Back to Home / #uml / 2007 / 05 / Prev Day | Next Day
#uml IRC Logs for 2007-05-30

---Logopened Wed May 30 00:00:46 2007
02:21|-|Ancalagon [] has quit [Quit: ChatZilla [Firefox]]
02:39|-|motp [~motp@] has joined #uml
02:44|-|newb_prog [] has quit [Remote host closed the connection]
03:45|-|newb_prog [] has joined #uml
03:46|-|newb_prog [] has quit []
03:53|-|newb_prog [] has joined #uml
03:54|-|newb_prog [] has quit []
03:55|-|motp [~motp@] has quit [Quit: Leaving]
04:31|-|LemyDanger [] has joined #uml
04:31<LemyDanger>i need some information about the hwclock
04:33<LemyDanger>my gameserver running on uml machine shows me ever twice the latency as the ping is
04:34<LemyDanger>normally the host kernel runs with 250HZ an i compiled it with 1000HZ
04:36<LemyDanger>i dont know how the time is measured, but perhaps thats the problem/solution?
04:37<LemyDanger>hello again
04:38<LemyDanger>perhaps i should mention that the latency should be equal to ping
04:40<LemyDanger>it isnt a hardware bottleneck or so, i tried a ping from within the uml
04:40<LemyDanger>i have installled already ntp and it runs
04:41<LemyDanger>unfortunately i dont know exactly how the halflife counterstrike measures the time
04:41<LemyDanger> .... server
04:44<LemyDanger>the ghosts of uml are sleeping
06:57|-|krau [~cktakahas@] has quit [Read error: Connection reset by peer]
07:05|-|krau [~cktakahas@] has joined #uml
08:22|-|krau [~cktakahas@] has quit [Quit: Leaving]
08:26|-|krau [~cktakahas@] has joined #uml
09:17|-|albertito [] has quit [Ping timeout: 480 seconds]
09:47|-|jdike [] has joined #uml
09:47<jdike>Hi guys
10:01|-|Urgleflogue [] has quit [Ping timeout: 480 seconds]
10:04|-|silug [] has joined #uml
10:10<kokoko1>bonjour jdike
11:35|-|krau [~cktakahas@] has quit [Quit: Leaving]
11:39|-|krau [~cktakahas@] has joined #uml
11:50|-|ram [] has joined #uml
12:14|-|albertito [] has joined #uml
12:18|-|Electric1lf [] has quit [Read error: Operation timed out]
12:24<kokoko1>what would be the advantage to compile eg, squid with-pthreads ?
12:25<dgraves>morning jdike!
12:33<caker>jdike: so, lotsa performance stuff went into 2.6.22-um -- have any gut feel or rough benchmark for its improvement over 2.6.21, say during a kbuild?
12:34<caker>and, is 2.6.23 The Time for SMP?
12:36<kokoko1>hi caker
12:47|-|newb_prog [] has joined #uml
12:47|-|the_hydra [~mulyadi@] has joined #uml
---Logclosed Wed May 30 12:55:31 2007
---Logopened Wed May 30 12:56:27 2007
12:56|-|mikegrb [] has joined #uml
12:56|-|Ekipa kanalu #uml: Wszystkich: 29 |-| +op [0] |-| +voice [0] |-| normalnych [29]
12:58|-|Kanal #uml zsynchronizowany w 155 sekundy
12:58|-|VS_ChanLog [] has joined #uml
13:15<LemyDanger>hello jdike
13:17<LemyDanger>i have a question with the realtime clock
13:18<LemyDanger>the uml with steam half life cs 1.6 server shows ever twice the latency as the ping is
13:18<jdike>OK, what?
