1231376289 J * hparker ~hparker@2001:470:1f0f:32c:212:f0ff:fe0f:6f86 1231376667 Q * dowdle Remote host closed the connection 1231380577 J * takeru ~takeru@nttkyo888227.tkyo.nt.ftth.ppp.infoweb.ne.jp 1231380696 M * Bertl off to bed now .. have a good one everyone! 1231380702 N * Bertl Bertl_zZ 1231381857 Q * takeru Quit: takeru 1231383299 J * neofutur ~neofutur@xena.ww7.be 1231383342 J * takeru ~takeru@nttkyo370117.tkyo.nt.ftth.ppp.infoweb.ne.jp 1231387197 Q * mugwump Remote host closed the connection 1231387355 M * TimLyth :S My DHCP server still doesn't want to start. I've specified 255.255.255.255/32 as a 2nd IP for the guest (I used the same interface as the existing IP), given NET_BROADCAST and NET_ADMIN capabilities to the guest, yet the dhcpd is still failing with Jan 8 14:58:16 dhcp dhcpd: Open a socket for LPF: Operation not permitted 1231387364 M * TimLyth Any suggestions on what further to try? 1231387393 M * daniel_hozac CAP_NET_RAW is what you want. 1231387411 M * daniel_hozac ISC dhcpd doesn't use the Linux network stack, so you can remove that IP address. 1231387439 M * TimLyth Thanks. :D 1231387452 M * daniel_hozac you can remove CAP_NET_ADMIN too. 1231387516 M * TimLyth In /etc/vserver/dhcp/bcapabilities, should it literally be CAP_NET_BROADCAST on one line and CAP_NET_RAW on another? 1231387531 M * daniel_hozac yes. 1231387541 M * daniel_hozac or comma-separated. either works. 1231387591 M * TimLyth Looks a lot more promising. I didn't get any error messages scream out at me when I've just done `vserver dhcp restart`. :D 1231388268 N * balbir_ balbirsingh 1231389156 Q * Aiken Quit: Leaving 1231390370 J * ghislainocfs2 ~Ghislain@adsl2.aqueos.com 1231390703 Q * ghislainocfs21 Ping timeout: 480 seconds 1231393534 Q * balbirsingh Ping timeout: 480 seconds 1231398777 J * doener_ ~doener@i577BAE8B.versanet.de 1231398878 Q * doener Ping timeout: 480 seconds 1231401585 J * jsambrook ~jsambrook@aelfric.plus.com 1231402143 J * kir ~kir@swsoft-msk-nat.sw.ru 1231402156 N * ag- Guest594 1231402163 J * ag- ~ag@fedaykin.roxor.cx 1231402289 N * pmenier_off pmenier 1231402414 Q * Guest594 Ping timeout: 480 seconds 1231402990 J * dna ~dna@77-207-103-86.dynamic.dsl.tng.de 1231403608 J * friendly ~friendly@ppp118-208-186-144.lns10.mel4.internode.on.net 1231404796 J * cga ~weechat@94.36.117.245 1231404972 P * jsambrook 1231405849 J * Aiken ~Aiken@ppp118-208-102-132.lns3.bne4.internode.on.net 1231407897 Q * hijacker Remote host closed the connection 1231408133 J * ktwilight__ ~ktwilight@199.118-66-87.adsl-dyn.isp.belgacom.be 1231408133 Q * ktwilight_ Read error: Connection reset by peer 1231408135 J * hijacker ~hijacker@213.91.163.5 1231409587 Q * friendly Quit: Leaving. 1231410851 J * sharkjaw ~gab@149-67-194.231210.adsl.tele2.no 1231411713 J * gnuk ~F404ror@pla93-3-82-240-11-251.fbx.proxad.net 1231411895 M * independence any gentoo users around? what happened to baselayout-1.13 ? 1231413007 M * hparker Dead... Use baselayout2 and openrc 1231413258 M * independence hm, should I use baselayout2 on the host system too then? 1231413269 Q * pmenier Read error: Connection reset by peer 1231413440 M * hparker I've got one that's not, just in the guests 1231413473 M * hparker It's getting about time to upgrade it thought, I've got it running on several other systems without problem 1231413478 M * hparker s/ght/gh 1231413607 J * pmenier ~pme@LNeuilly-152-22-72-5.w193-251.abo.wanadoo.fr 1231413949 Q * independence Quit: maintanance 1231416139 M * transacid you should read that http://www.gentoo.org/doc/en/openrc-migration.xml 1231416145 M * transacid ah he's gone already 1231416272 Q * Adrinael_ Read error: Connection reset by peer 1231416292 J * Adrinael adrinael@rid7.kyla.fi 1231417401 J * indy ~independe@cobra.lysator.liu.se 1231417422 M * indy I got "EXT3-fs: Unrecognized mount option "tagxid" or missing value" since upgrading to 2.6.27, any ideas? 1231417484 M * indy should I use -o tag instead of tagxid? 1231417526 M * indy oh, it was in the FAQ even.. sorry :P 1231418610 Q * jrdnyquist Quit: Leaving 1231419377 J * independence independen@titan.blinkenshell.org 1231419697 M * independence hm, can I reload ccapabilities while running? 1231419707 M * cehteh no 1231419833 M * independence hm, wasn't there some tool that could do it while running though? or cat something to proc? 1231419867 Q * Aiken Quit: Leaving 1231419867 M * transacid [13:02:19] ( transacid) you should read that http://www.gentoo.org/doc/en/openrc-migration.xml 1231420068 N * Bertl_zZ Bertl 1231420109 M * Bertl morning folks! 