Back to Home / #uml / 2007 / 04 / Prev Day | Next Day
#uml IRC Logs for 2007-04-17

---Logopened Tue Apr 17 00:00:38 2007
01:21|-|Urgleflogue [] has quit [Ping timeout: 480 seconds]
01:51|-|shze [] has joined #uml
02:25|-|Ancalagon [] has joined #uml
---Logclosed Tue Apr 17 02:43:52 2007
---Logopened Tue Apr 17 02:45:09 2007
02:45|-|mikegrb [] has joined #uml
02:45|-|Ekipa kanalu #uml: Wszystkich: 33 |-| +op [0] |-| +voice [0] |-| normalnych [33]
02:45|-|Kanal #uml zsynchronizowany w 33 sekundy
03:41|-|motp [~motp@] has joined #uml
04:03|-|shze [] has quit [Remote host closed the connection]
04:48|-|aroscha [] has quit [Quit: aroscha]
04:58|-|motp [~motp@] has quit [Quit: Leaving]
05:06|-|besonen_mobile_ [] has joined #uml
05:09|-|besonen_mobile [] has quit [Ping timeout: 480 seconds]
05:22|-|kokoko1 [~Slacker@] has quit [Ping timeout: 480 seconds]
05:24|-|kokoko1 [~Slacker@] has joined #uml
06:39|-|polyonymous [] has quit [Ping timeout: 480 seconds]
06:50|-|polyonymous [] has joined #uml
08:02|-|Ancalagon [] has quit [Quit: ChatZilla [Firefox]]
08:32|-|Urgleflogue [] has joined #uml
08:32|-|Urgleflogue [] has quit [Remote host closed the connection]
08:36|-|Urgleflogue [] has joined #uml
09:10|-|krau [~cktakahas@] has quit [Read error: Operation timed out]
09:16|-|aroscha [] has joined #uml
09:30|-|krau [~cktakahas@] has joined #uml
09:50|-|dgraves [] has joined #uml
10:12|-|jdike [] has joined #uml
10:12<jdike>Hi guys
10:15[~]caker is stupid
10:15<caker>I aldready knew strip did exactly that :)
10:16<caker>that works fine, btw
10:19|-|silug [] has joined #uml
10:27|-|hfb [] has joined #uml
10:28<jdike>why does my test box have 6G of kernel modules?
10:28<caker>supplied by Fedora, or your own?
10:29<jdike>that might be where the 20G on that partition went
10:29|-|baroni [] has quit [Remote host closed the connection]
10:29<jdike>about to find otu
10:29|-|baroni [] has joined #uml
10:31<jdike>should probably get rid of some of the dozen or so test kernels on that box
10:31<jdike>plus associated modules
10:31<caker>... one of my favorite things to do
10:32<jdike>starting with 2.6.11
10:54<jdike>hey dude
10:54<dgraves>jdike: we had 14G of core files on one of our boxes. :)
10:54[~]kokoko1 fighting spam :(
10:55<kokoko1>boss is upto my ass
10:55<jdike>and how did you discover them
10:55<jdike>my box wouldn't upgrade for the last week and I couldn't figure out why
10:55<jdike>yum just silently exited with status 1
10:55<dgraves>kokoko1: that's an issue. :)
10:55<dgraves>jdike: our builds started failing. people yelled at me. i poked around.
10:56<kokoko1>dgraves, yeah you now how to added SPAM-x headers using amavid-new ?
10:56<kokoko1>oh sorry
10:56<kokoko1>off topic
10:56<dgraves>kokoko1: blurk bllink teepee fire gargle stink!
10:56[~]kokoko1 hides
10:56<dgraves>hah. if people got in trouble for offtopic, i'd be fired.
10:56<dgraves>just ask jdike.
10:57<jdike>he's been fired for ages, he just won't go away
10:58<dgraves>i bbbabelieve you h--h-have muh-muh-my s-st-s-stapler...
10:58<jdike>and that peech imspediment
10:59[~]caker burns down the building
10:59[~]dgraves goes to enjoy his island. :)
10:59<kokoko1>heh you got an island?
11:00<kokoko1>man you are rich
11:00<jdike>in the middle of the Merrimack River
11:00<jdike>when the water is low
11:00<jdike>which it isn't right now, I bet
11:00|-|cmantito [] has joined #uml
11:04<caker>Caker can set ants on fire with a magnifying glass. At night.
11:04<caker>Caker can count backwards from infinity.
11:05<caker>Oxygen requires Caker to live.
11:07<caker>Jdike can slam revolving doors.
