertl inet 127.0.0.1/8 scope host lo 1109819092 M * Bertl 3: dummy0: mtu 1500 qdisc noop 1109819095 M * Bertl link/ether ca:82:b9:61:e3:04 brd ff:ff:ff:ff:ff:ff 1109819237 M * hillct cool 1109819347 M * Bertl Doener: now what is what you do on the host to make it respond to arp requests? 1109819390 M * Doener atm, i do: ip r a local 172.16.0.2 dev eth0 1109819460 M * Bertl hmm, but that seems not sufficient for a ping from outside ... 1109819502 M * Bertl (the lookup/arp of 172.16.0.1 fails here) 1109819534 M * Bertl hmm, maybe my fault ... sec checking 1109819557 M * Doener is enough for a ssh connection for me ;) 1109819637 M * Doener 04:13:28.798118 arp who-has 172.16.0.2 tell 172.16.0.1 1109819637 M * Doener 04:13:28.984059 arp reply 172.16.0.2 is-at 52:54:00:12:34:56 1109819651 M * hillct because I imagine someone has a script lying around that does this, I figure I'd ask 1109819659 M * Bertl yep, looks like an issue with my prep script 1109819685 M * hillct is there a script that provides total directory size that completely excludes hard linked files? 1109819706 M * hillct du does it natively only if the hard linked target is within the tree being tested 1109819731 M * Bertl hehe, good thinking ... but not that I know of, and it would not do what you want to do, at least not perfectly ... 1109819771 M * Bertl but IIRC, somebody modified du and other tools to 'know' about unification and to calc it properly 1109819855 M * Doener hillct: vdu 1109819864 M * hillct I'll just write a perl wrapper for it. I dislike the idea of modifying core system tools for this sort of thing 1109819865 M * hillct oh 1109819865 M * Doener comes at least with 0.30.204 1109819867 M * hillct cool 1109819874 M * hillct thanks 1109819876 M * Bertl really? cool! 1109819880 M * Doener vdu directory ... 1109819880 M * Doener Compute the size of a directory tree, ignoring files 1109819880 M * Doener with more than one link. 1109819897 M * hillct hmm 1109819911 A * hillct tests 1109819913 M * Bertl hmm, well, not perfect, but probably fine ... 1109819937 M * Bertl s/fine/sufficient/ 1109819957 M * hillct LOL 1109819969 M * hillct [root@laptop vservers]# vdu fourth/ 1109819969 M * hillct Can't open directory fourth//dev (No such file or directory) 1109819969 M * hillct Can't open directory fourth//etc (No such file or directory) 1109819969 M * hillct Can't open directory fourth//proc (No such file or directory) 1109819969 M * hillct Can't open directory fourth//bin (No such file or directory) 1109819970 M * hillct Can't open directory fourth//lib (No such file or directory) 1109819970 M * hillct Can't open directory fourth//lib64 (No such file or directory) 1109819972 M * hillct Can't open directory fourth//sbin (No such file or directory) 1109819972 M * hillct Can't open directory fourth//usr (No such file or directory) 1109819974 M * hillct Can't open directory fourth//home (No such file or directory) 1109819974 M * hillct Can't open directory fourth//root (No such file or directory) 1109819976 M * hillct Can't open directory fourth//tmp (No such file or directory) 1109819976 M * hillct Can't open directory fourth//var (No such file or directory) 1109819978 M * hillct fourth/ 0K 1109819978 M * hillct [root@laptop vservers]# 1109819988 M * hillct it didn't like that much 1109820027 M * Bertl maybe an xfs issue ... 1109820037 M * hillct perhaps 1109820068 M * Doener hm, no seems it does not work on subdirectories 1109820081 M * Doener doener ~ # vdu . 1109820081 M * Doener Can't open directory ./.loki/rt2 (No such file or directory) 1109820088 M * hillct [root@laptop vservers]# vdu --help 1109820088 M * hillct Can't open directory --help (No such file or directory) 1109820088 M * hillct --help 0K 1109820088 M * hillct [root@laptop vservers]# 1109820100 M * Doener (on ext3) 1109820216 M * hillct it just reports OK, for everything. It doesn't take any of the standard du args 1109820219 M * hillct hmm 1109820226 A * hillct writes his own 1109820242 M * Bertl Doener: I did a ping from outside to the guest, after adding the route 1109820258 M * Bertl the ping is answered as expected? ... 