1516060929 Q * sannes Quit: Leaving. 1516066802 M * Bertl_oO arekm: please try if the following patch fixes the compile issues for you: 1516066805 M * Bertl_oO http://vserver.13thfloor.at/Experimental/delta-ts64-fix01.diff 1516075867 M * Bertl_oO off to bed now ... have a good one everyone! 1516075871 N * Bertl_oO Bertl_zZ 1516080338 M * arekm Bertl_zZ: will check. In meanwhile someone else at pld played with this and made such patch: http://git.pld-linux.org/gitweb.cgi?p=packages/kernel.git;a=blob;f=kernel-vserver-fixes.patch;h=8284edfe4f62c34690fd2a9e479c5b3473e266ed;hb=refs/heads/LINUX_4_9 1516080351 M * arekm Bertl_zZ: and with that patch it did build fine 1516080372 M * arekm Bertl_zZ: (all hunks after first one) 1516080682 J * druschka_domaintechnik ~druschka@85.118.185.154 1516080683 Q * druschka_domaintechnik 1516084959 J * druschka_domaintechnik ~druschka@85.118.185.154 1516085415 P * druschka_domaintechnik 1516085431 J * druschka_domaintechnik ~druschka@85.118.185.154 1516087525 J * fstd_ ~fstd@xdsl-85-197-45-186.netcologne.de 1516087972 Q * fstd Ping timeout: 480 seconds 1516087973 N * fstd_ fstd 1516088257 M * Ghislain hi, cannot build the 4.9 here, i got stuck in lots of ipv6 error that is weird because i disable ipv6 in my kernel. When i re enable all ipv6 i still got https://pastebin.com/raw/yZjFF25X 1516088843 M * Aiken suppose it works for me does not help 1516088876 M * Aiken this is the config on the machine I was testing 4.9.76 on http://rilapy.jbmb.net/vserver/config-4.9.76 1516088900 M * Aiken my connection is dual and ipv6 was working 1516089364 M * Ghislain ok thx, i probably messed up somewhere :) 1516090436 M * Aiken if you want I could try compiling with your config to see what happens 1516090704 M * Ghislain well i dont want you to loose time but if you have any here it is:https://pastebin.com/raw/bxn2sEFt 1516090780 J * nikolay ~nikolay@149.235.255.3 1516090878 M * Aiken at the moment doing a good job of wasting time, just had a kernel compile blow up with No space left on device 1516090890 M * Ghislain got that too ;p 1516091718 M * Aiken some carefully placed rm -rf dealt with the space problem. Just started a compile with your config 1516091727 M * Aiken gentoo 1516092183 M * Ghislain thanks, i use debian mkkpkg on my side, i just switched to gcc8 so perhaps it is a little itchy for some change 1516092216 M * Ghislain i will have you sign a non disclosure agreement on any silly config mistake i could have done 1516092272 M * Aiken gcc 6.4.0 1516092534 N * Bertl_zZ Bertl 1516092537 M * Bertl morning folks! 1516092572 M * Bertl Ghislain: I did a test build of 4.9.x with and without ipv6 yesterday and both went fine, can you provide the config which is failing for you? 1516092601 M * Ghislain https://pastebin.com/raw/bxn2sEFt but this is probably one of my silly mystakes 1516092611 M * Ghislain aiken is trying it too 1516092630 M * Aiken it went boom 1516092648 M * Ghislain oh so this is my config that is not liked, what error did you get ? 1516092721 M * Aiken looks like same error as you starting with In file included from drivers/net/ethernet/broadcom/cnic.c:41:0: 1516092781 M * Aiken https://pastebin.com/2sXzXZqb 1516092841 M * Bertl why does the broadcom driver include ip6_route.h? 1516092855 M * Ghislain well you tell me ! :p 1516092876 M * Ghislain and my servers use bnx2 so i cannot remove them. 1516092919 M * Bertl can you try just commenting out that include for a test? 1516092926 M * Ghislain yep 1516092930 M * Aiken https://pastebin.com/SEdJ6dc1 1516093038 M * Ghislain can i put a #if IS_ENABLED(CONFIG_IPV6) 1516093039 M * Ghislain before the ipv6 include and en #endif after ? 1516093054 M * Bertl just use // before the include 1516093069 M * Bertl it's just a test, proper fix will follow depending on the result 1516093155 M * Ghislain there is ipv6 reference everywhere in the code of this driver 1516093599 M * Ghislain got the same error on another driver 1516093633 M * Bertl so, removing the include fixed it for this driver? 