13:19<LemyDanger>on the host i compiled the kernel with the 1000HZ option instead of 250HZ
13:19<LemyDanger>i suppose that could be the reason for the wrong latency
13:20<LemyDanger>i guess that
13:20<LemyDanger>i tried a ping to my home computer
13:20<LemyDanger>the ping is 12ms , the latency of the server is 24ms
13:21<LemyDanger>i dont believe that there is a hardware bottleneck or something like that
13:22<jdike>HZ has nothing to do with it
13:22<LemyDanger>my problem is that i dont know how hl server takes the time, and i dont know how the clock in uml works
13:22|-|cmantito [] has joined #uml
13:22<LemyDanger>nothing to do with the kernel frequency? shit
13:23<LemyDanger>i installed ntp and uninstalled hwclock in the uml
13:23<LemyDanger>at the moment i have no idea what else i could do
13:24<jdike>what's the problem?
13:27<LemyDanger>the hl server takes the wrong time . UML time is twice as high as real time
13:28<jdike>As in, sleep 10 takes 20 seconds?
13:28<jdike>or you see 1/2 performance as the host?
13:29<LemyDanger>you coud connect to the server if you want
13:30<LemyDanger>i can try sleep 10 in console ?
13:30<LemyDanger>one moment
13:31<LemyDanger>no , with sleep 10 , 10 are 10
13:32<the_hydra>have you confirmed that with "time"?
13:33<LemyDanger>i wrote
13:33<LemyDanger>sllep 10
13:33<LemyDanger>and waited 10 seconds
13:33<LemyDanger>that goes right
13:34<LemyDanger>is there a way to find out how the hl server takes the time
13:35<LemyDanger>in the internet i found no information about it
13:35<jdike>what's the problem?
13:37<LemyDanger>i have no network problem (100 MBit), the cpu usage is at 12 %, but the server laggs incredible
13:37<LemyDanger>and i dont know why
13:37<jdike>how incredible?
13:39<LemyDanger>the only hint that i have whats going wrong is that server shows me 24ms latency in game, my ping is 12, my friends have a ping of 40, there latency is about 80ms
13:39<jdike>you would expect half that?
13:40<LemyDanger>normally the latency should be 12ms just like the ping 12ms
13:41<LemyDanger>perhaps 3 or 4ms more than ping
13:41<LemyDanger>and my hardware is fast enough
13:41<jdike>is it swapping?
13:42<jdike>how do you know?
13:43<LemyDanger>ok, i m not sure
13:44<LemyDanger>i start the uml with swap_fs, and i have mounted the tmpfs like you say it on your homepage
13:45<jdike>is the host swapping?
13:45<LemyDanger>could plz give me the right command, before i say something wrong
13:46<LemyDanger>i use
13:46<LemyDanger>and it swapped 2 mb
13:46<LemyDanger>on the host
13:46<kokoko1>free -m
13:47<jdike>also you can watch vmstat 1
13:47<LemyDanger>on the uml 0 0 0
13:48<the_hydra>and the host?
13:48<the_hydra>kokoko1: hi
13:48<LemyDanger>on the host 999 2 997
13:51<LemyDanger>vmstat on the um : swpd 0 ; free 410740; buff 3232 ; cache 82036
13:52<kokoko1>the_hydra, hello
14:00<LemyDanger>still there?
14:00|-|Urgleflogue [] has joined #uml
14:20|-|the_hydra [~mulyadi@] has quit [Quit: using sirc version 2.211+KSIRC/1.3.12]
14:23<jdike>what do pings look like?
14:25<LemyDanger>no i go ingame
14:26<jdike>the ping command from the shell
14:26<LemyDanger>in game the latency shown by the server is between 30 and 40
14:27<LemyDanger>the ping gives me 12ms average
14:29<LemyDanger>12 ms
14:30<LemyDanger>from uml host to my home
14:30<jdike>from some other host, ping the host and the UML
14:34<LemyDanger>12 ms to the host
14:35<LemyDanger>the uml cant be pinged directly because of the firewall
14:36<LemyDanger>i do NAT by Port
14:36<jdike>it makes it hard to diagnose if you're not allowing pings
14:37<LemyDanger>directly to the uml ?
