Back to Home / #uml / 2007 / 04 / Prev Day | Next Day
#uml IRC Logs for 2007-04-25

---Logopened Wed Apr 25 00:00:23 2007
00:45|-|ram [] has joined #uml
00:53|-|ram [] has quit [Remote host closed the connection]
01:01|-|ram [] has joined #uml
03:14|-|shze [] has quit [Quit: Konversation terminated!]
04:08|-|polyonymous [] has quit [Ping timeout: 480 seconds]
04:18|-|polyonymous [] has joined #uml
05:11|-|kokoko1 [~Slacker@] has joined #uml
07:41|-|aroscha [] has quit [Ping timeout: 480 seconds]
07:41|-|baroni [] has quit [Ping timeout: 480 seconds]
07:45|-|aroscha [] has joined #uml
08:01|-|aroscha [] has quit [Ping timeout: 480 seconds]
---Logopened Wed Apr 25 09:38:35 2007
09:38|-|mikegrb_ [] has joined #uml
09:38|-|Ekipa kanalu #uml: Wszystkich: 33 |-| +op [0] |-| +voice [0] |-| normalnych [33]
09:38|-|Kanal #uml zsynchronizowany w 1 sekundy
09:39|-|mikegrb_ Your nick is now mikegrb
09:41<caker>jdike: whatcha hacking on of late?
09:41<caker>btw, we deployed a bunch of new servers with core_pattern goodness
09:42<caker>and sending out a debug-enabled (stripped) -um today ..
09:53|-|baroni [] has joined #uml
10:11|-|hfb [] has joined #uml
10:11|-|hfb [] has left #uml []
10:14|-|the_hydra [~mulyadi@] has joined #uml
10:20|-|wyvern [] has quit [Read error: Operation timed out]
10:20|-|wyvern [] has joined #uml
10:21<jdike>no dumps for me to look at yet>
10:21<caker>not yet
10:22<jdike>what UML version are you using?
10:32<jdike>hacking on merging rc7-mm1 and rc6-mm1 isn't backing out cleanly
10:37[~]jdike grabs full tarball
11:05|-|the_hydra [~mulyadi@] has quit [Ping timeout: 480 seconds]
11:16|-|the_hydra [~mulyadi@] has joined #uml
11:25|-|ram [] has quit [Ping timeout: 480 seconds]
11:34<peterz>jdike: I found my bug the other day, I messed up sys_munmap(), but somehow only UML triggered it.
11:34<jdike>I'd expect that sort of thing with ptrace
11:34<jdike>but not with mmap
11:35<peterz>hehe, yeah, I fully traced all the funny ptrace stuff first, didn't find anything
11:35<jdike>you don't know what UML was doing differently?
11:35<peterz>no, nothing very weird
11:35<peterz>I'm sure other apps would have triggered the same thing
11:36<peterz>but uml has a few clusters of mappings
11:36<peterz>and that triggered it
11:36|-|aroscha [] has joined #uml
11:37<peterz>small mmaps clusters separated by a region > PMD span
11:38<jdike>UML is unusual in mapping lots of single pages
11:39<peterz>yeah, all single page mappings; what happened is that it got the ceiling in free_pmd_range wrong
11:40<peterz>and thereby missed to free the pmd proper
11:40<jdike>that's a confusing piece of code
11:40<peterz>yeah, I know it well now
11:40<jdike>better you than me
11:41<caker>jdike: I've got another hang at "VFS: Mounted root (ext3 filesystem) readonly." for you, WITH the "Host VDSO" patch
11:41<jdike>what kind of debugging is there with this one?
11:41<jdike>I had another report of that but I couldn't get access to the host
11:42<caker>Want me to ask him to dupe his filesyetem for you, so he can lob off any personal stuff?
11:42<caker>Then I can set you up with the works
11:43<jdike>that'll work
11:43<jdike>anything unusual about the fs?
11:45<caker>Ubuntu ... not sure what he's done
11:45<caker>I expect a reply from him shortly
11:46<caker>do you remember which host I had you debugging in last time (perhaps history)
11:47<caker>found it, nm
11:51|-|ram [] has joined #uml
11:57<jdike>unless he made his own init, it'll be an Ubuntu thing
11:58[~]kokoko1 downloading ubuntu-7.04 Feisty
12:00<kokoko1>hi jdike :)
12:00<jdike>you've been quiet lately
12:01<kokoko1>Yes i was off, now i bring my family where I work :)
12:02<kokoko1>jdike, hows UML :) ?
12:03[~]kokoko1 was reading article on kvm lately on
12:03<jdike>been hacking on I/O
12:03<the_hydra>kokoko1: i hope your kids get comfort with the new neighbourhood
12:04<kokoko1>the_hydra, I am still without one :-S
12:04<the_hydra>funny, if I remove "U" in Ubuntu, that means "dead end" in Indonesian
12:05<kokoko1>the_hydra, you tell mate hows life in Indonesian, any plane of flying else where :P ?
12:05<the_hydra>kokoko1: oh you mean planes dropping?
