1181780571 M * coderanger So what is a valid argument to vc_create_ctx?
1181780578 M * coderanger er, vc_ctx_create
1181780605 N * Bertl_oO Bertl
1181780626 M * Bertl coderanger: assuming the latest version, a flagmask
1181780667 Q * FireEgl Read error: Connection reset by peer
1181780677 M * Bertl coderanger: you know vcmd by now?
1181780687 M * coderanger It claims that it takes an xid
1181780707 M * coderanger vcmd?
1181780847 M * Bertl http://linux-vserver.org/VCMD_HowTo
1181780864 M * Bertl vcmd is a hack tool, which implements all the kernel syscall commands
1181780872 J * zLinux ~zLinux@88.213.35.226
1181780878 M * coderanger Where do I get it?
1181780900 M * Bertl http://vserver.13thfloor.at/Experimental/TOOLS/
1181780939 M * Bertl you can also use it to show you the arguments (struct) for a specific syscall command
1181780963 M * coderanger Cool, thanky
1181780992 M * Bertl np, take a look at the examples, they will give you an idea how to create a process context
1181781202 M * coderanger vc_ctx_create((1<<34)+(1<<33)+(1<<32)) == VC_NOCTX
1181781334 M * Bertl you need top specify a context id too
1181781360 M * Bertl which is not part of the struct, but a syscall argument
1181781378 M * coderanger The function only takes one arg though
1181781407 M * coderanger This is using the libvserver function, not a raw syscall
1181781409 M * Bertl then it is an implementation of the previous (older API)
1181781427 M * coderanger Lovely, so we should just do a syscall ourselves?
1181781453 M * Bertl you have to chat with daniel_hozac about that, I do not know much about libvserver
1181781459 M * coderanger Okay
1181781462 M * coderanger daniel_hozac: Ping
1181781649 J * FireEgl FireEgl@2001:5c0:84dc:1:ff:39ff:fed1:1f2d
1181781689 M * coderanger Bertl: The libvserver function vc_ctx_create seems like it has data hardcoded as 0
1181781720 J * DoberMann_ ~james@AToulouse-156-1-162-230.w90-38.abo.wanadoo.fr
1181781734 M * Bertl coderanger: which would point to the v0 interface
1181781746 M * coderanger Bertl: Yep, VCMD_ctx_create_v0
1181781822 M * Bertl well, not directly legacy, but not the latest API either
1181781828 Q * DoberMann[ZZZzzz] Ping timeout: 480 seconds
1181782015 M * coderanger So what do you pass as the xid arg to create?
1181782020 M * coderanger Is it the xid you want to making
1181782026 M * coderanger (and if so, how do you pick those)
1181782039 M * Bertl yep, where 0 and 1 is special  ...
1181782048 M * coderanger Ahh, so I should try 2
1181782051 Q * rgl Quit: Leaving
1181782058 M * Bertl for compatibility reasons you should pick something between 2 and 49151
1181782075 M * Bertl although recent versions allow up to 65535
1181782104 M * coderanger Okay, 2 works
1181782120 M * coderanger We had tried 0, 1, and then some really big numbers (-1 and such)
1181782134 M * coderanger Is there a way to ask for "the next free xid"?
