Back to Home / #linode / 2006 / 10 / Prev Day | Next Day
#linode IRC Logs for 2006-10-11

---Logopened Wed Oct 11 00:00:58 2006
00:03<warewolf>abso9lutely awesome.
00:03<warewolf>thriller in legos.
00:04<warewolf>LEGO FIGURE
00:04<warewolf>CO STARRING
00:04<warewolf>MORE LEGO FIGURES
00:04[~]warewolf dies laughing
00:05<warewolf>I think I have a copy of "The Making of Thriller" on VHS somewhere at my parents house.
00:05<warewolf>they even include the coincidential resemblance clause at the end of the video
01:10|-|spr [] has quit [Quit: This computer has gone to sleep]
02:04|-|SpaceHobo [] has quit [Ping timeout: 480 seconds]
03:01|-|internat [] has joined #linode
03:02|-|SpaceHobo [~spacehobo@] has joined #linode
03:37|-|internat [] has quit [Quit: Leaving]
03:37|-|internat [] has joined #linode
03:53<encode>how can i figure out why my host load is going up and down like a windows server?
03:54<encode>ive been watching top, and nothing is using that much cpu
03:54<encode>nor io tokens
03:54<encode>and my host load goes from 0.1 to 1 or even as high as 3 or 4
03:54<encode>then slowly goes back down again
03:55<encode>im using ~96MB of swap
03:55<encode>on a linode with 160MB ram
03:57<internat>`stupid question, is host load depending on ur individual instance? or the actual host?
03:58<thelsdj>maybe using iostat to check how much activity the swap is getting? 96mb of swap sounds like theres stuff getting swapped in and out
03:58<thelsdj>also watched the 'wa' column on the cpu in top, will show how much stuff is waiting for disk io
04:00<encode># iostat
04:00<encode>-su: iostat: command not found
04:02<encode>also, i dont have a wa column in top
04:03|-|marcel [] has left #linode []
04:05<encode> 19:05:24 up 239 days, 13:12, 10 users, load average: 0.37, 0.29, 0.83
04:05<encode>maybe a reboot would fix things?
04:05<encode>although thats such a bandaid solution
04:11<thelsdj>top doesnt have this line?:
04:11<thelsdj>Cpu(s): 8.5%us, 1.9%sy, 1.5%ni, 84.7%id, 2.8%wa, 0.3%hi, 0.3%si, 0.0%st
04:11<thelsdj>also iostat is usually in sysstat package
04:13<encode>Cpu0 : 1.9% user, 1.7% system, 0.0% nice, 96.4% idle
04:18<encode>iostat output:
04:18<encode>avg-cpu: %user %nice %sys %iowait %idle
04:18<encode> 1.64 0.00 0.49 0.00 97.88
04:18<encode>Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn
04:18<encode>dev98-0 3.03 61.41 31.68 1271071812 655801096
04:22<encode>i've been running iostat at 60 sec intervals
04:23<encode>%iowait never goes above 0
04:23|-|graham [~graham@] has joined #linode
04:23<encode>caker: is host49 getting hammered? am i swap thrashing?
04:24<graham>Is there anyone else here who is on host39 ? The one we migrated to from 53 recently ?
04:24<graham>I ask because the swap doesn't seem to be working.
04:24<encode>graham: have you activated it?
04:24<graham>Euh, how do I do that ?
04:24<encode>umm, dont remember offhand
04:25<encode>what does the following command say?
04:25<encode>cat /proc/swaps
04:25<graham>graham@rex ~ $ cat /proc/swaps Filename Type Size Used Priority
04:25<graham>graham@rex ~ $
04:25<graham>No swap :-(
04:26<encode>then no, you haven't activated it
04:26<encode>i suggest you google for how to do that
04:26<encode>since i dont know offhand, and im about to go
04:26<graham>Ok thanks a lot.