1109820278 M * Bertl what I really wonder about is the last few actions 1109820295 M * Bertl vxD: neigh_create(803008c0,839b2c00[en0,#10]) : 010010ac 1109820295 M * Bertl vxD: __vx_fib_get_table(255,#10) = 83c83d20 1109820295 M * Bertl vxD: inet_addr_type(172.16.0.1, #10) 1109820295 M * Bertl vxD: ip_route_output_slow() = 0 1109820295 M * Bertl vxD: vnet_get(101) = 839b2e40[101,#10] 1109820298 M * Bertl vxD: __vx_fib_get_table(255,#0) = 81105e80 1109820301 M * Bertl vxD: inet_addr_type(172.16.0.1, #0) 1109820303 M * daniel_hozac vdu really does look broken. 1109820303 M * Bertl vxD: neigh_lookup(803008c0,81158000[eth0,#0]) : 010010ac 1109820306 M * Bertl vxD: __neigh_lookup(803008c0,81158000[eth0,#0],1) : 83281b80 1109820308 M * Bertl vxD: vnet_real_xmit(83d4bb60) [101,#10] 1109820311 M * Bertl vxD: netif_receive_skb(83d4bc20[#10]) = 2 1109820343 M * Bertl Doener: especially the neigh lookup in xid=0 looks suspicious to me ... shouldn't that happen in xid=10? 1109820572 M * hillct this looks like it's a good starting point http://mirrors.paul.sladen.org/sam.vilain.net/vserver/unify-dirs 1109820641 M * Bertl definitely, but look at the tools too, there is a vunify (or should be) 1109820685 M * Doener Bertl: what's before neigh_create? 1109820688 M * hillct there is. looks neat, but all I want is to find out how much disk space each of my vservers is uniquely using 1109820707 M * hillct excluding the links 1109820746 M * Bertl excluding unified links that is 1109820759 M * hillct yah 1109820780 M * Doener ah, neigh_lookup it is 1109820795 M * hillct which as far as I'm concerned is any hard link where the link target is outside the scanned tree 1109820827 M * Bertl no, which is any link which has iunlink + immutable set ;) 1109820838 M * hillct true 1109820847 M * hillct the former is easier to test though 1109820854 M * hillct and faster to test 1109820863 M * Bertl but not necessarily correct 1109820870 M * daniel_hozac do hardlinks even have the concept of a target? 1109820874 M * hillct expecience over accuracy 1109820889 M * hillct is ok on occasion 1109820902 M * Bertl daniel_hozac: no, all 'hard links' are created equal ;) 1109820912 M * hillct yah 1109820915 M * hillct hmm 1109820925 M * hillct maybe harder to test 1109820947 A * hillct rips apart unify-dirs 1109820998 M * Bertl Doener: probably way to late, but here it is: http://vserver.13thfloor.at/Experimental/NGNET/ping_from_guest.txt 1109821046 M * daniel_hozac if you're using tagxid, you could just use du --xid ... ;) 1109821067 M * Bertl which would not be correct either ;) 1109821114 M * hillct what an unfortunate last name... 1109821118 M * hillct "No, mr. Rumsfeld. The name is Vilain, not Vilan!" 1109821154 M * daniel_hozac it wouldn't? i can see the cases where it wouldn't be, but generally speaking, that should be quite right, no? 1109821224 M * Bertl no, because when you create a vserver, basically all files will belong to xid=0 1109821237 M * Bertl regardless if they are unified or not 1109821261 M * Bertl most of them files will change into the xid later ... 1109821274 M * Bertl (unless you run some program to change them in the first place) 1109821298 M * daniel_hozac yeah, if you want to get usage statistics, you'd want to change the xids as early as possible. 1109821313 M * Bertl btw, daniel_hozac, do we have links to your modifications/patches from the wiki yet? 1109821345 M * Bertl if not, it would be great if you could add some ... 1109821360 M * daniel_hozac where should i add those? 1109821377 M * hillct is there any user management in that wiki, or are all edits anonemous? 1109821426 M * Bertl preferences to set your name 1109821459 M * Bertl daniel_hozac: I guess patches and tools would be a good place 1109821473 M * Bertl tools and patches that is ... 1109821477 M * Bertl http://linux-vserver.org/Tools+and+patches 1109821526 M * Bertl maybe make a new section called 'Adapted Linux Tools' ? 1109821611 M * daniel_hozac sounds sane. 1109821695 M * daniel_hozac maybe GNU tools to be PC? :) 1109821744 M * hillct bah GNU 1109821772 M * hillct Administration Tools and Scripts 1109821795 M * Bertl daniel_hozac: if you like, add a link to the FSF ;) 1109821914 M * daniel_hozac hehe 1109822093 M * daniel_hozac there, added. :) 1109822325 M * Bertl excellent, thanks! 