1181782182 M * Bertl no, that is userspace 
1181782195 M * Bertl although you can check for 'used'
1181782243 M * coderanger Okay,
1181782358 M * Bertl we had dynamic context numbers (49152-65535) but this was too racy ...
1181782378 M * Bertl context setup is quite complicated and locking the kernel for a long time is not an option
1181782715 M * coderanger Ahh, so thats the DYNAMIC_XID business
1181782822 M * coderanger Bertl: Thanks again
1181782826 A * coderanger offlines
1181782834 M * Bertl np
1181783243 Q * meetra Ping timeout: 480 seconds
1181783290 Q * Aiken Quit: Leaving
1181783306 J * Aiken ~james@ppp121-45-192-48.lns1.bne1.internode.on.net
1181783309 Q * coderanger Ping timeout: 480 seconds
1181784483 Q * FireEgl Quit: Bye...
1181785863 J * infowolfe_ ~infowolfe@c-67-164-195-129.hsd1.ut.comcast.net
1181786264 Q * infowolfe Ping timeout: 480 seconds
1181792186 Q * meandtheshel1 Quit: Leaving.
1181793154 Q * pusling Remote host closed the connection
1181793220 J * pusling pusling@88.212.70.38
1181794714 M * Bertl okay, off to bed now .. have a good one everyone! cya!
1181794720 N * Bertl Bertl_zZ
1181799779 J * ktwilight ~ktwilight@166.199-66-87.adsl-static.isp.belgacom.be
1181800097 J * sharkjaw ~gab@158.36.45.236
1181801618 N * DoberMann_ DoberMann
1181802042 Q * ktwilight Quit: dead
1181802931 N * DoberMann DoberMann[PullA]
1181803013 Q * harry Ping timeout: 480 seconds
1181804038 J * harry ~harry@d54C2508C.access.telenet.be
1181804119 Q * mountie Ping timeout: 480 seconds
1181804451 J * mountie ~mountie@CPE000f66950c89-CM000a739acaa4.cpe.net.cable.rogers.com
1181805405 Q * ruskie Read error: Connection reset by peer
1181805415 M * eyck_ I think I know why vserver is so much more efficient then vmware in mulit-core scenarios
1181806093 J * FireEgl FireEgl@2001:5c0:84dc:1:ff:39ff:fed1:1f2d
1181806097 J * matthew-_ ~matthew@81.168.74.31
1181806111 M * matthew-_ morning.
1181806126 M * matthew-_ This morning, I notice dmesg containing lines like:
1181806127 M * matthew-_ vxW: xid=105 did lookup hidden ffff81007c41ecf8[#0,4026531918] �/proc/acpi�.
1181806150 M * matthew-_ is this related to vserver? And if so, what does it mean?
1181806459 J * dlezcano1 ~dlezcano1@blueice4n2.uk.ibm.com
1181806535 Q * harry Quit: reboot (2.4.34.4 -> 2.4.34.5)
1181806701 Q * dlezcano1 Read error: Connection reset by peer
1181807120 J * harry ~harry@d54C2508C.access.telenet.be
1181807298 M * Hollow matthew-_: it means that guest with context id 105 tried to access /proc/acpi inside which was hidden
1181807555 J * dna ~naucki@27-241-dsl.kielnet.net
1181807574 J * dlezcano1 ~dlezcano1@blueice3n1.uk.ibm.com
1181807818 M * matthew-_ Hollow: thank you
1181807869 M * matthew-_ right, the time stamps show it was when I was upgrading the guests. So it must have been some debian installation script