04:38<thelsdj>well obviously your linode website config needs to include a swap, then you need to know the device name like /dev/sda2 or whatever, you can try 'swapon /dev/whatever'
04:38<SpaceHobo>sudo fdisk -l
04:38<SpaceHobo>you may already have one
04:39<graham>I think the migration might not of created my swap properly, so I'm deleting and re-adding via the linode web interface
04:41<graham>That worked. Excellent
05:10|-|jekil [] has joined #linode
05:18<erikh>ok guys, email hilarity time!
05:18<erikh>And what about that? We live in XXI and todays no one cares about
05:18<erikh>the EULA
05:18<erikh>Last I checked it was MMVI. Although, if you live in XXI and have e-
05:18<erikh>mail, I bet Paul the Apostle would love to hear about it.
05:18<erikh>then, to top it all off, the reply:
05:18<erikh>Find difference between a year and a century. No one tells year in Arabic
05:18<erikh>numbers, mister "I'm a big troll who lives in USA and like being an ass".
05:18<erikh>And that "live in XXI" was a rhetorical one.
05:18<erikh>Arabic numbers!
05:29<erikh>ok, well, I guess I have to do my migration
05:29<erikh>is moving from 53 to 39 going to change my IP address?
05:35<graham>erik: No, your IP address moves with you.
06:04|-|internat [] has quit [Quit: This computer has gone to sleep]
06:05<erikh>oh, sweet.
06:05<erikh>well, might as well move it now.
06:07|-|erikh [] has quit [Quit: Terminated with extreme prejudice - dircproxy 1.0.5]
06:09|-|erikh-cgiirc [] has joined #linode
06:30|-|Newsome [] has quit [Quit: Linux: Now with employee pricing!]
06:46|-|erikh [] has joined #linode
06:47|-|erikh-cgiirc [] has quit [Quit: CGI:IRC 0.5.7 (2005/06/19)]
06:47|-|erikh [] has left #linode []
06:47|-|erikh [] has joined #linode
07:24|-|cmantito [] has joined #linode
07:29|-|FireSlash [~FireSlash@] has joined #linode
07:30|-|graham [~graham@] has quit [Quit: Leaving]
07:37<encode>so after doing a big apt-get update a few weeks ago (at least, thats the only major change i can think of), my cpu usage is significantly higher than before
07:37<encode>any ideas?
07:37<encode>err apt-get upgrade
07:39<encode>massive change
07:44<linbot>encode: Linode100 - 7, Linode150 - 1, Linode200 - 5, Linode300 - 37, Linode400 - 37
07:56|-|linville [] has joined #linode
08:08|-|Dreamr_3 [] has joined #linode
08:10|-|FireSlash [~FireSlash@] has quit [Read error: Connection reset by peer]
08:14|-|Dreamr3 [] has quit [Ping timeout: 480 seconds]
08:28|-|spr [] has joined #linode
08:29|-|internat [] has joined #linode
08:30|-|internat [] has quit [Remote host closed the connection]
09:02|-|schultmc_ [] has joined #linode
09:15|-|verticalf [] has joined #linode
09:18|-|verticalf [] has quit []
09:19|-|verticalf [] has joined #linode
09:22|-|verticalf [] has quit []
09:27|-|konoko [~Vampire@] has joined #linode
10:02|-|cmantito [] has quit [Ping timeout: 480 seconds]
10:04|-|erikh [] has quit [Quit: Terminated with extreme prejudice - dircproxy 1.0.5]
10:05|-|erikh [] has joined #linode
10:09|-|cmantito [] has joined #linode
10:22|-|linville [] has quit [Quit: Leaving]
11:19|-|iamback changed nick to JavaWoman
11:37|-|fendrish [] has joined #linode
11:47|-|SpaceHobo [~spacehobo@] has quit [Ping timeout: 480 seconds]
12:00|-|SpaceHobo [~spacehobo@] has joined #linode
12:09|-|Newsome [] has joined #linode
12:36<cmantito><tiny minor flood>
12:36<cmantito>Current Billing Cycle: October 01, 2006 - October 31, 2006
12:36<cmantito>Next Payment Due Date: November 01, 2006
12:36<cmantito>Current Balance: $0.00
12:36<cmantito>that menas I have to pay again in November
12:47|-|Newsome [] has quit [Quit: Linux: Now with employee pricing!]