1109822357 M * Bertl okay, guess I'm off to bed for now ... 1109822384 M * daniel_hozac night. 1109822409 M * Bertl Night everyone! night daniel_hozac! night Doener! night hillct! .... 1109822415 M * Doener night Bertl! 1109822422 N * Bertl Bertl_zZ 1109824300 Q * shuri Quit: 1109835974 J * Nik ~Nik@cable-153-130.online.bg 1109835981 M * Nik hi all 1109836466 Q * Nik Ping timeout: 480 seconds 1109837574 J * chairuou ~chairuou@210.245.69.234 1109837578 P * chairuou 1109837732 J * nox ~nox@213.39.135.248 1109837793 Q * tchan Read error: Operation timed out 1109837851 Q * sith Read error: Operation timed out 1109837940 J * tchan ~tchan@c-24-13-81-164.client.comcast.net 1109837997 J * sith sith@aaronp.com 1109840004 J * _are_ ~are@gateway-dsl.lihas.de 1109840008 M * _are_ Hi 1109840146 J * prae ~prae@ezoffice.mandrakesoft.com 1109840272 M * _are_ Hi prae 1109840326 M * _are_ hmm, if I remember right, someone created debianized util packages, any hint about a download location? 1109840344 M * prae Hi _are_ :) 1109840693 J * rs ~rs@194.98.28.2 1109840705 M * rs moin dudes 1109840751 J * erwan_taf ~erwan@81.80.43.68 1109840757 J * BWare ~bware@office.intouch.net 1109840805 M * prae erwan_taf: !! 1109840807 M * prae rs: pwet 1109840818 M * erwan_taf prae: 1109841031 M * BWare morning 1109842024 J * mcp ~hightower@www.c-tera.de 1109842349 J * alex|ikse ~alexx@82.225.136.176 1109842703 Q * alexx Ping timeout: 480 seconds 1109843192 J * Nik ~Nik@cable-153-130.online.bg 1109843676 Q * Nik Ping timeout: 480 seconds 1109845202 M * spocki http://www.openbeans.org/~mk/ 1109845232 M * spocki upps, wrong channel 1109845690 J * ruth VooDoo@pclab7.informatik.uni-ulm.de 1109846263 Q * TheSeer Remote host closed the connection 1109846491 J * TheSeer ~theseer@border.office.salesemotion.net 1109850479 J * Nik ~Nik@cable-153-130.online.bg 1109850961 Q * Nik Ping timeout: 480 seconds 1109857039 J * jsambrook ~jsambrook@host-62-69-64-93.bsve.net 1109857692 J * Nik ~Nik@cable-153-130.online.bg 1109858176 Q * Nik Ping timeout: 480 seconds 1109858919 Q * matti Ping timeout: 480 seconds 1109860036 Q * sladen Ping timeout: 480 seconds 1109861703 J * sladen paul@starsky.19inch.net 1109862199 P * ccooke 1109863705 N * Bertl_zZ Bertl 1109863710 M * Bertl morning folks! 1109863828 M * _are_ Hi Bertl 1109863834 M * erwan_taf lo Bertl 1109864196 M * prae someone has tested vserver with lvm ? 1109864220 M * Bertl yes, I'm using linux-vserver with lvm volumes (1 and 2) 1109864233 M * prae it's work perfectly ? 1109864269 M * Bertl as far as I can tell, yes (the lvm volumes are on the host of course) 1109864300 M * Bertl did you encounter any issues? 1109864402 M * prae without lvm, I encounter some issues 1109864409 M * prae with postfix 1109864423 M * prae oups! 1109864442 M * prae without vserver, I encounter some issues with postfix 1109864469 M * Bertl hmm, postfix on lvm gives issues? 1109864488 M * prae yes 1109864507 M * Bertl interesting, which kind of issues? 1109864547 M * prae I don't remember, but postfix became crazy 1109864566 M * prae some modules (rewrite, postdrop, etc...) crashed 1109864586 M * Bertl sounds strange to me ... 1109864703 M * prae me too 1109864904 J * Nik ~Nik@cable-153-130.online.bg 1109864945 M * Bertl wb Nik! 1109865183 M * hillct Bertl is there some particular documentation I should read as a primer on context IDs and quotas? 1109865202 M * hillct thought I saw something like that out there... 1109865217 M * Bertl quotas as in Disk Limits or as in 'User/group Quotas' 1109865238 M * hillct as in limits per vserver 1109865254 M * hillct disk space allocation per vserver 1109865280 M * Bertl so Disk Space Limits, right ... 1109865291 M * Bertl http://linux-vserver.org/Linux-VServer-Paper-04 (04.5) 1109865301 M * hillct ah 1109865303 M * hillct thanks 1109865353 M * Bertl http://linux-vserver.org/Quota+and+Disk+Limits 1109865375 M * Bertl (the Quota part is not available in 2.6/1.9.x yet) 1109865382 M * hillct ah 1109865386 Q * Nik Ping timeout: 480 seconds 1109865400 M * Bertl Quota as in User/group quotas that is ... 