14:38<LemyDanger>but within the virtual lan the ping is 1ms
14:39<LemyDanger> <- the host and with NAT the packages goes to the um
14:40<LemyDanger>should i NAT the ping packages to the UML?
14:40<jdike>whatever you do with the HL packets
14:41<LemyDanger>my holy firewall script :)
14:43|-|ram [] has quit [Remote host closed the connection]
14:48<LemyDanger>i did it
14:49<LemyDanger>lines like $IPTABLES -A FORWARD -i $EXTDEV -o $INTDEVGAMEHOST -p ICMP --icmp-type 0 -j ACCEPT and so on
14:49<LemyDanger>the ping is the same
14:50<LemyDanger>and the problem is still there
14:51<LemyDanger>i just was ingame to see what hl dedicated server says
14:51<LemyDanger>no change
14:54<jdike>12 to the host, 24 to the UML?
14:55<LemyDanger>no 12 ,12
14:55<LemyDanger>the network is ok
14:58<jdike>so the UML network stack is OK
14:58<jdike>the problem is in userspace somewhere
14:59<jdike>what version of UML?
14:59<LemyDanger>you can mail my ssh
15:01<LemyDanger>kernel host 2.6.20, kernel uml i tried yesterday night the and before i had the 2.18
15:01<LemyDanger>skas is running
15:01<jdike>those should be pretty good
15:02<jdike>even without skas
15:02<LemyDanger>i thought too :=)
15:02<jdike>with hl on the host, what pings do you see?
15:02<LemyDanger>i could mail you the ssh account
15:02<kokoko1>root me :D
15:02<LemyDanger>i didnt try it
15:03<LemyDanger>because hl is very insecure
15:03<LemyDanger>i use the email address on your homepage ?
15:12<LemyDanger>ok the mail is sent
15:12<LemyDanger>good luck, i stay in chat
15:18<LemyDanger>did you get the mail ?
15:19<LemyDanger>i send it to jdike at karaya dot com
15:21<LemyDanger>the ip is not in the mail, its up there in chat, did you see?
15:23<jdike>what pings do you see with HL on the host?
15:24<LemyDanger>shoud i really install it there ?
15:24<LemyDanger>that would take some time
15:24<LemyDanger>30 minutes
15:24|-|besonen_mobile [] has quit [Read error: Connection reset by peer]
15:25|-|besonen_mobile [] has joined #uml
15:25<jdike>how do we know it's slow until we know how it does on the host?
15:25<LemyDanger>ok i will do it
15:25<LemyDanger>30 minutes left :)
15:49<LemyDanger>i m back and on the host the latency is just like the ping
15:54<jdike>what does vmstat inside UML look like?
15:57<LemyDanger>i saw that on the host the server he says to me 999fps and on the uml 0fps
15:58<LemyDanger>i m not sure about it
15:59<LemyDanger>when i start the uml cs server the uml crashes now, first i must rewrite my firewall skript, to see the message
15:59<LemyDanger>5 min
16:07<LemyDanger>no I was wrong
16:08<LemyDanger>on the host and on the uml the server says 999fps
16:10<LemyDanger>my nerves :)
16:10<LemyDanger>i need a coffee
16:11|-|hfb [] has joined #uml
16:12|-|hfb [] has left #uml []
16:26<LemyDanger>back again
16:26<LemyDanger>vmstat moment
16:26|-|ram [] has joined #uml
16:31<LemyDanger>procs -----------memory---------- ---swap-- -----io---- --system-- ----cpu----
16:31<LemyDanger> r b swpd free buff cache si so bi bo in cs us sy id wa