1231420166 M * fb hello Bertl! :) 1231420318 M * Bertl independence: with capability masking you can change the upper cap limit at runtime 1231420450 M * independence I wanted to add RAW_ICMP, can I do that? 1231420537 M * Bertl as I said, if your Linux-VServer version supports capability masking (see feature matrix) then you can raise the upper bound 1231421154 Q * takeru Quit: takeru 1231421623 M * independence I'm using 2.3, and it says yes in the feature matrix 1231421713 J * docelic_ ~docelic@78.134.192.144 1231421794 M * independence I don't find much info on how to use it on the wiki though, it just says the mask is 0x00000100 1231422109 Q * docelic__ Ping timeout: 480 seconds 1231422263 M * Bertl the capability, yes, you want to add that to the existing bcapabilities 1231422535 M * independence but it says RAW_ICMP is ccaps? not bcaps? 1231422549 J * glen__ ~glen@elves.delfi.ee 1231422557 M * glen__ how do i rename vserver? 1231422701 M * nox glen__: stop > rename rootdir , etcdir , nodename > start 1231422750 M * nox *edit nodename 1231422756 M * Bertl glen__: what do you actually want to change when you say 'rename'? 1231422767 M * glen__ rename the internal representation of vserver NAME 1231422770 M * glen__ not DNSNAME 1231422792 M * glen__ so i could say vserver NEWNAME stop instead of vserver OLDNAME stop 1231422926 M * Bertl then renaming the /etc/vservers/ should suffice 1231422957 M * Bertl if you want to keep it consistant, rename the /vservers/ too, but you have to adjust the vdir link 1231423006 M * glen__ only vdir link needs to be changed? as there are quite dozen of links 1231423031 M * glen__ can i do it without stopping vserver? or then my vserver would go orphaned? 1231423032 M * Bertl double check with daniel_hozac and maybe urge him to add a rename command :) 1231423072 M * glen__ like mv /etc/vserver/{OLDNAME,NEWNAME}; ln -s NEWNAME /etc/vservers/NEWNAME could do? 1231423183 M * Bertl the second part is definitely wrong 1231423250 M * Bertl you want to adjust vdir and probably run (but that should be auto-updated) iif you move the location of the guest data 1231423300 M * glen__ no i won't move data, just control files 1231423311 M * glen__ the symlink is there so i could later stop the vserver, i guess 1231423331 M * glen__ or it is not needed, as it will take just ctx_id from /etc/vserver/NAME and be fine? 1231423357 M * Bertl then just the mv /etc/vserver/{OLDNAME,NEWNAME} is enough 1231423397 M * Bertl note that it will still be listed with the name entry in e.g. vserver-stat 1231423537 M * TimLyth :S I've successfully got my DHCP server to start up without any errors, but it isn't wanting to hand out IP's AFAIK. 1231423586 M * TimLyth As I want to keep the guests as light as possible, I'd like to avoid installing tcpdump in the guest. Is there another way to confirm if the broadcast packets are hitting the guest? 1231423773 Q * zbyniu Read error: Connection reset by peer 1231423826 J * zbyniu ~zbyniu@ip-62.181.188.13.static.crowley.pl 1231424332 M * Bertl TimLyth: slowly, first, you know that the guest uses the hosts network stack 1231424370 M * Bertl TimLyth: dhcp is below the IP layer, so you have to give special caps to the guest to access that (basically host network capabilities) 1231424456 M * Bertl now, what 'broadcast' packets are you talking about? 1231424457 M * TimLyth Bertl: I followed daniel's advise above - allowed CAP_NET_BROADCAST and CAP_NET_RAW for the guest. 1231424478 M * TimLyth The broadcast packets that a DHCP client send outs. 1231424530 M * Bertl they will be received by your dhcp server, when it attaches to the raw socket 1231424557 M * Bertl make sure that you do not limit the guest to an IP subset 1231424607 M * TimLyth IP for guest is 192.168.232.24/27 1231424622 M * TimLyth Host is 192.168.232.1/27 1231424731 M * TimLyth IPtables on host has no rules and a policy to ACCEPT all. 1231424732 M * Bertl as I said, you do not want to limit the guest to certain IPs, if you are running protocols below the IP layer 1231424759 M * Bertl btw, tcpdump on the host is as good as on the guest ... 1231424825 M * TimLyth Ok. 1231424934 M * TimLyth The network has always been a /27. 1231424971 M * TimLyth Are you suggesting that I don't even configure an IP for my guest? 1231425005 M * Bertl well, you want to run protocols below the IP layer, so limiting the guest to IPs on the IP layer doesn't make much sense 1231425018 M * Bertl (just leads to malfunction) 1231425031 M * Bertl specify 0.0.0.0 for the dhcpd guest 1231425038 M * TimLyth Ok. 1231425048 M * independence how do I add RAW_ICMP to bcaps while running? is it enough to just put it in /etc/vservers/x/bcapabilities ? 1231425050 M * glen__ Bertl: nope. if i rename /etc/vservers/NAME, then i can no longer see the running vserver 1231425082 M * glen__ Bertl: and now it's orphaned, vserver-stat doesn't show it's name after i renamed it back 1231425097 M * TimLyth What prefix do I give my dhcpd guest? /0 or /32? 