12:05<the_hydra>kokoko1: that stops for a while
12:05<the_hydra>now we have train crash
12:05<the_hydra>quite "amusing" too
12:05<kokoko1>my bad
12:05<kokoko1>typo dude
12:06<the_hydra>yeah, train, planes, ships, you name it, we have it all
12:06<the_hydra>we all plan to use becak instead
12:06<the_hydra>safer...but real slow
12:06<kokoko1>you still didn't get me :)
12:06<the_hydra>no, I was mumbling
12:06<the_hydra>oh plan? my plan?
12:07<the_hydra>no plan to fly elsewhere... I don't want to end up like Munir...
12:07<the_hydra>with arsen..
12:07<kokoko1>I duno they story of Munir?
12:08<the_hydra>kokoko1: anyway, how's work?
12:08<the_hydra>kokoko1: Munir is top person when you speak about human rights here
12:08<kokoko1>the_hydra, work sux, jsut like school sux
12:08<the_hydra>he was the icon
12:09<the_hydra>he mostly messed up with military doodz
12:09<the_hydra>kokoko1: sure
12:09<kokoko1>kinda off here :D
12:09<the_hydra>kokoko1: yeah
12:13|-|kos_tom [] has joined #uml
12:24[~]kokoko1 pokes jdike and disappears
12:28<caker>jdike: 2.86.ds1-6ubuntu32 <-- sysvinit .. want me to roll an fs for you?
12:28<the_hydra>jdike: i found that poking is the new way people say hi these days
12:28<jdike>I guess gug isn't catching on
12:29<jdike>as technically correct as it may be
12:29<the_hydra>jdike: "gug" is really an innovative one
12:30<jdike>sounds like you're drowning
12:30|-|apic [] has quit [Quit: *duck*]
12:44<kokoko1>Does anyone know an url for a linux benchmark that shows relative compile speed differences between processors?
12:45<the_hydra>compile speed? a simple make -j test maybe?
12:45<the_hydra>kokoko1: check LTP (Linux test project)
12:46<the_hydra>but that really depends on what do you compile...
12:47<the_hydra>bunch of inheritance?
12:48<kokoko1>I hope that won't make much difference
12:49<the_hydra>i hope so
12:51<caker>well, crap. I force-downgraded init to his same version, but it doesn't hang
12:51<caker>jdike: said he can dupe a copy of his fs off this evening...
12:52<the_hydra>caker: does init has anything to do with the way a fs is mounted?
12:53<caker>the_hydra: no, but that's what cause the hang last time
12:53<the_hydra>caker: hm ok
12:54<jdike>the fact that the last thing you see is 'VFS mounted...' is irrelevant
12:54<jdike>has nothing to do with the hang
12:54<jdike>all it means is that init is taking infinite segfaults before it managed to print anything
13:03|-|krau [~cktakahas@] has joined #uml
13:04<peterz>kokoko1: if you want to torture c++ compilers, compile some of the boost::spirit examples
13:04<peterz>be sure to have _lots_ of memory
13:06<kokoko1>heh right
13:09<the_hydra>peterz: damn long inheritance?
13:09<peterz>the_hydra: nah, template meta programming
13:09<the_hydra>hm ok
13:10<peterz>c++ templates are turing complete
13:12<peterz>so, much like how you can make the printer calculate your fractal with PS, you can make the compiler write your program
13:12|-|ram [] has quit [Ping timeout: 480 seconds]
13:26|-|aroscha [] has quit [Quit: aroscha]
13:44<caker>I love it.. a copy of his root boots fine
13:47|-|wyvern [] has quit [Read error: Operation timed out]
13:47|-|wyvern [] has joined #uml
13:56<jdike>everything else is the same?
13:58<jdike>you moved to a different host, right?
14:39|-|d-media [] has joined #uml
14:40<d-media>hi every one
14:40<d-media>is the owner of in right now? had a quick question?
14:41<jdike>that would be me
14:43<the_hydra>jdike: suggestion, stick your IRC name clearly in UML web site ;)
14:43<jdike>or people could just ask questions
14:43<jdike>I'm not the only one with answers
14:46<the_hydra>d-media: and your question is?
14:47|-|aroscha [] has joined #uml
14:48|-|baroni [] has quit [Ping timeout: 480 seconds]
14:49|-|baroni [] has joined #uml
14:50|-|Netsplit <-> quits: ElectricElf, linbot, d-media
14:50|-|Netsplit over, joins: d-media
14:50|-|Netsplit over, joins: linbot, ElectricElf
14:53|-|aroscha_ [] has joined #uml
14:53|-|aroscha [] has quit [Read error: No route to host]
14:56|-|aroscha_ [] has quit []
14:59|-|the_hydra [~mulyadi@] has quit [Quit: gtg everyone...]