1181808007 M * matthew-_ thanks for your help
1181808008 Q * matthew-_ Quit: [BX] Its not TV. Its BitchX.
1181809065 J * Vudumen ~vudumen@perverz.hu
1181814361 J * chand ~chand@212.99.51.254
1181814367 M * matti Morning.
1181814377 M * daniel_hozac morning.
1181814745 M * matti ;)
1181815072 J * ruskie ruskie@ruskie.user.oftc.net
1181815410 J * lilalinux ~plasma@dslb-084-058-199-034.pools.arcor-ip.net
1181816643 J * pmenier ~pmenier@LNeuilly-152-22-72-5.w193-251.abo.wanadoo.fr
1181817002 J * weasel_ weasel@asteria.debian.or.at
1181817002 Q * weasel Remote host closed the connection
1181817011 N * weasel_ weasel
1181818728 Q * Vudumen Read error: Connection reset by peer
1181818729 J * Vudumen 82c340e573@perverz.hu
1181819591 J * _mcp ~hightower@wolk-project.de
1181820039 Q * mcp Ping timeout: 480 seconds
1181820332 Q * _mcp Read error: Connection reset by peer
1181820422 J * mcp ~hightower@wolk-project.de
1181820794 Q * pmenier Read error: Connection reset by peer
1181820979 Q * zLinux Remote host closed the connection
1181821311 Q * mcp Read error: No route to host
1181821377 J * mcp ~hightower@wolk-project.de
1181822963 J * pmenier ~pmenier@LNeuilly-152-22-72-5.w193-251.abo.wanadoo.fr
1181823451 Q * neuralis Ping timeout: 480 seconds
1181824298 J * ktwilight ~ktwilight@254.203-66-87.adsl-static.isp.belgacom.be
1181824472 J * ema ~ema@rtfm.galliera.it
1181826914 Q * sharkjaw Quit: Leaving
1181828653 N * Bertl_zZ Bertl
1181828663 M * Bertl good morning!
1181828669 M * daniel_hozac morning Bertl!
1181828746 M * Bertl how's going?
1181828805 M * daniel_hozac quite alright, you?
1181828840 M * Bertl fine, fine ... just got up, have to get caffeine and sugar levels up first :)
1181828891 M * daniel_hozac hehe
1181829157 Q * Aiken Quit: Leaving
1181829504 J * marcfiu ~mef@aegis.CS.Princeton.EDU
1181830374 M * Bertl welcome marcfiu!
1181831484 Q * FireEgl Quit: Bye...
1181832975 J * FireEgl FireEgl@2001:5c0:84dc:1:ff:39ff:fed1:1f2d
1181834014 Q * ntrs_ Ping timeout: 480 seconds
1181834589 Q * bragon_ Ping timeout: 480 seconds
1181834710 J * ntrs ntrs@68-188-55-120.dhcp.stls.mo.charter.com
1181834739 J * bragon ~bragon@sam.geeknode.org
1181834811 M * marcfiu hey Bertl
1181834813 M * marcfiu any news
1181835095 J * ntrs_ ntrs@68-188-55-120.dhcp.stls.mo.charter.com
1181835095 Q * ntrs Read error: Connection reset by peer
1181835226 M * matti Hi Bertl 
1181835296 J * lastdreamer0 ~andrea@host165-126-dynamic.1-87-r.retail.telecomitalia.it
1181835319 J * ntrs__ ntrs@68-188-55-120.dhcp.stls.mo.charter.com
1181835418 J * awk ~awk@kia.inet-corp.com
1181835433 M * awk hi, any idea when support will be built into the 2.6.20 kernel ?
1181835470 M * mnemoc ?
1181835576 M * lastdreamer0 www.freepenguin.it/index-en.html
1181835576 M * lastdreamer0 www.freepenguin.it/index-en.html
1181835576 M * lastdreamer0 www.freepenguin.it/index-en.html
1181835576 M * lastdreamer0 www.freepenguin.it/index-en.html
1181835576 M * lastdreamer0 www.freepenguin.it/index-en.html
1181835576 F * FloodServ +q *!*@host165-126-dynamic.1-87-r.retail.telecomitalia.it
1181835580 P * lastdreamer0 
1181835703 M * awk idiot
1181835721 M * mnemoc thanks
1181835736 Q * ntrs_ Ping timeout: 480 seconds
1181835799 M * blizz hey
1181835805 M * blizz meh, spambots suck
1181836049 M * Bertl awk: support for?
1181836087 M * awk vserver patch
1181836197 M * Bertl you mean: when will a patch be available? or when will it be included in mainline 2.6.20?
1181836213 M * awk naa, a patch be available
1181836232 M * ard eh
1181836240 M * harry awk: use perl!