1231425119 M * Bertl shouldn't matter 1231425123 M * TimLyth glen__: vtop should allow you to see the processes and kill them off. 1231425135 M * Bertl glen__: what exactly did you do? 1231425143 M * glen__ 16:02:37 Bertl> then just the mv /etc/vserver/{OLDNAME,NEWNAME} is enough 1231425150 M * glen__ and then vserver-stat 1231425171 M * Bertl and you stopped the guest before, and restarted it afterwards? 1231425188 M * glen__ no i did not stop it 1231425208 M * glen__ 15:57:11 glen__> can i do it without stopping vserver? or then my vserver would go orphaned? 1231425211 M * glen__ 16:01:51 glen__> the symlink is there so i could later stop the vserver, i guess 1231425218 M * Bertl well, that's the problem ... 1231425219 M * glen__ but you said later the symlink is not needed 1231425233 M * Bertl kill the guest with vkill 1231425234 M * glen__ you didn't realize i wanted to do on running server? 1231425240 M * Bertl no 1231425243 M * glen__ :( 1231425247 M * glen__ well i can ssh there 1231425247 M * Bertl you never ever rename a running guest 1231425255 M * glen__ no 1231425260 M * glen__ i wanted to be prepared for it's rename 1231425272 M * glen__ so next time somebody it or host is restarted new name gets used 1231425352 M * glen__ ah. somewhy vdir got changed too 1231425360 M * glen__ i fixed vdir and now can enter with new name 1231425385 M * glen__ vserver-stat however still doesn't show name. but that part i don't care as long as commands work 1231425397 M * TimLyth I'm going to give up on the DHCP server for now. It's 01:36 where I am, so I'm heading off to bed. 1231425404 M * Bertl glen__: it depends on the run symlink 1231425418 M * Bertl glen__: i.e. that one is wrong for the vserver-stat 1231425434 M * Bertl TimLyth: have a good night sleep then! 1231425434 M * glen__ Bertl: it (vdir) was changed somehow automatically. dunno how, i didn't do it 1231425450 M * glen__ Bertl: run in /etc/ ? 1231425477 M * glen__ ln -sf /var/run/vservers/devportal run 1231425482 M * glen__ still doesn't show name 1231425512 M * Bertl you have a link in /etc/vservers//run, which points to 1231425534 M * Bertl /var/run/vservers/, which contains the context id of that guest 1231425547 M * glen__ yes, it contains context id of the guest 1231425583 M * Bertl and you have an entry in /var/run/vservers.rev/ 1231425596 M * Bertl (with the context id) which points back to the dir 1231425622 M * Bertl if both are adjusted properly, the utils should be fine 1231425660 M * glen__ ahaa, the .rev one is wrong 1231425685 M * glen__ now link in .rev is ok, but vserver-stat still has no name :) 1231425745 M * Bertl then something else is missing, which is done correctly on guest startup :) 1231425883 M * independence I still get "ping: icmp open socket: Operation not permitted", kernel 2.6.27vs2.3.0.36.4, util-vserver 0.30.215.. any ideas? 1231425988 M * Bertl first check that you have the required bcaps 1231426015 M * Bertl wait, RAW_ICMP is a flag 1231426051 M * Bertl i.e. you have to add that to the cflags, not bcaps, further its not related to capability masking at all 1231426109 A * Bertl is now confused ... 1231426116 M * independence me too... 1231426123 M * Bertl what are you trying to accomplish? 1231426124 M * independence it says on the wiki on sereval pages that it's a ccap 1231426129 M * independence I'm trying to ping from a guest 1231426143 M * Bertl that doesn't need any capabilities, just a flag, sec 1231426201 Q * sharkjaw Quit: Leaving 1231426218 M * independence hm, I can ping if I'm root in the guest 1231426240 M * Bertl okay, so you want to ping as non root 1231426265 M * independence yeah 1231426268 M * Bertl then you have to set the suid bit of the guest or give the capability to the binary 1231426277 M * Bertl suid bit of ping in the guest 1231426323 M * Bertl alternatively, if you have filesystem based posix capabilities enabled, you can give just the CAP_NET_RAW to ping 1231426325 Q * doener_ Read error: Connection reset by peer 1231426394 J * doener ~doener@i577BAE8B.versanet.de 1231426399 M * independence I'd like to give it the least privileges required 1231426433 M * independence I'm running ext3.. not sure if I have posix capabilities enabled 1231426445 J * takeru ~takeru@nttkyo888227.tkyo.nt.ftth.ppp.infoweb.ne.jp 1231426536 M * Bertl check with getfcaps/setfcaps 1231426834 M * glen__ blah. is there any known problems with networking with many vservers/ip's 1231426847 M * glen__ we've like replaced all hardware and still having some weird timeouts 1231426865 M * Bertl nope, on older kernels, there was a 16ip limit, but that was lifted some time ago 1231426876 M * independence hm, I don't have a getfcaps command even 1231426888 M * Bertl independence: you need a recent libcap 1231426892 M * glen__ well. it's not permanently timeouting, but time to time 1231426909 M * glen__ # ip a|grep inet|grep -v inet6|wc -l 1231426909 M * glen__ 35 1231426911 M * independence Bertl: I have latest in stable portage, 2.15 1231426943 M * glen__ but any clues how to debug the network timeout problems? 