1109865416 M * hillct and if I limit it usind seperate partitions, I'm SOL for unification 1109865421 M * hillct hmm 1109865495 M * Bertl as I said, Disk Limits work out of the box with 1.9.x, just the user/group quotas are not there yet 1109865515 M * hillct oh. misread that 1109865524 M * Bertl basically everybody does ... 1109865539 M * hillct disk limits != quotas 1109865541 M * hillct K 1109865580 M * hillct what I care about are disk limits anyway 1109865591 M * Bertl precisely, so everything is fine ;) 1109865596 M * hillct hup 1109865599 M * hillct yup 1109865603 A * hillct reads 1109865609 M * hillct thanks 1109865615 M * Bertl you're welcome! 1109866092 M * hillct Bertl as of the writing of http://linux-vserver.org/Linux-VServer-Paper-04 aparently inode context ID was not implemented for XFS. does 'working out of the box' mean XFS or just ext2/3 ? 1109866148 Q * _are_ Quit: Disconnecting 1109866271 A * hillct keeps reading 1109866297 A * Doener yawns 1109866301 M * Doener morning 1109866512 M * Bertl morning Doener! 1109866535 M * Bertl hillct: hmm, that does just affect the internal taggings 1109866553 M * Bertl xfs has 32 bit uid/gid support I presume, so the UID/GID taggings should work fine 1109866567 M * hillct K 1109866576 M * hillct still getting backgound 1109866649 M * Bertl you want to select UID24/GID24 for xfs ... 1109866653 M * Bertl (well, it's the default anyways) 1109866744 J * sannes ~ace@home.skarby.no 1109866770 M * sannes :) 1109866784 M * Bertl hey sannes! 1109866789 M * Bertl LTNS! 1109866802 Q * ruth Quit: 1109866846 M * sannes yeah, things seem to happen at a rappid pace in my life these days.. 1109866869 M * Bertl is that inherently bad? 1109866892 M * sannes no, not at all, got a gf now that spend a lot of time with :) 1109866923 M * Bertl ahh, real life! well, that's something ;) 1109866949 M * sannes which is all good, and then I have a courses that take up the rest of the time, I think they've all ganged up on me to make sure I don't have any spear time.. heh 1109867058 M * sannes and now I'm sick, and, did I tell you, that I got the same panic on another machine (not mine), .. and the only thing they have in common is that they use LVM instead of LVM2 .. 1109867091 M * sannes and I saw in the changelog that there was fixed a kernel panic in dm stuff, so I might give it whirrel next weekend (easter) .. :) 1109867129 M * Bertl sick? how unfortunate! hopefully nothing critical? 1109867133 M * sannes anyways, 1.9.4 is out and 1.9.5-rc1 and ngn9 .. 1109867145 M * sannes Bertl : nothing crucial, but I have to stay in the vicinity of a toilet.. heh :> 1109867193 M * sannes anyways, got a new machine for vserver 2.6 testing which hasn't got any productions things on it.. :) 1109867226 M * sannes hey, I saw that you posted a new bind mount patch on lkml .. a week ago or so? 1109867318 M * Bertl yep, right! with a few changes (closer to the current vfs) 1109867347 M * sannes greate! :) I look forward for it to end up in mainline .. 1109867377 M * Bertl if they will not take it this time, I'll add it to linux-vserver and that's it ... 1109867396 M * sannes was there a follow up on it? (just looked very briefly at it because I was in a hurry) 1109867424 M * Bertl well, Trond was so kind to comment on a few things ... 1109867453 M * Bertl (which are fixed and posted as 0.06.1) 1109867475 M * Bertl but no general comment or statement, neither from viro nor morton yet ... 1109867553 M * Bertl of course, no user comments (like yeah, please include it right now!) either ... 1109867568 M * sannes ah, I can do that, if I didn't have lkml-phobia .. 1109867714 M * sannes :) 1109867734 M * sannes don't think everyone knows what is actually does, heh :) 1109867738 M * hillct Bertl I think I understand the theory behind disk limits now, but I don't see the relevent tools 1109867776 M * hillct I'm probably not looking in the right place 1109867790 M * Bertl the relevant tool(s) is vdlimit. 