1181836240 M * ard they are available from the site?
1181836248 M * ard or am I missing something?
1181836267 A * ard just came back from the datacenter
1181836268 M * awk harry i'd like harry's vserver + grsec patch
1181836278 M * awk err - 1 harry
1181836280 M * harry still /nick perl :p
1181836283 M * ard so I am a little dull right now
1181836293 M * Bertl awk: ah, well, then you have to talk to him
1181836294 M * ard ah the +grsec
1181836300 M * harry yes... talk to me... :)
1181836318 M * matti Heh.
1181836333 M * harry you want a grsec patch for 2.6.20?
1181836340 M * matti Me?
1181836341 M * matti No.
1181836342 M * matti ;]
1181836345 M * harry nono, awk
1181836347 M * awk harry yes.
1181836368 M * matti I am using Solaris at work :<
1181836368 M * harry awk: problem is... there is no official grsec for 2.6.20
1181836377 M * harry matti: slowlaris... that's nice :)
1181836389 M * awk harry aahh, makes sense :)
1181836396 M * harry well.. i could take an unofficial one
1181836401 M * harry but then again... would be useless
1181836405 M * matti harry: Well... Working for corporation makes me down a bit...
1181836408 M * harry since there is no more support for 2.6.20
1181836413 M * harry it's all 2.6.21 and soon 2.6.22
1181836421 M * matti harry: They just want Cisco and Sun and all this and that...
1181836424 M * harry but iirc, there are some vserver probs with 2.6.21
1181836447 M * harry matti: company policy is company policy ;)
1181836480 M * awk harry ahh so you have a vserver + grsec for 21 ?
1181836483 M * matti harry: This sounds like a death sentence almost.
1181836483 M * matti ;]
1181836509 M * matti harry: I don't mind them paying for my training, but I get bored ;]
1181836521 M * matti harry: Everythink works, almost 100% SLA...
1181836521 M * matti ;p
1181836523 M * Bertl harry: no Linux-VServer problems with .21 only mainline issues
1181836525 M * harry 17:53 < harry> but iirc, there are some vserver probs with 2.6.21
1181836534 M * harry ahaa
1181836537 M * harry i'm wrong!!!!!
1181836546 M * matti harry: I'll become the greatest sudoku solver ever... soon ;p
1181836549 M * harry i remember not so correctly then! :)
1181836564 M * harry matti: nice :)
1181836572 M * harry want some 5x5 sudokus? ;)
1181836578 M * harry they are with alphabet :)
1181836583 M * awk ok, so no issue with 21 then?
1181836583 M * matti harry: I switched to hexadocu.
1181836589 M * matti ;]
1181836592 M * matti s/c/k/
1181836597 M * awk harry is it a mission to make a patch?
1181836600 M * harry awk: besides from the ones Bertl has in mainline 2.6.21
1181836609 M * harry awk: it is! :)
1181836613 M * harry i'll do it!
1181836618 M * matti awk: Yes! That's the HOLY MISSION!
1181836624 M * matti awk: We have two goals.
1181836625 M * harry i WOULD prefer to wait for 2.6.22 tough
1181836637 M * harry in the hope that those 2.6.21 issues are solved then..
1181836648 M * harry Bertl: what's the status on those patches/bugs in 2.6.21?
1181836660 M * awk ok kewl
1181836664 M * awk thanks alot..
1181836666 M * awk &
1181836686 M * Bertl matti: http://www-cs-faculty.stanford.edu/~knuth/preprints.html (Dancing links)
1181836729 M * Bertl harry: haven't checked 2.6.21.5 (yet) ...
1181836742 M * harry there is an __init bugfix
1181836751 M * harry and some bluetooth+lvm fixes
1181836763 M * harry (those are backported to 2.4, which i compiled yesterday :))
1181836782 A * harry now supersafe on my server @ home ;)
1181836785 M * harry Linux lucifer.homelinux.com 2.4.34.5-grsec #2 Thu Jun 14 01:51:31 CEST 2007 i586 i586 i386 GNU/Linux