1516093634 M * Aiken trying your config on a vanilla 4.9.76 1516093652 M * Bertl no need, it is related to a Linux-VServer change 1516093677 M * Ghislain it seems let me recheck all the errors 1516093733 M * Ghislain yes i do not see the cnet.c i have a bunch of others now 1516093741 M * Bertl I presume it 'works' because IPv6 is disabled in your config 1516093832 M * Ghislain https://pastebin.com/raw/nzjfDBgr 1516093876 M * Ghislain well as i do not use ipv6 yet i disabled it, ipv6 is so 2012 you know 1516094567 M * Bertl you are using TPM? 1516094618 M * Ghislain no but it is surely left in the config 1516094635 M * Ghislain CONFIG_TCG_TPM=y 1516094677 M * Bertl so what are you doing with the integrity framework? just curious 1516094717 M * Ghislain i do not use TPM that i know of, this is just a default config 1516094732 M * Bertl ah, you are building a kitchen-sink kernel 1516094769 M * Ghislain well i took the default one and disabled quite some thing like sound, ipv6, PATA, 100mbps network card etc.. 1516094804 M * Bertl in general, it is way better to work the other way around 1516094818 M * Ghislain if you see silly thing tell me i will add them to the list of thing to remove 1516094819 M * Bertl i.e. start with an allnoconfig and enable what you actually need 1516094825 M * Ghislain does it hinder anything ? 1516094831 M * Aiken the kernel config I posted earlier is mostly just enough to run most things here 1516094845 M * Ghislain well i do not know a lot about the kernel and i fear i will forget essential bits 1516094855 M * Aiken do not remember if it started as a allnoconfig or was very agressive with disabling things 1516094866 M * Bertl it just adds bloat and complexity to have stuff you don't need and often it causes potential attack vectors which can be easily avoided 1516094884 M * Aiken configuring kernels I have spent a lot of time staring at lspci and lsusb 1516094932 M * Bertl worst case scenario is you have to try a few times till it works (if you forget critical stuff) 1516095074 M * Aiken your config compiles after enabling ipv6 1516095264 M * Ghislain the noconfig remove only drivers or all ? including the weird cpu related optimisation and memory and pci setting and apci etcc.. 1516095373 M * Ghislain it remove all..so for exemple should i use membarrier call and madvise/fadvise, no fricking idea :) 1516095643 M * Bertl that's what the help is for 1516095672 M * Bertl you also get a 'default' option in case the help doesn't help you :) 1516095713 M * Aiken what about defconfig ? 1516095753 M * Bertl kind of okay, but still enables a lot of stuff you do not necessarily need 1516095790 M * Bertl but I guess starting with defconfig and eliminating stuff is a reasonable approach too 1516096339 J * romster ~romster@158.140.215.184 1516096465 M * Bertl btw, is hoppy still 'alive'? 1516096547 Q * romster_ Ping timeout: 480 seconds 1516097419 M * Bertl Ghislain: http://vserver.13thfloor.at/Experimental/delta-ipv6-fix04.diff 1516097426 M * Bertl this should fix the IPv6 issue 1516097571 M * Aiken the name lives on as a test minecraft server but the alpha is somewhere in the house collecting dust 1516097624 M * Bertl what a shame ... http://irc.13thfloor.at/LOG/2005-06/LOG_2005-06-16.txt 1516097691 M * Aiken the flintstones theme never changed 1516098552 M * Aiken from memory I was starting to have trouble with the alpha with kernel size getting a bit big 1516098560 M * Aiken life getting in the way did not help 1516099747 J * Gremble ~Gremble@cpc1-aztw34-2-0-cust397.18-1.cable.virginm.net 1516106438 M * Ghislain Bertl: trying the patch right now 1516108642 M * Ghislain failed with the same error i do a rm -Rf of the tree and start over to be sure i have not done some silly stuff 1516108647 M * Ghislain cd 1516113429 M * Bertl off for now ... bbl 1516113435 N * Bertl Bertl_oO 1516115969 J * kimel^ ~kimel^@7MPAAAF7R.tor-irc.dnsbl.oftc.net 1516115986 M * kimel^ Hello 1516116010 M * Ghislain hi kimel 1516116022 M * Ghislain Bertl_oO: ok with the patch it compiled fine 1516116063 M * kimel^ Is this chat for asking questions about linux in general? 1516116120 M * Ghislain not really, this is a chan that speak of a virtualisation technology called "linux vserver" 1516116166 M * gnarface kimel^: there's one on freenode, but you're better off picking a distro and going to their channel directly instead 1516116168 M * kimel^ Ok, that means I'm a fool for a minute XD. Thanks anyway! 