16:31<LemyDanger> 0 0 0 376756 1276 72192 0 0 59 2 120 46 1 1 98 0
16:31<LemyDanger>inside uml
16:34<jdike>vmstat 1
16:34<jdike>while HL is running
16:35<LemyDanger>procs -----------memory---------- ---swap-- -----io---- --system-- ----cpu----
16:35<LemyDanger>i have it runnung
16:36<LemyDanger> r b swpd free buff cache si so bi bo in cs us sy id wa
16:36<LemyDanger>2 0 0 375336 1276 72192 0 0 45 2 127 86 1 1 98 0
16:36<LemyDanger> 0 0 0 375336 1276 72192 0 0 0 0 98 190 1 0 99 0
16:36<LemyDanger> 1 0 0 375336 1276 72192 0 0 0 0 107 199 0 0 100 0
16:39|-|linbot [] has quit [Ping timeout: 480 seconds]
16:40|-|linbot [] has joined #uml
16:40<LemyDanger>procs -----------memory---------- ---swap-- -----io---- --system-- ----cpu----
16:40<LemyDanger> r b swpd free buff cache si so bi bo in cs us sy id wa
16:40<LemyDanger> 2 0 0 375668 1276 72192 0 0 39 1 127 103 1 1 98 0
16:40<LemyDanger> 0 0 0 375668 1276 72192 0 0 0 0 149 193 1 0 99 0
16:40<LemyDanger> 0 0 0 375668 1276 72192 0 0 0 0 151 190 1 0 99 0
16:40<LemyDanger> 1 0 0 375668 1276 72192 0 0 0 0 147 188 0 0 100 0
16:40<LemyDanger> 2 0 0 375684 1276 72192 0 0 0 0 154 198 1 1 98 0
16:41<LemyDanger>tunctl -u $uid
16:41<LemyDanger>ifconfig $HOST_IFACE $IP netmask up
16:41<LemyDanger>bash -c 'echo 1 > /proc/sys/net/ipv4/ip_forward'
16:42<LemyDanger>route add -host $UML_IP dev $HOST_IFACE
16:42<LemyDanger>bash -c 'echo 1 > /proc/sys/net/ipv4/conf/$HOST_IFACE/proxy_arp'
16:42<LemyDanger>arp -Ds $UML_IP eth0 pub
16:42<LemyDanger>./linux ncpus=8 ubd0=root_fs ubd1=swap_fs mem=$MEMORY eth0=tuntap,$HOST_IFACE
16:42<LemyDanger>this is the start skript of the uml
16:42<LemyDanger>i have root_fs and swap_fs in the folder with linux
16:47<LemyDanger>do you think its a performance problem of the uml ?
17:07|-|aroscha [] has joined #uml
17:07<aroscha>what tool would you recommend to monitor cpu usage / swapping / paging / etc on a sub second level
17:08<aroscha>i would also need the raw numbers
17:08<aroscha>not just a nice graph
17:41<caker>13:33 < caker> jdike: so, lotsa performance stuff went into 2.6.22-um -- have any gut feel or rough benchmark for its improvement over 2.6.21, say during a kbuild?
17:42<caker>Should I take that as a no? :-p
17:53|-|ElectricElf [] has joined #uml
18:13|-|ElectricElf [] has quit [Quit: Lost terminal]
18:17<LemyDanger>i go to bed now, good night all
18:17<LemyDanger>computer stays online
18:27|-|cmantito [] has quit [Quit: I aer quitted.]
18:36|-|ElectricElf [] has joined #uml
19:20<LemyDanger>i stopped everything but the one uml and no improvement
19:40|-|kokoko1 [~Slacker@] has quit [Ping timeout: 480 seconds]
20:29|-|LemyDanger [] has quit [Ping timeout: 480 seconds]
21:30|-|silug [] has quit [Ping timeout: 480 seconds]
22:11|-|newb_prog [] has quit [Quit: Leaving]
22:26|-|cmantito [] has joined #uml
22:32|-|ram [] has quit [Ping timeout: 480 seconds]
22:58|-|jdike [] has quit [Quit: Leaving]
22:59|-|VS_ChanLog [] has left #uml [Rotating Logs]
22:59|-|VS_ChanLog [] has joined #uml
---Logclosed Thu May 31 00:00:21 2007