1231427045 M * Bertl independence: try getcap/setcap 1231427058 M * Bertl glen__: what times out and when? 1231427087 M * independence Failed to get capabilities of file `/vservers/titan/bin/ping' (Operation not supported) 1231427087 M * glen__ Bertl: time to timei get timeouts with world. the timouts persisst few seconds sometimes minutes 1231427103 M * glen__ and if target has also ipv6 address, i get address not supported errors 1231427106 M * Bertl anything in dmesg? 1231427106 M * independence so I guess I should go for suid then? 1231427109 M * glen__ Bertl: nope 1231427130 M * Bertl glen__: yep 1231427144 M * Bertl s/glen/independence/ 1231427158 M * glen__ only thing vserver related i see in dmesg is not related: 1231427159 M * glen__ Jan 8 16:34:40 lumpy kernel: [ 8203.714234] vxW: [ps,21180:#1225|1225|1225] did lookup hidden devpts:ffff8801ed96d380[#0,2] /dev/pts. 1231427162 M * glen__ Jan 8 16:34:40 lumpy kernel: [ 8203.714245] vxW: [ps,21180:#1225|1225|1225] did lookup hidden devpts:ffff8801ed96d380[#0,2] /dev/pts. 1231427175 M * glen__ Jan 8 16:57:36 lumpy kernel: [ 9579.802630] vxW: [xid #234] !!! limit: ffff88020ccf0090[,20] = 9 on exit. 1231427179 M * glen__ Jan 8 16:57:36 lumpy kernel: [ 9579.802638] vxW: [xid #234] !!! limit: ffff88020ccf0090[,21] = 9 on exit. 1231427197 J * jrdnyquist ~jrdnyquis@slayer.caro.net 1231427236 Q * takeru Quit: takeru 1231427250 M * Bertl well, as Linux-VServer doesn't change the network stack, just imposes some limits (IP subset) timeouts are not directly affected 1231427297 M * glen__ weird thing is that it is always problem with eth0 1231427299 M * independence Bertl: ok, it workes now, thanks :) 1231427312 M * glen__ even if new network card is inserted and motherboard changed twice 1231427336 M * Bertl what kind of card is eth0? 1231427358 M * daniel_hozac and the cable? and the switch? 1231427392 M * jrdnyquist How do you guys monitor vserver instances for yum updates? I use yum-updatesd on the host but if I try into install that into an instance it wants to install yum and many other things that I do not want to internalize in my instances 1231427394 M * glen__ [ 10.912572] sky2 eth0: Link is up at 1000 Mbps, full duplex, flow control rx 1231427420 M * glen__ err, dunno how to look more than the sky2 1231427471 M * Bertl I had a sky2 onboard some time ago, it was replaced by an intel card because it caused strange packet loss 1231427507 M * glen__ well we put sky2 today there, as we tought the onboard ones suck 1231427510 M * glen__ this sky2 is external 1231427526 M * Bertl and the onboard card was what type? 1231427528 M * glen__ 03:00.0 Ethernet controller: D-Link System Inc DGE-560T PCI Express Gigabit Ethernet Adapter (rev 13) 1231427531 M * glen__ 05:00.0 Ethernet controller: Intel Corporation 80003ES2LAN Gigabit Ethernet Controller (Copper) (rev 01) 1231427535 M * glen__ 05:00.1 Ethernet controller: Intel Corporation 80003ES2LAN Gigabit Ethernet Controller (Copper) (rev 01) 1231427543 M * glen__ e1000 1231427563 M * glen__ switch had no errors on the port, so it is not switch error? 1231427565 M * Bertl that should be fine, no problems with e1000* so far here 1231427579 M * glen__ and there are no errors on ifconfig errors for rx: tx: 1231427582 M * Bertl what kind of traffic times out? 1231427593 M * glen__ well, i'm usually stuck with cvs straffic 1231427596 M * glen__ others use ssh 1231427616 M * Bertl so tcp traffic? 1231427621 M * glen__ yeah 1231427638 M * Bertl tcp has a rather long timeout, something around 3 minutes or more 1231427655 M * Bertl that would suggest that your routes suddenly change 1231427671 M * Bertl can you a tracepath for the route in question and upload that? 1231427691 M * Bertl (feel free to anonymize the addresses) 1231427707 M * glen__ have to wait for the timeouts to start 1231427731 M * Bertl the situation right now could already clarify things 1231427740 M * glen__ ok. i'll try to tracepath the gateway 1231427762 M * Bertl tracepath from your ssh origin to the guest/server 1231427768 M * glen__ 1: x.x.x200 0.143ms pmtu 1500 1231427768 M * glen__ 1: x.x.x254 5.585ms reached 1231427768 M * glen__ 1: x.x.x254 5.561ms reached 1231427768 M * glen__ Resume: pmtu 1500 hops 1 back 255 1231427777 M * glen__ this is when things are Ok 1231427783 M * Bertl i.e. the same path which is timing out 1231427786 M * Bertl (please use paste.linux-vserver.org for everything longer than 3 lines) 1231427799 M * glen__ .200 is host addr and .254 is the gw 1231427818 M * Bertl two hops on the same IP? 1231427859 M * mnemoc drunk router? 