1181836786 M * Bertl for my personal use, I use the backported sysfs patch (which seems to work fine)
1181836813 M * harry vserver would be useless... since i have no real cpu power/use for it... noone has accounts :)
1181836821 M * harry it's a p200, 64MB ram
1181836827 M * harry (yes... still that old box ;))
1181836917 Q * ensc Ping timeout: 480 seconds
1181836925 M * matti Bertl: I am too supid for that. But thanks anyway.
1181836970 M * eyck_ 34.5? oh my, am I out of date
1181836971 M * matti Bertl: In free time I'd practice playing piano :) So it is not so bad with me :)
1181836990 M * Bertl http://sudokugarden.de/en/solve/dancing-links
1181836993 M * eyck_ what's a 'free time' ? 
1181837015 M * matti eyck_: Latest invention.
1181837025 M * harry http://kernel.org/pub/linux/kernel/v2.6/ChangeLog-2.6.21.5
1181837040 M * Bertl harry: congrats! you found the changelog :)
1181837043 M * harry tnx
1181837044 M * matti HEheh
1181837050 M * harry i pasted it, so you can just click and check
1181837057 M * harry saves you 15 seconds!
1181837086 M * harry or 5, depending on your mental freshness!
1181837090 M * harry ;)
1181837103 A * harry proud of himself
1181837534 Q * FireEgl Read error: No route to host
1181837579 J * stefani ~stefani@tsipoor.banerian.org
1181837586 M * Bertl welcome stefani!
1181837593 M * stefani aloha B.
1181837601 M * blizz are there "recommended" example settings for the new token bucket interface somewhere?
1181837625 J * bonbons ~bonbons@2001:5c0:85e2:0:20b:5dff:fec7:6b33
1181837627 M * Bertl there is a wiki page explaining the TB scheduler
1181837644 M * Bertl (gives some examples there)
1181837671 M * blizz hmm, i just need to limit a server which is goin armageddon at the moment ;-)
1181837680 M * blizz (browser game with too much ajax running on it)
1181837705 M * Bertl well, for that the system defaults might be enough
1181837725 M * Bertl they give 25% of cpu to each guest, just enable the hard cpu scheduler
1181837750 M * blizz sched_prio cap?
1181837769 M * Bertl nope, that is the priority scheduler, sched_hard :)
1181837777 M * blizz ahhh, okay :-)
1181837804 M * blizz i really need to talk to the programmer of that php-browser-game thing
1181837873 M * blizz ahh, perfect.. thanks for the fast help, again
1181837884 M * Bertl you're welcome!
1181837984 M * blizz some day i'll ship you and daniel a beer crate of augustiner hell (as long as you're resident in germany and as long as you like augustiner) ;-)
1181838012 M * Bertl unfortunately neither is true ...
1181838016 M * blizz lol
1181838028 M * blizz crap :-)
1181838169 M * blizz i always asume that everybody loves beer
1181838196 M * Bertl can't speak for daniel ... 
1181838228 Q * rob-84x^ Ping timeout: 480 seconds
1181838304 J * ensc ~irc-ensc@p54B4CE15.dip.t-dialin.net
1181838314 M * Bertl wb ensc!
1181838502 M * blizz good to see mysql limited to 25% load
1181838560 M * arachnist np: The Pogues - - Beer, Beer, Beer (Irish drinking song) [0:05/2:02]
1181838566 M * arachnist beer!
1181838678 J * rob-84x^ rob@submarine.ath.cx
1181838687 Q * rob-84x^ 
1181838694 J * rob-84x^ rob@submarine.ath.cx
1181839026 N * phedny_ phedny
1181839121 M * blizz lol
1181839128 M * matti arachnist: :)
1181839132 M * blizz arachnist: is that song royality free? ;D
1181839132 M * arachnist matti: :)
1181839137 M * arachnist blizz: dunno
1181839144 M * arachnist rob-84x^: :)
1181839158 M * harry Beauty Is in the Eye of the Beerholder
1181839168 M * arachnist blizz: probably not
1181839206 M * harry q: when is a person TOO lazy?
1181839218 M * harry like... i could do something for work... 
1181839228 M * harry and... at work, i have a shell opened on that server, as root
1181839238 M * harry if i want to do it now... i have to log in, su to root
1181839239 M * harry then do it
1181839249 M * harry is it bad when i say... naaaaah... it'll have to wait
1181839569 J * FireEgl FireEgl@2001:5c0:84dc:1:ff:39ff:fed1:1f2d
1181839870 M * blizz i want f00:f00:f00:f00...
1181840045 M * rob-84x^ arachnist: hi
1181840251 Q * pmenier Quit: pmenier
1181840854 Q * dlezcano1 Read error: Connection reset by peer
1181841473 Q * chand Ping timeout: 480 seconds
1181842208 Q * lilalinux Remote host closed the connection
1181842740 N * DoberMann[PullA] DoberMann
1181842772 J * _mcp ~hightower@wolk-project.de
1181842772 Q * mcp Write error: connection closed
1181843816 Q * ema Quit: leaving
1181843868 J * onox ~onox@kalfjeslab.demon.nl
1181844020 J * coderanger_ ~laptop@wireless-25.media.mit.edu
1181844022 N * coderanger_ coderanger
1181844951 Q * awk Ping timeout: 480 seconds
1181845108 Q * FireEgl Quit: Bye...
1181845622 J * neuralis ~krstic@solarsail.hcs.HARVARD.EDU
1181846424 J * Darkglow ~pdesnoyer@208.71.184.41
1181846524 J * FireEgl FireEgl@2001:5c0:84dc:1:ff:39ff:fed1:1f2d
1181846555 M * Darkglow Hi, Ive got a weird problem : I installed 2 vservers. Then followed procedure to vhashify. all is fine to this point. I then installed a software and I can see all it's files, but if I run a binary I get "no such file or directory" altho I can view it with let's say less...