11:16<jdike>it's my claim to fame
11:17<caker>The chief export of jdike is pain
11:17<caker>oldies but still fun
11:37|-|silug [] has quit [Ping timeout: 480 seconds]
11:45|-|kos_tom [] has joined #uml
11:56|-|krau [~cktakahas@] has quit [Ping timeout: 480 seconds]
11:56|-|aroscha [] has quit [Quit: aroscha]
12:07|-|nikkne [] has joined #uml
12:21|-|aroscha [] has joined #uml
12:22<aroscha>jdike: i wanted to ask you if my theory concerning memory usage is correct
12:22<aroscha>jdike: can I safely assume that each instance consumes the amount of memory that it really consumes , or should I assume that it consumes the full amount of mem=xM that I specify in the command line
12:27|-|ram [] has joined #uml
12:28<aroscha>the docu was not quite clear for me on this topic
12:31<caker>aroscha: a simple test run would show you
12:32<aroscha>according to a test run with 400 UMLs I lose quite a lot of virtual memory but RSS stays quite small
12:32<caker>are you using tmpfs for the UML's memory file
12:33<caker>in that case, the (unlinked) memory file grows up to the max memory that the UML kernel is using
12:33<aroscha>but that sounds not so bad I guess
12:36<dgraves>jdike: there's a river where my roads are.
12:37<dgraves>in fact, they closed off my road yesterday.
12:37<jdike>aroscha, it consumes what it uses
12:37|-|pgstudy [] has joined #uml
12:38|-|hojuruku [] has quit [Ping timeout: 480 seconds]
12:41|-|pgstudy [] has quit []
12:51<aroscha>jdike: thx! perfect
12:52<dgraves>yes, yes, jdike is perfect.
12:52<aroscha>jdike: so , seems we can really scale up then. Hm, let's see if we can improve VNUML :)) And yes, jdike is a perfect help on IRC. Thanks a lot.
13:25|-|nikkne [] has quit [Ping timeout: 480 seconds]
13:29|-|silug [] has joined #uml
13:31<BillyCrook>yeah, thanks from here too for the idea yesterday on bridging nics
13:31|-|krau [~cktakahas@] has joined #uml
13:56|-|dominik [] has joined #uml
14:09|-|blackasylum [] has joined #uml
14:19<aroscha>i tried to compile the UML kernel without debuging infos (make ARCH=um etc) . Did not work. Is that a bug or a feature?
14:20|-|dominik [] has quit [Remote host closed the connection]
14:43<jdike>not a feature
14:44<jdike>what's the error?
14:44|-|blackasylum [] has quit [Quit: BitchX: now with 42 percent more random quit messages!]
15:00<dgraves>its a fug!!!
15:02<jdike>who let him in
15:05<aroscha>jdike: ok, I will reproduce it hang on
15:08|-|kos_tom_ [] has joined #uml
15:09|-|kos_tom_ [] has quit [Remote host closed the connection]
15:10<aroscha> MODPOST vmlinux
15:10<aroscha>WARNING: vmlinux - Section mismatch: reference to .init.text:free from .bss between 'stderr@@GLIBC_2.2.5' (at offset 0x60224380) and 'stdout@@GLIBC_2.2.5'
15:10<aroscha>WARNING: vmlinux - Section mismatch: reference to .init.text:parse_header from .bss between 'stdout@@GLIBC_2.2.5' (at offset 0x60224388) and 'completed.6100'
15:10<aroscha> LINK linux
15:12<dgraves>jdike: your bouncer bounced.
15:12<aroscha>what i did is i simply disabled the kernel hacking features in make menuconfig
15:15<jdike>those aren't errors
15:15|-|Urgleflogue [] has quit [Read error: Connection reset by peer]
15:15<jdike>they are WARNINGs
15:15<jdike>and as far as I can see, they are bogus
15:17[~]dgraves watches jdike throw down on the errors.
15:17[~]jdike places ads for a new bouncer
15:19|-|Urgleflogue [] has joined #uml
15:21<aroscha>yup, you are right
15:22<aroscha>hm... I was to quick with copy&paste of the "error" message :)) i had some errors before... I will try to recreate them. The guest also segfaulted then. But... well, will try to recreate
16:08|-|mjf [] has joined #uml
16:08|-|mjf [] has quit []
16:10|-|kos_tom [] has quit [Ping timeout: 480 seconds]
16:22<aroscha>jdike: btw: running 1000 instances now on a single uml_switch
16:22<jdike>that was the goal, right?
16:22<aroscha>and i could not reproduce my experience with the kernel hacking options
16:23<aroscha>yes! well 2000
16:23<aroscha>but 1000 is also fine
16:23<jdike>well, you're halfway there
16:23<jdike>how's the host's memory holding out?