1109867794 M * hillct ah 1109867804 A * hillct was looking for mquota 1109867810 M * hillct that was for 2.4.x 1109867825 M * sannes Bertl : have you implementet per vserver quotas now? ... I'm impressed :) 1109867842 M * Bertl not for 2.6 (for 2.4 it's there since ever ;) 1109867873 M * sannes ah, heh .. was beginning to think time were experienced differntly at where you are.... hehe 1109867881 M * hillct sannes I'm playing with disk limits (not quotas, as was pointed out to me earlier) 1109867924 M * sannes ah, disk limits, hm.. ok, I'm not in the know how of what that is (was that sentence broken?) .. what is it? 1109867931 M * Bertl correct ...but unfortunately most folks do not know what 'quotas' (the traditional ones) are ... 1109867944 M * Bertl unless they explicitely need it ;) 1109868003 M * sannes maybe I've got what quotas are wrong too... my first guess is that disk limits just limit how much space can be used on a partition or mount if you like? and quota is a more per user thing? 1109868027 M * Bertl yep, that's it, per user and group 1109868053 M * sannes well, then I know what it is:) 1109868083 M * sannes so I could put a disk limit on a partition for a vserver so that they don't fill the partition completly up (less work for the filesystem?) .. 1109868101 M * Bertl yes, that's in 2.4 and 2.6 1109868129 M * hillct Bertl is vdlimit a tool (if so it's not part of 0.30.204) or just an element of the vserver config file structure? 1109868133 M * Bertl *2.4/1.2.x+quota patches and *2.6/1.9.x by default 1109868158 M * Bertl hillct: it's a tool, and it's slowly getting a part of 0.30.20x 1109868162 M * sannes ah, have been thinking that filling up a region to its max must be bad for fs performance.. 1109868213 M * Bertl well, yes, but the main point of application is to have shared (unified) vservers on the same partition 1109868244 M * Bertl hillct: http://www.13thfloor.at/vserver/d_rel26/v1.9.4/ (last link) 1109868286 M * sannes I understand that, just wondering if I could use it for something completly different.. :) but, then I remember there was a setting somewhere were I could do something similiar .. but that was space available for root and for other users.. 1109868323 M * hillct aha 1109868327 M * hillct thanks 1109868344 M * Bertl my pleasure! 1109868350 A * sannes is just ircing out loud :) 1109868423 M * Bertl Doener: I'm pretty sure that the 'add local route' thing is not what we want ... had some thoughts on the trace we did get yesterday ... 1109868454 M * Bertl what we actually want is that the arp happens _inside_ the context and the host just _relays_ it to it ... 1109868501 M * Doener Bertl: i didn't really find out what way we're going through the kernel in that trace... 1109868517 M * Doener but yep, i see that local route thing as a work around, too 1109868540 M * Doener i guess some extra vserver table would be a nice way to go 1109868561 M * Bertl the main issue with arp is the thing that you do not know where to relay a reply to ... 1109868621 M * bro hiya all :> 1109868629 M * Bertl welcome bro! 1109868715 M * bro Bertl, by any chance would u have any idea why box, which is vserver's host, cant see more than 1gb of memory? but bios shows me that i have 2gigs there 1109868764 M * bro and, when there was only 1gig it was reporting that there was only 884mbytes there ;/ 1109868785 M * Doener Bertl: why is there no vnet_start_xmit in your trace? did i add that debug output? 1109868821 M * daniel_hozac bro: default memory split and highmem disabled? 1109868826 M * Doener no, is there in 9.1... 1109868840 M * Bertl bro: sounds like two questions mixed into one 1109868863 M * bro hmmm 1109868873 M * Bertl bro: first, the question is: does you machine see _all_ the memory (which the bios and the hardware knows about) 1109868889 M * Bertl *your even, and check in the dmesg output for that 1109868906 M * Bertl (if you want, upload it somewhere, I'll have a look at it) 1109868930 M * bro bios, knoppix shows all of the memory, that is 2gigs, used memtest, went smoothly for some time, then i ran out of patience... 1109868939 M * Bertl the second question is: with 1GB of discovered memory, why do I just get 884/992 MB 1109868971 M * Bertl and the answer to that is what daniel_hozac suggested, change the memory split or use highmem ... 