15:01|-|baroni [] has quit [Ping timeout: 480 seconds]
15:04|-|aroscha [] has joined #uml
15:11|-|baroni [] has joined #uml
15:23|-|aroscha [] has quit [Quit: aroscha]
15:24|-|ram [] has joined #uml
15:57|-|baroni [] has quit [Remote host closed the connection]
15:58|-|baroni [] has joined #uml
16:44|-|aroscha [~aroscha@] has joined #uml
17:01|-|kos_tom [] has quit [Quit: I like core dumps]
17:13<caker>jdike: no, same host that his Linode is on. He duped his root fs and gave it to me
17:14<caker>only real difference is that I'm not feeing in the usual cmdline args (for network, etc)
17:25|-|krau [~cktakahas@] has quit [Quit: Leaving]
17:29|-|d-media [] has quit [Ping timeout: 480 seconds]
17:30|-|weasel [] has quit [Remote host closed the connection]
17:30|-|weasel [] has joined #uml
18:12|-|ram [] has quit [Ping timeout: 480 seconds]
18:23<aroscha>jdike: I will now try with HZ=100 and I also disabled the preemption patch
18:23<aroscha>let's see
18:48|-|apic [] has joined #uml
18:51|-|baroni [] has quit [Ping timeout: 480 seconds]
19:34|-|linbot [] has quit [Read error: Operation timed out]
19:36|-|VS_ChanLog [] has quit [Ping timeout: 480 seconds]
19:37|-|tasaro [] has quit [Ping timeout: 480 seconds]
19:38|-|caker [] has quit [Ping timeout: 480 seconds]
19:40|-|ram [] has joined #uml
19:55|-|strontian [] has quit [Ping timeout: 480 seconds]
20:08|-|caker [] has joined #uml
20:10|-|VS_ChanLog [] has joined #uml
20:13|-|linbot [] has joined #uml
20:16|-|tasaro [] has joined #uml
20:18|-|ram [] has quit [Ping timeout: 480 seconds]
20:29|-|jdike [] has quit [Quit: Leaving]
20:39|-|tasaro [] has quit [Quit: leaving]
20:45|-|tasaro [] has joined #uml
21:06|-|aroscha [~aroscha@] has quit [Ping timeout: 480 seconds]
21:15|-|tasaro [] has quit [Quit: leaving]
21:23|-|aroscha [] has joined #uml
21:43<aroscha>does anybody know by chance if openVZ has individual IP stacks in each instance?
21:45|-|arun [~arun@chobie.cs.Virginia.EDU] has quit [Read error: Connection reset by peer]
21:46|-|baroni [] has joined #uml
21:49<caker>aroscha: I have no clue, but I kinda doubt it based on what I've read about it
21:49<caker>it's basically container-type jails, from what I understand
21:49<aroscha>yes, sounds very much like vserver
21:49<caker>single host kernel
21:49<caker>er, single kernel, period.
21:50[~]caker sees the wheels spinning
21:51|-|tasaro [] has joined #uml
21:53<aroscha>yeah, I just simply see much to much starvation with so many instances, have to do something about that
21:53<aroscha>I was wondering if I could simple make the instances RTPrio level 1
21:53<aroscha>my bash (for stopping them hehe), level 2
21:53<aroscha>and the measurement tools like also level 2
21:54<aroscha>now all level 1 processes get scheduled FIFO round robin
21:54<aroscha>FIFO or round robin
21:54<aroscha>no! sorry, just round robin
21:54<aroscha>with a fixed quantum
21:55<aroscha>that way they will all get a little bit of CPU
21:55<aroscha>hm... question is if that is enough for sending the network packets
21:56<aroscha>but in theory I would get 2GHz/1000 = 1000 x 2MHz Linuxes :)
21:57<aroscha>wel, sure, the customer gets what he pays for, n'est pas? ;-)
21:58<aroscha>(well, for me there is no customer actually, just joking), but I could simulate evenly.. interesting thought. Anything that I am missing with this idea?
21:58<caker>I've never done any of the RT sched stuff, but sounds good to me
21:58<caker>nice won't cut it, eh?
21:59<aroscha>nice helps nothing against RT sched
21:59<caker>I meant with non-rt
22:00<aroscha>ok. Well, the instances are already running with nice. But ... still they unevenly compete for the CPU
22:00<aroscha>meaning my simulation with many instances is quite ... well,... uneven
22:00<caker>daring enough to try the CFS patches?
22:01<caker>heard about it lately?
22:01<caker>not sure what state it's in
22:01<aroscha>no, not yet
22:01<caker>or where it is
22:01<aroscha>no, where?
22:01<caker>but there's been a crapload of conversation on LKML about it lately
22:01<caker>it doesn't use runqueues or timeslices
22:01<caker>it's some wacked out, new fangled technology scheduler
22:02<caker>Let me know how it goes :)
22:03<albertito>aroscha: me too =)
22:19|-|aroscha [] has quit [Quit: aroscha]
22:20|-|linbot [] has quit [Remote host closed the connection]
22:31|-|linbot [] has joined #uml
22:59|-|VS_ChanLog [] has left #uml [Rotating Logs]
22:59|-|VS_ChanLog [] has joined #uml
---Logclosed Thu Apr 26 00:00:40 2007