16:24<aroscha>no problem
16:24<aroscha>root@texas:~# free
16:24<aroscha> total used free shared buffers cached
16:24<aroscha>Mem: 8180348 8051216 129132 0 235656 6956908
16:24<aroscha>-/+ buffers/cache: 858652 7321696
16:24<aroscha>Swap: 3004112 0 3004112
16:24<aroscha>0 swap
16:25<jdike>almost all of it is page cache
16:26<aroscha>but I do notice a message "send_sock sending to fd4 Resource temporarily unavailable" on the same xterm where i started uml_switch
16:26<aroscha>many many of these messages
16:26<jdike>yeah, comment it out if you don't like it
16:27<aroscha>so that means nothing.
16:27<jdike>actually, you might want to fix it at some point
16:27<jdike>the switch uses a single socket to send out all packets to all instances
16:27<aroscha>what is the cause?
16:27<jdike>when that fills up, you get those messages
16:28<jdike>one socket per port might be a better plan
16:28<aroscha>i guess so
16:28<aroscha>or a really high backlog :)
16:28<jdike>hehe, yeah
16:28<aroscha>ah! and the stuff that i am simulating actually is full of broadcasts
16:29<aroscha>hm, means the packets don't get sent to the guests which connect to that overloaded switch?
16:29<jdike>could be
16:31<aroscha>in that case i have to do something there
16:43<caker>No two snowflakes are alike thanks to an algorithm written by Jdike.
16:44<aroscha>ok, starting 2000 now
16:44<aroscha>some crashed
16:45<aroscha>but memory is still ok
16:45<aroscha>load avg 241 ;-)
16:45<aroscha>oops > 300
16:46<aroscha>hehe and still starting instances
16:46<aroscha>and now my macbook 's fans started because there is so much happening in the xterms haha!
16:53|-|mjf [] has joined #uml
16:53|-|mjf [] has quit []
17:01|-|aroscha [] has quit [Quit: aroscha]
17:04|-|kokoko1 [~Slacker@] has quit [Ping timeout: 480 seconds]
17:05|-|kos_tom [] has joined #uml
17:09|-|alb [] has quit [Ping timeout: 480 seconds]
17:18|-|aroscha [] has joined #uml
17:21|-|BillyCrook [] has quit [Ping timeout: 480 seconds]
17:25<caker>jdike can divide by zero.
17:33<mikegrb>jdike can boot in 3 ms
17:34|-|kos_tom [] has quit [Quit: I like core dumps]
17:41<aroscha>does linux have something like the BSD netgraph framework?
17:44<aroscha>jdike: and basically in uml_switch you really re-programm the complete functionality of a switch?
17:44<aroscha>or a hub
17:59<caker>jdike: have you noticed anything weird about loadavgerages in recent 2.6-um kernels. That and/or process accounting -- people report stuff like Apache shows many more threads in ps/top than it does under 2.4 (and older 2.6 kernels, pre-NPTL support, I think)
17:59<caker>weird loadaverages as in higher than previous 2.6/all 2.4-um kernels
18:04|-|albertito [] has joined #uml
18:11|-|alb [] has joined #uml
18:11|-|albertito [] has quit [Quit: Reconnecting]
18:12<jdike>there shouldn't be any differences like that
18:12|-|alb changed nick to albertito
18:12<jdike>aroscha, uml_switch has basic switch and hub functionality
18:12<jdike>no STP, for example
18:13<aroscha>ok got it. Well, I will try with a huge backlog now. Let's also see what setsockopt offers
18:16<aroscha>hm... i guess for good network simulations we will need better virtual switches/hubs/routers
18:19<caker>Jdike can win a game of Connect Four in only three moves.
18:19<caker>ok, I'll stop now.
18:19<caker>jdike: :)
18:19[~]caker blows kisses
18:21<aroscha>jdike: yes, by the way uml_switch does lose packets. Many in fact! I can see that because the routing daemon that I want to test only sees a few other instances but not nearly as many as I started
18:22<aroscha>ok, more tomorrow. it is late in CET+1 tz
18:22|-|aroscha [] has left #uml []
18:22|-|aroscha [] has joined #uml
18:32|-|baroni [] has quit [Remote host closed the connection]
18:34|-|hfb [] has quit [Quit: Leaving]
18:36|-|baroni [] has joined #uml
19:01|-|jdike [] has quit [Quit: Leaving]
19:23|-|baroni [] has quit [Ping timeout: 480 seconds]
19:25|-|baroni [] has joined #uml
19:33|-|baroni [] has quit [Ping timeout: 480 seconds]
19:35|-|baroni [] has joined #uml
20:17|-|wyvern [] has joined #uml
20:18<wyvern>Where are the syscall numbers for UML defined? arch-um/unistd.h doesn't have them like it does for other arch's
21:04|-|ram [] has quit [Ping timeout: 480 seconds]
22:59|-|VS_ChanLog [] has left #uml [Rotating Logs]
22:59|-|VS_ChanLog [] has joined #uml
---Logclosed Wed Apr 18 00:00:15 2007