1109868980 M * bro omg 1109868984 M * bro okay 1109869037 M * bro tnx for the hints again, Bertl and daniel_hozac ;> 1109869067 M * Bertl sidenote: memory split changes are possible due to a patch in linux-vserve which allows to set it differently (not just 3/1 as default) 1109869095 M * Bertl (recent linux-vserver 1.9.x that is) 1109869124 J * erwan_ho ~erwan@lns-vlq-39f-81-56-133-136.adsl.proxad.net 1109869134 M * sannes Bertl : ever tried to get that into mainline? :> 1109869178 M * Bertl yes, at least twice ... was ignored! 1109869185 M * Bertl Doener: http://vserver.13thfloor.at/Experimental/NGNET/ping_from_guest.txt 1109869193 M * Bertl we are talking about that one, right? 1109869205 M * Doener yep 1109869212 M * Bertl guess the filename is misleading ... 1109869224 M * Bertl it was a ping from real -> guest 1109869224 M * Doener it's ping to guest, right? 1109869242 M * Bertl yep precisely, was half sleeping I guess ... 1109869272 M * Bertl I'll rename it to ping_real_to_guest.txt 1109869300 M * Bertl (done) 1109869769 M * Bertl Doener: I'm inclined to add the 2.6 vroot device to 1.9.5 and fix it up later, what is your opinion on that? 1109869810 M * Doener hm, what would be the 'fix it up' part? 1109869844 M * Bertl it's a basic port from 2.4 to 2.6 which lost some checks/warnings 1109869864 M * Bertl (not very well tested, but it _seems_ to work) 1109869916 M * Bertl http://vserver.13thfloor.at/Experimental/diff-2.6.10-vs1.9.3.13-vr0.21.diff 1109870352 Q * prae Read error: Operation timed out 1109870432 J * prae ~prae@ezoffice.mandrakesoft.com 1109870620 M * Doener Bertl: should be fine then... i'll spend some time with my gf and my sister... back later 1109870626 N * Doener Doener|gone 1109870658 M * Bertl cya, have fun! 1109870666 M * erwan_ho ++ 1109871412 Q * prae Quit: Client exiting 1109871570 J * prae ~prae@ezoffice.mandrakesoft.com 1109871860 J * shuri sjnesjd@dsl.speedline209.226.electronicbox.net 1109871909 Q * prae Quit: Client exiting 1109872155 J * Nik ~Nik@cable-153-130.online.bg 1109872184 M * Bertl welcome shuri! 1109872194 M * shuri hello Bertl 1109872366 Q * erwan_ho Read error: Operation timed out 1109872641 Q * Nik Ping timeout: 480 seconds 1109873491 Q * hillct Read error: Connection reset by peer 1109874089 J * erwan_ho ~erwan@lns-vlq-39f-81-56-133-136.adsl.proxad.net 1109876206 N * Doener|gone Doener 1109876214 M * Doener short visit ;) 1109876241 M * Doener Bertl: about that call to vnet_real_xmit in the trace, where does that come from? 1109876304 M * Bertl hmm, sec 1109876359 M * Bertl vxD: vnet_real_xmit(83d4bb60) [101,#10] <.. this one? 1109876363 M * Doener yep 1109876409 M * Doener vnet_start_xmit should have also emitted a message... 1109876425 M * Bertl hmm .. good point! 1109876435 M * Bertl ah, that has: 1109876437 M * Bertl vxdprintk(VXD_CBIT(net, 5), 1109876442 M * Bertl (which is a bug ;) 1109876456 M * Doener hm, IIRC my trace had both set to 1023 1109876467 M * Bertl yeah, but not mine ... 1109876470 M * Doener and no vnet_start_xmit there... 1109876481 M * Bertl will fix that and redo ... 1109876482 M * Doener but maybe i've also been half asleep 1109876568 M * Doener anyway, that just bothered me, time for dinner... back soon 1109876573 N * Doener Doener|gone 1109876580 M * Bertl okay, enjoy your dinner! 1109876584 M * Doener|gone thx! 1109876845 M * Bertl ping_real_to_guest.txt was updated ;) 1109877669 J * duckx ~Duck@dyn-83-157-199-211.ppp.tiscali.fr 1109877730 Q * rs Quit: leaving 1109878456 Q * jsambrook Ping timeout: 480 seconds 1109879266 J * prae ~prae@sherpadown.net 1109879363 J * Nik ~Nik@cable-153-130.online.bg 1109879413 M * Bertl wb prae! Nik! 1109879590 M * prae hi* 1109879722 N * Doener|gone Doener 1109879851 Q * Nik Ping timeout: 480 seconds 1109880352 M * Bertl wb Doener! 1109880693 J * hillct ~hillct@client200-5.dsl.intrex.net 1109880718 M * Bertl wb hillct! 