1181847090 M * Darkglow got it... damn... 32 bit binaries on a 64 bit platform without the 32b libs...
1181847115 M * Darkglow weird error tho... no such file or directory... anyway, sorry for the spam :-)
1181847117 P * Darkglow Konversation terminated!
1181848198 J * ema ~ema@rtfm.galliera.it
1181852973 Q * _mcp Read error: Connection reset by peer
1181852973 J * mcp ~hightower@wolk-project.de
1181853563 Q * Hunger Remote host closed the connection
1181853786 J * Hunger Hunger.hu@Hunger.hu
1181854866 J * yarihm ~yarihm@84-75-109-39.dclient.hispeed.ch
1181854909 J * chand ~chand@m167.net81-64-156.noos.fr
1181854911 M * yarihm hi everyone
1181854921 M * Bertl wb yarihm!
1181854990 M * yarihm my boss asked me today (we use VServer to provide virtual servers and more demanding hosting environment) whether i could ask if there are any plans for live-migration-features with VServer?
1181855016 M * yarihm in fact he asked way before already, but now i just remember :)
1181855032 M * Bertl not really, we spent some serious thought on that and considered it not worth the efford
1181855065 M * Bertl the problem is, there are actually 3 cases where you want to migrate guests 'live'
1181855107 M * Bertl the first one is on sheduled host downtimes, the second one is to do load balancing and the third would be to increase fault tolerance
1181855153 M * Bertl in the first case, you can achieve the same (or better) by using Xen with a single vserver domain
1181855193 M * Bertl in the last scenario, you have lost already .. as in the failure case, the guest cannot be migrated (unless you keep periodic snapshots)
1181855215 M * yarihm yeah. in our case it would be probably a sub-case of the second one. It's not about live-load-balancing in the conventional sense, but we just move our custmers around when they use too much ressources - to a certain limit at least, if one just really overdoes it, then he is forced to upgrade.
1181855246 M * Bertl the second scenario might look interesting to balance, but, as a matter of fact, the guest migration is quite resource intensive, so balancing out the guests will cause significant overhead itself
1181855300 M * yarihm Bertl: well, fault-tolerance can in that sense not be achieved by any virtualisation-solution i know of, but using DRDB comes close to that. alternatively a different shared-storage backend like NFS or an iSCSI-based cluster-lvm would do the job there too
1181855309 M * Bertl we are more planning to combine Linux-VServer and some kind of cluster system (think mosix) than spending time on migration with limited useability
1181855325 M * yarihm mhm ... 
1181855336 M * yarihm vserver on mosix would be very nice indeed
1181855344 M * Bertl yeah, that is the thing, for fault tolerance, drbd and the current setup is more than sufficient
1181855381 M * Bertl actually the guest restart on recent machines is around a few seconds, so even 'migration' is an option without keeping the apps and connections
1181855388 M * Bertl (at least for most cases)
1181855399 M * yarihm as i said, for us it was more a question of easy ressource-allocation. move guests around if the host is overloaded or something the like
1181855429 M * yarihm we do it in fact using rsync, it works nicely. the question rose because openvz providing the feature
1181855431 M * Bertl well, as I said, doesn't help you much if you add more load by moving guests around
1181855472 M * Bertl and for longer term moves, a scheduled move of a few seconds is probably acceptable
1181855492 M * yarihm well, if you have a single "bad guy" on a server then moving him away and have the both hosts be on even more load is preferred over having the other one stay there and block all other guests the whole night long or whatever
1181855545 M * yarihm yeah, i agree. for me personally (using VServer privately) there is no need for that either, and i think the additional but rather small workload is acceptable for business-cases as well
1181855553 M * Bertl if you have a 'bad guy' shutting down his guest and restarting it on the 'bad guys machine' is probably fine too, no?
1181855567 M * neuralis yarihm: i used openssi to great success for moving around apps that are using too much resources
1181855583 M * yarihm in particular because he is the bad guy ,) normally customers dislike downtime though
1181855586 M * neuralis unfortunately, openssi hasn't been updated past 2.6.11 or so, hp isn't putting resources into it anymore