1231427874 M * Bertl if that is correct, it looks like a transparent firewall 1231427884 M * glen__ hmm. it sure is two, and same with backup gateway 1231427891 M * Bertl which could cause all kinds of timeout 1231427892 M * glen__ but backup gateway is somewhat faster 1231428127 M * glen__ oddly enough, tracepath is OK from inside vserver 1231428135 M * glen__ i.e the gw ip is reached once 1231428158 M * glen__ 1: x.x.x.200 0.155ms pmtu 1500 1231428158 M * glen__ 1: x.x.x.254 5.689ms reached 1231428158 M * glen__ Resume: pmtu 1500 hops 1 back 1 1231428168 M * Bertl well, then I'd check with the gw admin :) 1231428184 M * glen__ this was from vserver, which has internal ip on other interface and iptable nat ruled to real world 1231428203 M * glen__ sth like: SNAT all -- 192.168.2.0/23 !192.168.2.0/23 to:x.x.x.200 1231428216 M * Bertl looks to me like your connections are dropped on the gw 1231428239 M * glen__ yeah, i start to think that too 1231428248 M * Bertl (which probably does some firewalling or specific accounting) 1231428266 M * glen__ it's ISP gw. dunno what they do there 1231428286 M * Bertl you could try setup openvpn or similar on the vserver host 1231428309 M * glen__ uh. what that help? 1231428310 M * Bertl and setup your private network, that should allow a strable connection 1231428316 M * Bertl *stable 1231428322 M * Bertl i.e. for ssh access 1231428345 M * glen__ the thing is that problems are only with internet interface, local lan interface is ok. it is from where i see that stuff is down 1231428350 M * Bertl when the gw drops the connection, the openvpn will reestablish it, and tunneled data will not see it 1231428353 M * glen__ which even more points to gw being problematic 1231428384 M * Bertl yeah, my best guess would be some kind of table-overflow-reset scenario on the gw 1231428420 M * Bertl the more traffic gets routed through, the sooner the gw will give up 1231428423 M * mnemoc Bertl: have you tried tinc-vpn with vservers? 1231428453 M * Bertl I tried tinc some time ago, but it didn't do what I wanted, so it got replaced 1231428472 M * glen__ this specific host doesn't do much traffic. i wonder why this host gets hit with gw problems 1231428509 M * glen__ maybe gw is so "smart" that it will first kick out hosts who have more ip's? 1231428537 M * glen__ ok. thanks for the discuss, now have something to research 1231428617 M * mnemoc i want to make a secured "mesh network" of vservers on different hosts and I was poudering between tinc-vpn and cloudvpn as openvpn only allows one server per network 1231428619 M * Bertl mnemoc: might have changed now .. was quite some time ago 1231428632 M * mnemoc but I haven't tested any yet 1231428667 M * meebey I added /proc/bus/usb/ to /etc/vservers/.defaults/apps/vprocunhide/files and it barfs about it like this: /proc/bus/usb/: Invalid argument /proc/bus/usb/devices: Invalid argument 1231428701 M * Bertl not unexpected 1231428721 M * Bertl /proc/bus/usb despite the path, is not a procfs 1231428766 J * dowdle ~dowdle@scott.coe.montana.edu 1231428799 M * meebey ups 1231428825 M * meebey Bertl: I tried bind mount the first time, but the target path is not there, so how can I mount it? 1231428839 M * Bertl so, you want to add that to the guest configs fstab 1231428888 M * meebey hm actually, mounting it a 2. time should work too 1231428890 A * meebey tries 1231428949 Q * sladen Ping timeout: 480 seconds 1231428982 M * meebey hm its not there 1231428984 M * meebey root@ups-apc_putin:/# ls /proc/bus/ 1231428984 M * meebey input pci 1231429004 M * meebey procbususb /proc/bus/usb usbfs defaults 0 0 1231429015 M * meebey my line in /etc/vserver/ups-apc/fstab 1231429057 M * meebey is that issue that the target mount directory is not present in the vserver? not sure how /proc handles that in general 1231429094 M * mnemoc meebey: unhide the mount point :) 1231429095 M * Bertl it seems you did unhide /proc/bus successfully, but you have to do that with /proc/bus/usb too 1231429106 M * Bertl (before you mount the usbfs on it :) 1231429150 M * meebey eeeks thats nasty 1231429153 J * er ~yakker@aegis.CS.Princeton.EDU 1231429175 M * Bertl hey er! back to the US? 1231429180 M * er Bertl, yep 1231429194 M * Bertl smooth flight? 1231429222 M * er that's _almost_ an oxymoron... 1231429233 M * Bertl hehe :) 1231429268 M * er apparently, it was -22 degrees in Paris, so my flight got canceled. then i had to go through amsterdam, which added about 5 hours to my travel time. 1231429292 M * er o well, I regained my mastery of Tetris, so not all is lost 1231429300 M * meebey ok I will remove the usb stuff from the host then 1231429303 M * meebey doesnt need it anyhow 1231429322 M * meebey that way /proc/bus/usb stays unmounted 1231429324 M * er Bertl, are you porting the PL kernel to 2.6.28, or to 2.6.27 ? 