1109880777 M * hillct hey 1109880799 M * hillct I got a chance ot mess with vdlimit some 1109880818 M * Bertl yes, and? 1109880820 M * hillct the command syntax is not clear at all 1109880845 M * hillct I can get it to throw errors quite easily though ;) 1109880865 M * hillct [root@laptop vdlimit-0.02]# ./vdlimit -a 600000 /vservers/fourth 1109880865 M * hillct vc_add_dlimit: Function not implemented 1109880865 M * hillct vc_get_dlimit: Function not implemented 1109880865 M * hillct 600000: 0,0,0,0,0 1109880865 M * hillct vc_add_dlimit: Function not implemented 1109880867 M * hillct vc_get_dlimit: Function not implemented 1109880900 M * Bertl well, kind of expected 1109880923 M * Bertl where do you specify the context you want to add the limit to? 1109880925 M * hillct yah 1109880934 M * Bertl what is the -a 60000 supposed to do? 1109880951 M * hillct I guess I'm not interpreting the help message properly 1109880952 M * daniel_hozac shouldn't that return -EINVAL or something? ENOSYS seems wrong. 1109880997 M * hillct so, I need to specify the context -x and the limit -a (in blocks) along with the path? 1109881026 M * daniel_hozac no. 1109881032 M * daniel_hozac -a means to add the disk limit. 1109881037 M * daniel_hozac -S ... sets it. 1109881054 M * hillct as in enable it 1109881076 M * hillct so you set the limit for a context ID then you enable it for a path? 1109881091 M * daniel_hozac you need to add it before you're able to set it. 1109881099 M * hillct yah 1109881100 M * Bertl hmm, guess I have to do a howto use for the vdlimit too ... 1109881124 M * Bertl was hoping that it will be implemented properly in the util-vserver before that happens ... 1109881135 M * hillct in other news, I though I could pull the context ID out of the config file tree 1109881139 M * hillct don't see it though 1109881158 M * daniel_hozac vserver-info XID 1109881163 M * hillct ah 1109881164 M * Bertl hillct: which most likely means that you are 'somehow' using dynamic contexts 1109881183 M * Bertl and if that is the case, I would not even think about touching vdlimits ;) 1109881276 M * Bertl http://www.mail-archive.com/vserver@list.linux-vserver.org/msg03045.html <-- good thread 1109881278 M * hillct hmm 1109881297 M * hillct each server has context ID of 9 1109881315 M * hillct er 0 1109881317 M * hillct [root@laptop uts]# vserver-info fourth XID 1109881318 M * hillct 0 1109881318 M * hillct [root@laptop uts]# vserver-info firsttry XID 1109881318 M * hillct 0 1109881318 M * hillct [root@laptop uts]# 1109881343 M * Bertl did you start any vserver yet? 1109881348 M * hillct I gather I should have specified the context ID when creating the vserver 1109881351 M * hillct er 1109881354 M * hillct heh 1109881359 M * daniel_hozac vserver-info does seem broken like that. 1109881361 M * hillct I rebooted with the new kernel 1109881363 M * hillct LOL 1109881388 M * daniel_hozac according to the help, that's supposed to work, but it returns 0 for any vserver. 1109881391 M * hillct nope. still 0 1109881429 M * hillct did i have the option to specify a context ID when I created the vserver? 1109881436 M * Bertl yes, of course 1109881447 M * hillct using an arg i don't know about 1109881458 M * Bertl --context ... the static context of the vserver [default: none; a dynamic 1109881462 M * Bertl context will be assumed] 1109881478 M * Bertl but you can assign a static context id at any time ... 1109881488 M * hillct aha 1109881508 M * Bertl just put it into the context file (see flower page for path) 1109881537 M * hillct yah 1109881540 M * hillct saw that 1109881543 M * hillct didn't see the file 1109881553 M * hillct so I can change the context after the server is created 1109881554 M * hillct ? 1109881574 M * Bertl yes, just make sure to stop the vserver before 1109881584 M * hillct K 1109881590 M * Bertl then change the xid to a static (unique) one (range 2-49151) 1109881934 Q * prae Quit: Pwet 1109881963 M * hillct vserver-info still reports 0 but I guess that brokenness is known 1109882038 M * Bertl try PXID 1109882070 M * Bertl and make sure that you have started the vserver before ... 1109882078 M * hillct is this the proper syntax for setting a limit? 1109882079 M * hillct vdlimit -x 2 -a 600000 1109882092 M * Bertl nope, definitely not, did you read the thread I posted? 