1181855596 M * yarihm openssi?
1181855609 M * yarihm never heard of that ... i'll have a look
1181855680 M * neuralis it's in many ways a better solution than mosix, imo
1181855690 M * Bertl yep, agreed
1181855733 M * yarihm well, if it is not maintained anymore ... 
1181855777 Q * phreak`` Quit: guess I'll have to quit here and now!
1181855777 M * neuralis there's ongoing community maintenance, but forward-porting it from 2.6.10 or so to a recent kernel is quite a lot of work
1181855792 M * neuralis last i looked, mosix wasn't in much better shape, but i haven't really looked in a while
1181855809 M * yarihm that's BTW an argument against openvz in our particular case: their stable-releases are so old that on newer machines you can hardly use them. bertl does a good job there with vserver, so good that debian even provides vserver-enabled kernels out of the box
1181855872 A * neuralis &
1181855890 M * yarihm well, there the question would be whether the work needed to maintain a mosix- or openssi-cluster would justify the things gained
1181855944 M * Bertl well, you basically have the same efford with a live migration standby, no?
1181856178 M * yarihm i've never set up a mosix-cluster, so i can't tell. we've written scripts that do the migration via rsync. they basically do an rsync, then stop the vserver, do another rsync and start it up on the remote host. Downtime is normally less than 20 seconds. Systems with higher uptime-demands should be redundant anyway IMHO
1181856346 N * DoberMann DoberMann[ZZZzzz]
1181856755 Q * bonbons Quit: Leaving
1181856874 J * Aiken ~james@ppp121-45-192-48.lns1.bne1.internode.on.net
1181857219 Q * dna Quit: Verlassend
1181857960 Q * ema Quit: leaving
1181858005 M * coderanger Bertl: Is there a nice (wiki?) document with all the syscall commands and their args?
1181858336 M * Bertl no, but there is the syscall matrix, and the reference in switch.c as well as in vmcd
1181858341 M * Bertl *vcmd even
1181858464 M * coderanger Bertl: Do we have patches yet? ;-)
1181858526 M * Bertl nothing final yet, still testing
1181858537 M * coderanger Okay
1181858758 M * coderanger Maybe I will start a wiki page as we work on this
1181858765 M * coderanger The basics of the bindings are coming together
1181858795 M * Bertl go ahead, probably the syscall command matrix is a good start
1181858812 M * Bertl (which, IIRC, wasn't ported to the new wiki yet)
1181859967 J * meetra ~xtreme@82.155.145.10
1181859971 Q * onox Quit: leaving
1181862599 J * meandtheshel1 ~markus@85-124-232-67.work.xdsl-line.inode.at
1181862626 J * Piet hiddenserv@tor.noreply.org
1181863938 Q * yarihm Quit: Leaving
1181865003 P * stefani I'm Parting (the water)
1181865089 Q * chand Quit: chand
1181865098 J * dallas ~dallas@sf.newdream.net
1181865157 M * Bertl welcome dallas!
1181865164 M * dallas Hello, thank you
1181865194 M * dallas I have been looking for some information about vsched and changing cpu scheduler limits while also maintaining hard scheduling
1181865210 M * dallas it seems as if when I run a vsched command it switches the guest to priority scheduling.  I could be wrong but it seems so
1181865234 M * Bertl sounds unusual .. what kernel/tool versions?
1181865258 M * dallas kernel is 2.6.19.7, tools are the ones provided by Debian etch.. I'll check version
1181865289 M * dallas  vsched --version
1181865289 M * dallas vsched 0.30.212 -- modifies scheduling parameters
1181865290 M * dallas This program is part of util-vserver 0.30.212
1181865300 M * dallas is that enough info?
1181865309 M * Bertl won't hurt to update that, but IMHO should be fine
1181865326 M * Bertl how do you test/observe the change?
1181865328 M * dallas so am I to understand using vsched should not change the scheduling method?
1181865368 M * Bertl as far as I know, no :)
1181865388 M * dallas a guest set up right after start will have a load of around 5 with the host load at an expected value.  then after running a vsched command the load on the guest starts dropping significantly and the host load goes up
1181865394 M * dallas the guest load will drop down to < 1
1181865403 M * dallas with the same incoming traffic, of course
1181865430 M * dallas I'll show you the command I am running... perhaps I don't need an option
1181865436 M * Bertl okay :)
1181865473 M * dallas vsched --xid whee --fill-rate 7 --interval 32 --tokens 500 --tokens-max 1000 --tokens-min 200 --idle-time --force
1181865507 M * dallas actually... I possibly just solved it for myself.  8-)
1181865538 M * dallas that 7 --fill-rate should be a 1 there... and the --interval should be 8.  I'll try it with the correct value and see what happens
1181865574 M * Bertl you are also setting the --idle-time flag, which might not be your intention
1181865589 M * dallas ah, ok.  I wasn't clear what that was for but it seemed like something I would want
1181865595 M * dallas what does it do?