1516116185 P * kimel^ 1516116341 Q * obeardly Remote host closed the connection 1516116583 M * Ghislain the kernel compile but the server do not reboot so i guess i deactivated a little too much things 1516117604 Q * Gremble Remote host closed the connection 1516118113 M * Ghislain oh its back, humm it perhaps fsck after the last 34803284 reboots 1516118186 M * Ghislain daniel_hozac: fun fact util vserver compile on a debian jessie but fail on a debian buster 1516119133 M * Ghislain daniel_hozac: here are all the warnings and the error, gcc8 this time on debian buster: https://pastebin.com/raw/sxUztG7x 1516119639 J * hjf288 ~smuxi@cpc109027-salf6-2-0-cust228.10-2.cable.virginm.net 1516119712 M * hjf288 o/ A couple of months back I was asking about an issue I was seeing with VServer and the patches applied to the latest 3.18 kernel at the time where /dev/null in my guests was being replaced - I've tried the latest BenG kernels and having the same issue 1516119789 M * hjf288 I believe the issue maybe related to "monit" as I've noticed that disabling the init script, recreating the character device and restarting the vserver guest, stops experiencing this behaviour 1516119837 M * hjf288 I noticed that https://pastebin.com/q64ns3V4 is displayed in the host syslog upon start as well 1516119878 M * hjf288 I don't suppose anyone would be able to try replicate this issue or be able to offer assistance in resolving it please 1516119931 M * Ghislain i have this sporadicly /dev/null and i bet yes there is a script that destroy this device and makes it a simple file. never found the culprit but replacing the device make all work again 1516119971 M * Ghislain for the proc message it is normal, i guess monit try to reach forbiden proc. You have then 2 solution, use the vprocunhide to let it see them or ignore those messages 1516120008 M * hjf288 It doesn't seem to be sporadic here - it's everytime the guest starts 1516120013 M * hjf288 when monit is enabled 1516120015 M * Ghislain oh ! 1516120046 M * Ghislain so you should check /dev/null in the start scripts to see if it mess with it 1516120059 M * Ghislain after, if it is the daemon itself then will be harder 1516120089 M * Ghislain bandaid would be to start after monit a script that recreates it 1516120107 M * hjf288 Well what confuses me is that the issue doesn't occur on an older kernel build - 3.18.11-vs2.3.7.4 1516120218 M * hjf288 and i don't get the monit inode syslog messages on the older kernel build 1516120236 M * Ghislain you've gone from wich to which version ? 1516120278 M * hjf288 from 3.18.11-vs2.3.7.4 to 3.18.91-vs2.3.7.5 1516120288 M * hjf288 However I have the same issue on the 4.1 kernels 1516120313 M * hjf288 I also had the same issue when I applied the vs2.4.7.4 patch to 3.18.55 (if i remember the version correctly) 1516120502 M * Ghislain and removing monit solve the thing completly ? 1516120565 M * hjf288 If I -x permissions from the monit init.d script to stop it from starting then /dev/null is fine 1516120585 M * hjf288 on the newer kernels 1516120603 M * hjf288 the older one I'm using has no issues nor does it print the inode messages I posted above 1516120627 M * Ghislain perhaps monit detect a new feature in the kernel that triggers something 1516120679 M * Ghislain you could prevent auto start 1516120716 M * Ghislain the strace monit 2>&1 |tee /root/monit.txt and see if it touches /dev/null and how 1516120724 M * Ghislain s/the/then/ 1516120791 M * Ghislain bert_oO, daniel_hozac: , why do we use dietlibc ? 1516120950 M * hjf288 I know it touches /dev/null as I greped the src and narrowed it to 3 files: https://pastebin.com/ikLu3Hb3 1516121069 M * hjf288 the guests can't have mknod capability so I can't just fix them in the guest and IIRC correctly last time I looked at this, even if I do recreate them - they will go missing again at some point :/ 1516121118 M * Ghislain you can use post post start perhaps 1516121185 M * Ghislain post-start 1516121186 M * Ghislain The scriptlet which will be executed after the vserver has been started. Before executing the script, the vserver root directory will be made the working directory. 1516121191 M * Ghislain bandaid but... 1516121212 M * Ghislain must go, ++ guys and good luck 1516121241 M * Ghislain /etc/vservers/vserver-name/scripts/post-start cf http://www.nongnu.org/util-vserver/doc/conf/configuration.html 1516121248 M * hjf288 I believe I tried that last time 1516121281 M * hjf288 and it wasn't practical 1516122006 Q * nikolay Quit: Leaving 1516122149 M * hjf288 Also the monit version is the same - 5.14 (Released on 9 June 2015) whereas kernel 3.18.55 was May 17th 2017. I've also tried an up to date version of monit 1516122229 M * hjf288 both versions are fine on the older vserver kernel and have the dev/null issue on the newer vserver kernels 1516122246 M * hjf288 which makes me skeptical as to it taking advantage of newer kernel features 1516123105 J * obeardly ~obeardly@12.153.3.34 1516123610 Q * druschka_domaintechnik Quit: druschka_domaintechnik 1516125007 J * Spookyneedles ~mind-ctrl@5PUAAAMSS.tor-irc.dnsbl.oftc.net 1516125491 Q * Spookyneedles autokilled: This host violated network policy. Mail support@oftc.net if you think this is in error. (2018-01-16 17:58:10) 1516127933 J * druschka_domaintechnik ~druschka@85.118.185.154 1516134927 M * Bertl_oO hjf288: interesting ... 