1231429348 M * er so that I know what I need to port Chopstix to. 1231429454 M * Bertl first port will be on 2.6.27.x, but I guess a 2.6.28 version shouldn't be out of reach :) 1231429463 A * meebey has to hack /etc/init.d/mountkernfs.sh 1231429479 M * Bertl er: 2.6.27 is said to be long-time maintained ... so that is probably a good target 1231429563 M * er kewl. 1231429628 M * meebey yay /proc/bus/usb is mounted in the vserver 1231429630 M * mnemoc Bertl: will it be "long-time maintained" by vs too? :) 1231429672 M * Bertl mnemoc: that's the idea :) 1231429811 M * mnemoc great =) 1231430862 J * balbir_ ~balbir@122.172.106.53 1231431192 J * sladen paul@starsky.19inch.net 1231433346 A * vasko is back. from sleep 1231433361 M * Bertl good morning vasko :) 1231433367 M * vasko hi :) 1231433501 M * vasko i am having i386 guest under amd64 host. in guest uname -m reports x86_64 what causes problem to some oracle scripts. is there an easy way to make it return 'i386'? 1231433548 M * vasko except editing /bin/uname, i mean to make it the way that syscal will return i386 1231433566 M * Bertl did you set the personality properly? 1231433648 M * Bertl http://www.nongnu.org/util-vserver/doc/conf/configuration.html (look for personality) 1231433714 Q * PowerKe Ping timeout: 480 seconds 1231433728 M * vasko no... 1231433747 M * Bertl then try that, linux32 is a good one 1231433751 M * vasko is there a way to set it in runtime 1231433754 M * vasko ? 1231433757 M * Bertl not really 1231433781 M * vasko ok, thanks 1231434150 M * Bertl np 1231434815 J * bonbons ~bonbons@2001:960:7ab:0:2c0:9fff:fe2d:39d 1231436089 M * meebey cat /dev/usb/hiddev0 1231436089 M * meebey cat: /dev/usb/hiddev0: Permission denied 1231436098 M * meebey is there a CAP for USB? not sure such API that uses 1231436123 M * Bertl no, there is no USB capability 1231436142 M * Bertl but I presume it will check for some other major capability 1231436149 M * Bertl like, e.g. CAP_SYS_ADMIN 1231436185 M * meebey let me try that one 1231436190 M * Bertl from what you described, you want to monitor/control an UPS from the guest, no? 1231436238 M * meebey once I succeed I should add apcupsd to the vserver wiki 1231436247 M * meebey Bertl: exactly 1231436266 M * Bertl well, you do not need or want the complete bus stuff and excessive caps for that 1231436285 M * Bertl you just identify the proper interface, probably serial via usb 1231436298 M * Bertl and then you copy just that device node into the guest 1231436327 M * meebey hiddev0 doesnt sound like serial over usb 1231436349 M * meebey and as it can't read it, the device nod is not enough 1231436392 M * meebey complete or limited usb bus doesn't make a difference in my case 1231436398 M * fb meebey: why you want to run apcupsd in the guest and not on the host? 1231436398 M * Bertl as usual, some tools think they need to disect the machine first, just to figure that all it needs is access to a serial device :) 1231436440 M * meebey fb: because the host is not running anything 1231436461 M * meebey besides linux + udev + sshd 1231436485 M * fb meebey: but the host can gracefully shut down all running guests at the moment 1231436527 M * fb it seems a natural place to run apcupsd to me, not the guest 1231436584 M * fb and this is my setup, sshd, apcupsd and ntpd on the host 1231436593 A * meebey is probably the most heavy vserver user, he got everything running inside vserver (somtimes had to use nasty caps of course) 1231436606 M * meebey fb: the machine will shutdown a bunch of servers so... 1231436635 M * meebey via ssh that is, so it will shutdown itself too (as last host) 1231436665 M * fb but why doing this simple task so complicated? 1231436715 M * meebey I am using vserver differently than the most probably :) 1231436741 M * meebey fb: because I prefere to make things as simple as they can be, thinking outside the box 1231436761 M * meebey this is not the only location where I will monitor the UPS and shutdown all hosts 1231436774 M * meebey once I have it working, I can spread the vserver everywhere ;) 1231436966 M * meebey at least Bertl isn't asking anymore why I do run so strange stuff inside vservers :-P 1231437065 M * meebey Bertl: looks like its not CAP_SYS_ADMIN, always the same fun guessing which CAP it is :( 1231437084 M * daniel_hozac meebey: permission denied sounds more like you're running it as non-root. 1231437101 M * meebey daniel_hozac: na, I am root :) 1231437117 M * meebey cat /dev/usb/hiddev0 1231437117 M * meebey ……(2 1231437121 M * meebey its happy on the host 1231437126 M * daniel_hozac and in the guest? 