12:52|-|SpaceHobo [~spacehobo@] has quit [Ping timeout: 480 seconds]
12:53|-|graham [~graham@] has joined #linode
13:04<graham>I have a Apache / TCP question I'm really hoping someone can help.
13:04<graham>Apache has a bunch of connection which it never closes. netstat says they are CLOSE_WAIT.
13:04<graham>In Apache's mod_status the 'Seconds since beginning of most recent request' is over 10 minutes ago.
13:04<graham>This on host39
13:11<iggy>could be keepalive's or the connections are still being (re)used
13:11<graham>My Apache KeepAliveTimeout is 4 seconds, and Timeout is 10 seconds.
13:12<iggy>any particular reason? that seems kind of low
13:12<graham>To try and fix this problem :-)
13:12<graham>The connection hang around much much longer then either of those two values. And eventually all the Apache connections are used up
13:13<iggy>what exactly are you trying to fix? I'm not entirely sure things being in CLOSE_WAIT is an issue unless it causes other problems.
13:13<iggy>okay, there's a problem to fix
13:15<iggy>did you try just turning off keepalive's?
13:15<graham>Yes, and I got the same result.
13:15<graham>At the moment the longest connection has been there for 10.5 minutes
13:18<iggy>you're sure the fact that apache's connections are used up is directly caused by this?
13:19<graham>In mod_status I see more and more connections over time which have been around for ever. So yes, I think so.
13:19<iggy>it's normal for connections to be in CLOSE_WAIT for a few mins after the connection is closed
13:21<graham>Humm. It's confusing. One of them is TIME_WAIT, and some of the apache connection don't show up in netstat at all.
13:25<iggy>what all are you using apache for? php? java? cgi?
13:26<iggy>also, is this new behavior or has it always done this?
13:26<graham>The site it has problems on is mod_python with Django.
13:26<graham>No, this is new behaviour. I _think_ it is since I migrated to host39 yesterday.
13:28<iggy>any mod_python changes recently?
13:29|-|linville [] has joined #linode
13:29<graham>Nope. There were changes to the Python code and HTML yesterday, but nothing that should hold a connection open indefinitely.
13:29<graham>Also when I browse the pages that the hanging connection is on it works ok for me.
13:31<graham>I tried changing my sysctl net.ipv4.tcp_keepalive_time setting to two minutes and bounced apache. Not sure if that will help.
13:32|-|MaconjaX [] has joined #linode
13:33<iggy>nothing interesting in apache's logs I suppose
13:34|-|SpaceHobo [] has joined #linode
13:35<graham>Except of course 'server reached MaxClients' eventually !
13:37<graham>I've also set the sysctl settings for net.ipv4.tcp_keepalive_probes and net.ipv4.tcp_keepalive_intvl - one can only hope.
13:42<graham>iggy: I've got to go home. Thanks for the help.
13:42|-|MaconjaX [] has quit [Quit: CGI:IRC (Ping timeout)]
13:43|-|graham [~graham@] has quit [Quit: Leaving]
13:51<iggy>if anyone sees that guy (graham) come back later, tell him to turn off kaspersky if he's using it
13:52|-|SpaceHobo [] has quit [Read error: No route to host]
13:55<kvandivo>graham the cracker?
13:56[~]iggy slaps kvandivo
14:01<bendy24>i thought he didnt say such things in this channel?