1109882094 M * hillct then enable with -S ,path> 1109882104 M * hillct that's the next stop 1109882149 M * hillct PXID returns nothing 1109882153 A * hillct reads 1109882257 M * daniel_hozac PXID is parent XID, and isn't implemented yet. 1109882280 M * Bertl but XID is supposed to work on running vserver (new style) right? 1109882297 M * daniel_hozac well, the --help would suggest that, but the code certainly doesn't. 1109882307 M * daniel_hozac it wants a pid. 1109882314 M * Bertl but it is used by vserver to start the helpers, no? 1109882340 M * hillct I did this by adding the file context to the config tree 1109882355 M * daniel_hozac i guess vserver-stat is the best way to get the xid of a running vserver... 1109882357 M * Bertl should be fine, check with vserver-stat 1109882357 A * hillct will try creating a new vserver with context specified shortly 1109882528 M * hillct vserver-stat reports context id correctly 1109882645 M * hillct ah 1109882652 M * hillct values must be fully specified 1109882666 M * Bertl you can use placeholders like 0,-1 and -2 1109882784 M * hillct brb 1109882953 M * Bertl nap attack ... I'm off for now ... back later I guess ... 1109882981 N * Bertl Bertl_zZ 1109883896 Q * erwan_ho Remote host closed the connection 1109885290 Q * ndim Ping timeout: 480 seconds 1109885330 M * Doener now it makes sense! 1109885356 M * Doener the netfilter stuff also uses the wrong debug setting... 1109885621 J * ndim hun@helena.bawue.de 1109886620 J * Nik ~Nik@cable-153-130.online.bg 1109886745 J * FgL1986 ~fgl1986@200.37.224.116 1109887101 Q * Nik Ping timeout: 480 seconds 1109887231 A * FgL1986 toi out. 1109888331 Q * shuri Read error: Connection reset by peer 1109888728 M * Doener Bertl_zZ: ok, i now know why the neigh lookup is down with ctx 0 1109888815 M * Doener in vnet_rebuild_header we temporarily replace the skb->dev with the real dev to call the real device's rebuild header. And the neigh_lookup is called within that path 1109888866 M * Doener so we basically give up virtualisation in that path 1109888969 M * Doener i hope i'll never forget that stuff, was a pita to find the right path through the code 1109890501 M * Doener of course that also means that it's not related to the local route entry 1109890909 M * Doener at least not directly 1109891407 M * Doener vxD: vnet_target(65535,#0) [101,3] 1109891407 M * Doener vxD: vnet_get(101) = 877de240[101,#10] 1109891407 M * Doener vxD: arp_find: skb(8623eac0, #10) dev(eth0, #0) 1109892245 M * hillct how flexible is the .distributions mechanism inside the config directory tree? 1109892255 M * hillct What governs the items I can place there? 1109892308 M * hillct is that mechanism part of the top level vservers shell script? 1109892319 A * hillct hasn't gone through it carefully yet 1109892527 M * Doener me neither, i'm more a kernel guy for linux-vserver (at least i try to be ;)... i guess most questions about how the tools work are best sent to the ml for now as the developer is not around here 1109892556 M * hillct ah 1109892565 M * hillct I'll go through the code first 1109892583 M * hillct easier to add stupid questions here than on a mailing list :) 1109892591 M * Doener heh 1109892601 M * hillct er ask 1109893207 M * daniel_hozac the .distributions is described on the flower page, isn't it? 1109893267 M * alex|ikse (the very famous beautifull flower page ;) ) 1109893294 N * alex|ikse alexx 1109893317 M * Doener css style gras1 is pretty nice IMHO 1109893368 M * alexx what is the story of this page (why the flower/grass theme) ? 1109893446 M * Doener guess you have to ask enrico to get to know that 1109893827 J * Nik ~Nik@cable-153-130.online.bg 1109893856 M * alexx Is enrico a flower lover ? :) 1109893858 M * hillct .distrivutions as defined on the flower page has two examples for yum and apt, but diesn't discuss extention to other files / data to be inserted based on the selected distribution 1109893878 M * hillct alexx it's a function of the type of flower 1109893887 M * hillct er plant 1109893916 M * Doener we all love flowers, don't we? 1109893924 M * alexx sure ... 1109893925 M * alexx ;) 1109894311 Q * Nik Ping timeout: 480 seconds