1231437145 M * meebey fails with: 1231437145 M * meebey root@ups-apc_putin:/# cat /dev/usb/hiddev0 1231437145 M * meebey cat: /dev/usb/hiddev0: Permission denied 1231437160 M * meebey crw-rw---- 1 root root 180, 96 Jan 8 15:57 /dev/usb/hiddev0 1231437248 M * meebey would be nice to have some kind of CAP tracing/logging 1231437275 M * meebey someone would need to convince about that linus I guess 1231437391 M * mnemoc that would be a pretty expensive feature 1231437407 M * mnemoc (i think) 1231437462 M * meebey the low level has_cap() function would just need to printk the result, or is that expensive? 1231437472 M * Bertl meebey: would be trivial to hack, just add a printk 1231437484 M * meebey Bertl: yeah thats what I thought 1231437489 M * Bertl but not really that useful ... 1231437490 M * daniel_hozac you could even do that with kprobes or similar. 1231437504 M * daniel_hozac to make it a temporary thing. 1231437507 M * mnemoc meebey: how many printk per second that would cause? :) 1231437518 M * meebey mnemoc: no idea, hopefully not too many :-P 1231437544 M * meebey so maybe limit it to a named namespace and be happy 1231437577 M * meebey or processspace? whatever its called, where the processes run in 1231437652 N * pmenier pmenier_off 1231437657 Q * pmenier_off Remote host closed the connection 1231438021 M * mnemoc meebey: http://sourceware.org/systemtap/kprobes/README.kprobes sounds like a good way :) 1231438160 M * meebey nice 1231438391 M * meebey doesn't look good, even with all CAPs listed in bcaps-v13.c it fails... I think I will inspect the hiddev code now 1231439252 M * mnemoc pathetic question, module_param(debug, int, 0); 1231439267 M * mnemoc implies modprobe mymod debug=1 to enable? 1231439277 M * mnemoc or just dbeug 1231439294 M * Bertl debug=N 1231439300 M * mnemoc thanks 1231439466 M * mnemoc freaking "sundance" :( ... it claims to find the hardware but no ethN appearce :( 1231439490 M * mnemoc appearse 1231439742 M * Bertl does something else appear in 'ip link ls'? 1231439773 M * mnemoc lo and the built-in eth 1231439794 M * mnemoc (same at /sys/class/net) 1231439918 M * mnemoc Bertl: http://rafb.net/p/UWBD4861.html, printk :( 1231440065 M * Bertl wll, doesn't claim anything about detected hardware there 1231440078 M * Bertl just lists the interrupts 1231440088 M * mnemoc uhm 1231440097 M * Bertl (for different pci ports :) 1231440109 M * mnemoc but it "found" them... 1231440125 M * Bertl yes, but they 'belong' to the PCI bridge :) 1231440178 M * Bertl did you try lspci -v (with a recent lspci)? 1231440226 M * mnemoc 2.2.9 ... Kernel modules: sundance 1231440243 M * mnemoc but, do you mean someone else kidnapped them? 1231440271 M * Bertl hehe, no, would be very unusual :) 1231440272 M * mnemoc i have had that problem with pata vs. sata in every machine i try .27 1231440296 M * mnemoc i have to preload the sata module to see the disks 1231440314 M * Bertl that is something the initramfs usually does 1231440314 M * mnemoc (or blacklist the pata) 1231440382 M * Bertl what's the pci id of your card (lspci -vn) 1231440412 M * Bertl (and the subsystem id, if present) 1231440484 M * mnemoc both 1186:1012 1231440490 M * mnemoc http://rafb.net/p/tHvQ3Z69.html 1231440520 M * mnemoc i'm really really lost here :( sorry to bother you with this OT 1231440640 M * Bertl good that they are not both 1186:1012 :) 1231440669 M * Bertl that's a four port pci card, yes? 1231440674 M * mnemoc yes 1231440704 M * mnemoc doh, I saw them identical :| 1231440783 M * Bertl D-Link DFE-580TX ? 1231440789 M * mnemoc exactly 1231440809 M * Bertl worked before? 1231440880 M * mnemoc it's brand new.... and 12k km (Mm :p) away from me :\ 1231440920 M * Bertl well, you are not the only one having troubles with that card 1231440952 M * Bertl please upload the output of 'modinfo sundance' 1231440963 M * mnemoc i'm supposed to setup a router/firewall/bridge there with a cache proxy inside a guest 1231441006 M * mnemoc sweet, no modinfo :p ... is it fair to run modinfo over that module on a different system? 1231441041 M * Bertl if it is the same kernel, sure 1231441072 M * mnemoc not the same kernel :\ 1231441174 M * mnemoc and busybox doesn't have modinfo :( 1231441194 M * daniel_hozac you can scp the module to a different system. 1231441305 M * mnemoc http://rafb.net/p/2MaMDe51.html 1231441313 M * mnemoc but the running kernel is not the same 1231441326 M * mnemoc (if that affects anything) 1231441327 M * daniel_hozac doesn't matter. 1231441331 M * mnemoc good :) 1231441375 M * mnemoc i'm trying to install from an ssh-powered initrd running from a pendrive :| 1231441478 M * Bertl http://webui.sourcelabs.com/kernel/issues/7821 1231441533 M * Bertl has your kernel CONFIG_SUNDANCE_MMIO=y set? 1231441628 M * mnemoc yes :\ 1231441647 M * mnemoc Bertl: thanks a lot for the hint! i'll recomplie 1231441658 M * Bertl you're welcome! 