14:01[~]mikegrb slaps bendy24
14:04[~]bendy24 takes it like a man
14:31|-|schultmc_ [] has quit [Remote host closed the connection]
14:50|-|konoko [~Vampire@] has quit [Quit: taisetsu to yuukoto wa subarashii tsukashi subarashii to yuukoto wa motto taisetsu desu]
14:56|-|konoko [~Vampire@] has joined #linode
14:58|-|SpaceHobo [] has joined #linode
15:15|-|konoko [~Vampire@] has quit [Read error: Connection reset by peer]
15:15|-|konoko [~Vampire@] has joined #linode
15:49|-|encode [] has quit [Quit: leaving]
15:59|-|Kurt [] has joined #linode
16:00|-|linville [] has quit [Quit: Leaving]
16:00|-|konoko [~Vampire@] has quit [Quit: taisetsu to yuukoto wa subarashii tsukashi subarashii to yuukoto wa motto taisetsu desu]
16:05|-|konoko [~Vampire@] has joined #linode
16:38|-|fendrish [] has quit [Quit: CGI:IRC (Session timeout)]
16:40|-|bkirkbri [] has joined #linode
16:55<bkirkbri>hi, caker or mikegrb around?
16:56[~]Kurt goes back to re-reading "Introduction to Objectivist Epistemology" by the eminent 20th-century Russian-American philosopher Ayn Rand
17:04|-|karuna [] has joined #linode
17:06<Kurt>I defy you to refute a single word she ever said or wrote
17:07<lucca>hardly on topic here
17:07<lucca>but I will point out that any refutation I or others might offer would be rooted in a Different epistemology
17:07<Kurt>is your last name "Brasi"?
17:07<Kurt>because if it is, then you sleep with the fishes
17:07<Kurt>in other words, it would be rooted in an incorrect epistemology
17:08<lucca>Rand's views are well-formed within her own epistemology
17:08<lucca>Interestingly this is true of most epistemologies, and should not be considered strong praise.
17:09<lucca>but no, we can't really gain anything by arguing about it
17:10<lucca>I have met at least one other Objectivist into computer science, if you're interested in communicating with such an individual
17:11|-|karuna [] has quit [Quit: CGI:IRC (Ping timeout)]
17:13<Kurt>why not?
17:13<Kurt>epistemologies aren't just a matter of opinion
17:13<Kurt>Rand proved that her epistemology was the only objectively correct one
17:16<lucca>s/objectively/Objectively/; there is a difference and it is crucial here.
17:18<lucca>but as I said, Objectivism isn't really on-topic here, and I'm not interested in somehow converting you from that path, so a useful argument is unlikely to happen.
17:25|-|jekil [] has quit [Ping timeout: 480 seconds]
17:33|-|encode [] has joined #linode
17:33<encode>caker: ping
17:35|-|konoko [~Vampire@] has quit [Quit: taisetsu to yuukoto wa subarashii tsukashi subarashii to yuukoto wa motto taisetsu desu]
17:36<bkirkbri>encode: did you say you're having trouble on host49?
17:37<encode>bkirkbri: yeah
17:37<encode>i decided to reboot my box about 20 mins ago
17:37<encode>its doing a forced fsck now :S
17:37<encode>415 days w/out a check
17:38<bkirkbri>encode: i'm on host49 and have found my host bogged down in iowait all the time
17:38<bkirkbri>small rsyncs taking 30 minutes and causing system load 10+
17:38<encode>but i wasnt having any iowait
17:38<bkirkbri>i haven't rebooted, but i'm not swapping and not using that many io_tokens
17:38<bkirkbri>oh, really?
17:39<encode>i used iostat and stuff, i never had anything waiting for io
17:39<bkirkbri>i found hdparm -t /dev/ubda gave me 500k/s
17:39<bkirkbri>i ran that with nothing running on the VPS, so I would guess it's indicative of lots of disk activity on the physical host
17:40<bkirkbri>what was going on with your node then (if not iowait, i mean)
17:40<encode>i dont know what the problem was
17:41<encode>but host load average was constantly going up and then slowly down
17:41<encode>so it would trickle down to say 0.3, but then spike up to 2 or 3
17:41<encode>it made for very slow webpage loads
17:41<bkirkbri>hmm, that's not so bad. I run a 20MB rsync and it's hitting 10+
17:42<bkirkbri>hopefully your reboot helps, maybe i'll try that although I doubt it will help
17:42<encode>i had an uptime of 240 days :S
17:43|-|Newsome [] has joined #linode
17:43<encode>semi related question - im running the latest 2.6.18 kernel available - do i need to disable tls?