1516134979 Q * obeardly Quit: Leaving 1516134993 M * Bertl_oO I did a quick check and there are no functional differences in the patch between 3.18.7-vs2.3.7.4 and 3.18.55-vs2.3.7.5 1516135037 M * Bertl_oO so whatever triggers your tool must be a mainline change 1516135897 J * obeardly ~obeardly@12.153.3.33 1516136933 M * Bertl_oO note that protecting the /dev directory should prevent any tool from removing entries there, so that might be a good thing to do in your case 1516137818 M * hjf288 Bertl_oO: Thanks, would you be able to advise the suitable manner on doing so? Would need stuff to be able to redirect to null but nto remove it 1516137837 M * hjf288 the inodes denied flagging in syslog had me curious 1516137845 Q * druschka_domaintechnik Quit: druschka_domaintechnik 1516137868 M * Guy- hjf288: chattr +i on dev/. should do it 1516137894 M * Bertl_oO well, a guest is allowed to remove dev entries by default (no harm in that, security wise) but typically it is not allowed to create device nodes, that's the reason for the messages 1516137897 M * Guy- but I think it should also work if you mount it readonly 1516137923 M * Bertl_oO probably should be enough to remove the 'w' flag from /dev 1516137953 M * Bertl_oO I really doubt that the tool will override the /dev dir permissions, but maybe it will, in which case the +i might be required 1516138420 M * hjf288 ok great, i'll give that a go :) 1516138480 M * hjf288 just wondering why the messages never appeared on the older release but if it's a mainline change to the behavior might explain it 1516138540 M * Bertl_oO what's the exact message you are seeing? 1516139077 M * hjf288 https://pastebin.com/q64ns3V4 1516139082 M * hjf288 just that 1516140242 M * Bertl_oO can you do a 'find /proc/ -inum 4026531846 2>/dev/null' for me on the host? 1516140476 M * hjf288 I'm afraid I'll have to check tomorrow as I don't have access to the machine atm 1516140500 M * hjf288 I did try to find the inode but didn't get a hit although I was juggling alot of stuff at the same time 1516140511 M * Bertl_oO no problem 1516140708 Q * obeardly Quit: Leaving 1516142218 Q * Aiken Remote host closed the connection 1516142329 J * Aiken ~Aiken@2001:44b8:2168:1000:b26e:bfff:fe2a:b951 1516142926 J * obeardly ~obeardly@12.153.3.33 1516142987 M * obeardly gnarface: You around in here? 1516143008 M * gnarface yes 1516143033 M * gnarface obeardly 1516143034 M * obeardly When building my first vserver on the host, where does it store all the guests at? 1516143061 M * gnarface it's either /vservers or /var/lib/vservers, i think, depending 1516143066 M * obeardly I need to know how to partition my 1TB of RAID10 on this server that's going to be a vserver host. 1516143108 M * obeardly depending on what? 1516143128 M * gnarface on how you built util-vserver, i think 1516143152 M * obeardly Well, I haven't done any of that yet. It's a fresh OS install. 1516143314 M * gnarface obeardly: the debian jessie instructions should work, minus the caveat about systemd http://linux-vserver.org/Installation_on_Debian 1516143486 M * obeardly gnarface: Thanks! 1516143558 M * obeardly I'm building the base OS for a production server, but then I'm going to build out on a test box, before I finish the build. 1516144335 M * obeardly In the build instructions, what does this mean: --i-know-its-there 1516144476 M * gnarface dunno 1516144481 M * gnarface but i assume you can ignore that for devuan 1516144543 M * gnarface i guess i haven't done this for a while 1516144613 M * gnarface i didn't use -n either 1516144620 M * obeardly Okay. 1516144769 M * obeardly I assume the / drive on the host can be small >10GB/<40GB? 1516144785 M * gnarface easily 1516144826 M * obeardly Cool. I'll be back with more questions later. Thanks for all of your help. 1516144837 M * gnarface no problem 1516145312 M * obeardly So question, if you build 4 guests on top of vserver, do they all have equal access to CPU/RAM? 1516145412 M * gnarface yea, unless you set quotas yourself 1516145424 M * gnarface same with disk space 1516145460 M * obeardly Can you point me to the docs on how to set quotas? 1516145503 M * gnarface check the FAQ 1516145512 M * gnarface under Resources Usage 1516145530 M * obeardly Thanks again. 1516145536 M * gnarface no problem