1231441705 M * mnemoc i was googling +2.6.27 and found nothing... i never thought it could be older 1231441734 M * Bertl well, the hardware is out of production for several years now 1231441743 M * mnemoc uh 1231441781 M * daniel_hozac i don't think so. 1231441798 M * daniel_hozac i've seen them being sold brand new as recently as a year ago. 1231441838 M * Bertl interesting ... I thought they were dropped in favor of PCIe cards 1231441896 M * Bertl (because PCI already has troubles with 4 fast ethernet ports) 1231441937 M * Bertl but yeah, I see d-link still lists them ... my fault 1231441947 M * daniel_hozac it's just 800 Mbps. 66 MHz PCI slots have 2 Gbps of bandwidth. 1231442019 M * Bertl so it might be worth telling mainline about that if that fixes the issue :) 1231442115 M * mnemoc Bertl: a patch removing mmio support would be fair for lkml? i have never send anything there 1231442145 M * Bertl I guess, a patch adding the proper pci access stuff would be more appreciated :) 1231442146 M * mnemoc and i fair to send something :p 1231442190 M * Bertl but raising the issue, and providing some testing (including contacting the maintainer) would probably do already 1231442328 M * mnemoc written down... i'll get this installed and working and then I'll try to get it fixed upstream 1231442336 M * mnemoc thanks a LOT for your support :) 1231442371 M * Bertl np 1231442515 M * mnemoc Bertl: OT: do you still use mandriva? 1231442524 M * Bertl zup 1231442532 M * Bertl *yup 1231442556 M * mnemoc de<->en layout conflict? :p 1231442562 M * mnemoc how is it? 1231442590 M * Bertl yeah, I have two keyboards here, one german one english, and gnome is too dumb to manage different layouts 1231442590 M * mnemoc i plan to buy a mips64 subnotebook which comes with mandriva 1231442599 M * mnemoc w 1231442600 M * mnemoc ow 1231442614 Q * gnuk Quit: NoFeature 1231442616 M * daniel_hozac isn't that an X thing? 1231442632 M * mnemoc it's nicer to blame gnome :p 1231442640 M * daniel_hozac of course. 1231442655 M * Bertl yes, it _is_ an x thing, but I suspect (no comment from the xorg folks yet) that xorg actually _can_ do that# 1231442712 M * Bertl (separate keyboard config for each keyboard, separate layout) 1231442736 M * Bertl but as gnome wants to configure it's own keyboard layout, that will get reset, I guess 1231442893 J * geb ~geb@92.4.82-79.rev.gaoland.net 1231442956 J * klikz_ mike@no.phear.eu 1231442956 Q * klikz Read error: Connection reset by peer 1231443349 M * meebey I found the issue with hiddev0, I reviewed the code and it doesn't use CAPs at all 1231443372 M * meebey then I copied the dev file into the vserver and I can read that one, without any CAPs 1231443383 M * meebey so it's simple mount bind issue... (again) 1231443400 M * Bertl what mount bind_ 1231443403 M * Bertl ? 1231443409 M * meebey /dev/usb to /dev/usb 1231443418 M * Bertl why did you do that? 1231443423 M * meebey udev populates that one 1231443433 M * Bertl and why did you forget to specify the 'dev' option? 1231443448 M * meebey Bertl: because in older days mount bind wasn't stupid like that! 1231443449 M * meebey Bertl: ;) 1231443454 M * Bertl (by default, all mounts get nodev :) 1231443464 J * xdr_ ~xdr@gote2.247.cust.blixtvik.net 1231443470 M * mnemoc fortunelly :) 1231443480 M * meebey root@ups-apc_putin:/# cat /dev/usb/hiddev0 1231443483 M * meebey works now 1231443485 M * meebey :) 1231443490 M * meebey gar 1231443496 M * meebey bind,dev beind the fix 1231443499 M * meebey -d+g 1231443526 M * mnemoc bing,gev ? 1231443550 M * meebey lol ups 1231443571 M * meebey mnemoc: you are supposed to apply such patch pattern smartly! :-P 1231443573 Q * xdr Ping timeout: 480 seconds 1231443583 M * mnemoc :) 1231443736 Q * TimLyth Ping timeout: 480 seconds 1231444971 M * Bertl nap attack ... bbl 1231444982 N * Bertl Bertl_zZ 1231447036 J * hijacker_ ~hijacker@87-126-142-51.btc-net.bg 1231449485 Q * hijacker_ Remote host closed the connection 1231449986 J * Aiken ~Aiken@ppp118-208-102-132.lns3.bne4.internode.on.net 1231450568 Q * dna Quit: Verlassend 1231450805 Q * gdm_ Quit: leaving 1231450928 Q * bonbons Quit: Leaving 1231451298 J * larsivi ~larsivi@9.80-202-30.nextgentel.com 1231451309 N * Bertl_zZ Bertl 1231451343 M * Bertl back now ... 1231452002 Q * cga Quit: WeeChat 0.2.6 1231453746 Q * Aiken Quit: Leaving 1231454970 J * matthes ~chatzilla@static-87-79-88-215.netcologne.de 1231454985 M * matthes l 1231455001 Q * matthes 1231456411 J * blues_ blues@acj116.neoplus.adsl.tpnet.pl 1231456428 J * rangaparmastan ~ohm@84.78.128.255 1231456529 Q * blues Ping timeout: 480 seconds 1231456813 J * maharaja raoul@80-64-143-52.rev.ipax.at 1231456820 M * maharaja re 1231456932 M * Bertl wb 1231457719 J * ghislainocfs21 ~Ghislain@adsl2.aqueos.com 1231457913 Q * geb Remote host closed the connection 1231458024 Q * ghislainocfs2 Ping timeout: 480 seconds 1231458673 J * TimLyth ~tux@202.134.227.227