17:45<encode>ok this really sux, just booted, hostload staying still at 1.70
17:45<bkirkbri>IIRC the 2.6.18 kernel works with tls
17:54<encode>do i need to run rngd?
17:54<encode>surely the host's randomness is enough
17:54<encode>i dont even know when it got installed
18:01<encode>so now i have values showing up in %wa in top - ranging from 3.3 to 6 or so
18:02<encode>maybe its the host thats getting thrashed
18:03<bkirkbri>i find that even ls can take a lot of time to run, due to iowait. But since my node isn't doing anything I chalk it up to others thrashing the host
18:07<encode>im thinking of moving to a linode 300 anyway
18:07<encode>mmm 300 MB ram
18:07<bkirkbri>the problem is that anything that accesses the disk (even a little) can get held up on a thrashing host
18:09<encode>which brings the hostload up
18:09<encode>for the guest
18:10|-|encode_ [] has joined #linode
18:10|-|encode [] has quit [Quit: leaving]
18:10|-|encode_ changed nick to encode
18:26|-|FireSlash [~FireSlash@] has joined #linode
18:35|-|Kurt [] has quit [Ping timeout: 480 seconds]
18:52<encode>the host load graph in the web interface - i presume thats the host as in host49, rather than my linode?
18:52<encode>in which case, maybe my issues are because the host is so busy
18:59<bkirkbri>i wondered about that, thought it was for the node... but i don't think mine has used nearly as much as the graph says...
18:59<encode> <-- :O :O :O
19:13|-|kvandivo [] has quit [Ping timeout: 480 seconds]
19:16|-|Kurt [] has joined #linode
19:19|-|jekil [] has joined #linode
19:29|-|bkirkbri [] has left #linode []
19:56|-|SpaceHob1 [] has joined #linode
20:02|-|SpaceHobo [] has quit [Ping timeout: 480 seconds]
20:11|-|Dreamr3 [] has joined #linode
20:15|-|SpaceHobo [] has joined #linode
20:16|-|Dreamr_3 [] has quit [Ping timeout: 480 seconds]
20:21|-|SpaceHob1 [] has quit [Ping timeout: 480 seconds]
20:22|-|cmantito [] has quit [Quit: Outta here, yo.]
20:24|-|kvandivo [] has joined #linode
20:30|-|konoko [~Vampire@] has joined #linode
20:36|-|jekil [] has quit [Ping timeout: 480 seconds]
20:37|-|al [] has joined #linode
20:47|-|al [] has quit [Quit: CGI:IRC (Ping timeout)]
21:05|-|Kurt [] has quit [Ping timeout: 480 seconds]
21:06|-|kokoko [~Vampire@] has joined #linode
21:10|-|konoko [~Vampire@] has quit [Read error: Operation timed out]
22:11|-|FireSlash [~FireSlash@] has quit [Read error: Connection reset by peer]
22:47|-|thelsdj [] has quit [Quit: leaving]
22:59|-|VS_ChanLog [] has left #linode [Rotating Logs]
22:59|-|VS_ChanLog [] has joined #linode
23:00|-|kokoko [~Vampire@] has quit [Quit: taisetsu to yuukoto wa subarashii tsukashi subarashii to yuukoto wa motto taisetsu desu]
23:23|-|Newsome [] has quit [Quit: Linux: Now with employee pricing!]
23:42|-|schultmc [] has quit [Read error: Connection reset by peer]
23:46|-|schultmc [] has joined #linode
---Logclosed Thu Oct 12 00:00:47 2006