lvm on centos is a bit nicer now that they (I think) stopped naming everythin VolGroupXX and LogVolXX by default.
pain in the ass when you're trying to put one machine's disks in another. (ie "hardware died, rescue time")
it's nice to nam disks by their hostname
making Debian 6 templates.. w00t
man, Debian boots quick
holy.shit.
was able to resize Debian auto install template on-the-fly during provisioning
just need to script that up...
and it'll be quite a bit less burdensome to manage the debian 6 templates
weird, i've found ubuntu to boot much faster than debian in general
really? i've noticed the opposite
that's server version of both
is debian using btrfs now?
i've noticed btrfs boots slightly faster than ext4
I think you can choose btrfs in the Debian 6.0 installer, but it's not the default
mercutio: default is ext3 still
ahh ok
i wonder why
debian still uses /etc/inittab, it is very conservative :)
why it boots fast i mean
there's not much going on with it
i like /etc/inittab
like, it doesn't do much on boot
hmm
neither does ubuntu
The Debian base install really is very small
i've noticed arch is pretty fast to boot
oh, /netbsd/ i found fast to boot
and VERY small
It's a good size!
yeh i'm sick of bloat
i'm wondering why my netbsd takes up so much now
oh
i installed all the source
lol
so you're templating images now?
oh you were before too weren't you?
i think i did a reinstall straight away
mercutio: we always templated, yeah
maybe you should update openbsd at same time :)
i'm just finally finding time to refresh all our templates, and jpalmer is helping me with centos, so that'll be in the mix now too
ahh true
mercutio: ?  we already have 5.1
oh real
you didn't when i did my install :)
i think it was 4.7
yeah we didn't have 5.1 right away
i don't think 5.1 was out when i installed
ah
gotta peel myself away from this screen, bbl
W00t at Debian 6. Next time I won't have to upgrade.
i feel like i have been running squeeze for like 3 years now
Guest49601: Services are back, you've be renicked.
Services are back.
Man
I moved to Gentoo from Arch a few months ago
and now I want to move all of my VPSes over to Gentoo, also :|
My ARP VPS is FreeBSD for the time being
 4:33PM  up 396 days, 18:41, 4 users, load averages: 0.10, 0.06, 0.02
there was a port of Portage to FreeBSD at one point
interesting
why's that?
portage...?
gentoo portagE?
arenlor: :)
nice netsplit
freebsd portage
not a fan of gentoo generally
and portage is a violation of poor little freebsd's bits :~(
(I'm also a snob)
phlux: what do u like about gentoo?
kraigu: violation of bits? :)
up_the_irons: yes! :D
up_the_irons: you know what I mean
it's all about the sanctity of the bits, man
yus
wait, that's not going on my bill for next month, is it? :D
i think kraigu needs more whiskey
I ALWAYS need more whiskey.
lol
what about the freebsd/debian thing kraigu
debian gnu/kfreebsd i think it is?
I've seen it
same here
it's at least not as gross =]
haha
I ... really don't see the point of tying a freebsd kernel to a linux userspace though
to me, part of the reason you run FreeBSD is for the integration between said kernel and the userspace
if you want to run linux... run linux
that said, it takes all kinds
heh
yeah
i must admit i haven't played much with freebsd any time recently
and the freebsd ports collection was terrible when i used it
in what way?
it's like they didn't have stable release system for ports
up_the_irons: I like the package management - specifically etc-update's way of managing config files that may need updating.
but they did for kernel/userland
mercutio: that's the whole point,r eally
i duno, lots of packages were broken
didn't work
dependency hell etc
mercutio: you *can* stick to just "what was released" with binary packages, but
(I wouldn't)
this was before freebsd gave my file system corruption
if that's your objection then it hasn't gotten any better :D
me
nd i switched to openbsd
which had much more stale pots
ports
stable
haha
stale too probably i dunno
yes :D
I like "stale pots" though ;)
i don't use many things
I switched the other way round, from openbsd to freebsd
i have used the same window manage for 11 years
I used to have an openbsd desktop
phlux: ah cool
+r
before that i use enlightenment/twm
i alternated a little
i dunno why openbsd defaults to fvwm
twm is a much saner default
i prefer fvwm actually
I was fvwm for a few years, then a friend showed me windowmaker, and that's what I use when I have to use a non-Mac desktop nowadays
back ages ago i tried them all
window maker, ovlwm or something, fvwm, twm, enlightenment, something box, umm
kde, gnome,
you guys are going to hate me
they were all terrible
but it's KDE4 here.
kde was actually one of the best :)
but it was just too slow on older gaer
gear
it at least seemed coherent
yeah it is slow on older equipment
then i found ion :)
but my laptop is powerful enough to handle it fine - so why not do it? :)
hmm
my laptop is too slow with linux
it's fine on windows
actuall that's before i upgraded the ram
phlux: dude, I run Macs as desktops by choice, why would I hate you for KDE4? ;)
linux really desn't work well as a desktop on 2gb ram
well, aside from your bad taste ;)
lol
mercutio: It will with the right WM
i dunno why but windows handles 2gb of ram better than windos for dektop use
phlux: nah
it's mostly uhh
chrome
and virtualbox
and uhh
all tha other memory hungry stuff
yeah virtualbox will eat right through 2GB of RAM
i think it ws mostly chrome actually
and X
when I first got this laptop, it had 2GB, but KDE4 ran perfectly
i don't think i actually use virtualbox on it
I ran Chromium at that time
virtualbox on 4gb was still sucky though
I use Chrome now
oh i use chromium
in linux
i think
with an alias on starting to make it store it's cache directory in ram
i have 16gb of ram though
it didn't seem to make much diff
I wonder if it makes any difference that I compiled from source, and you (I'm assuming here) compiled from binary?
chrome is faster under linux than windows for me
firefox is slower under linux than windows...
i didn' compile from binary
i used the provideded packages
on the ubuntu atp repistory from google
or whatever
yeah that's binary packaging
yeah
Gentoo's package manager compiles everything from source
i've built chrome from source on windows :)
I wonder how much of a difference that makes performance-wise?
oh god, you poor monkey
yeh i used to use gentoo
i dunno i wouldn't mind so much if i had a ssd
at hte time
but i found the constant disk grinding in the background distracting
what i like about gentoo is being able to change options around
and being able to have multiple options to provide similar functionality
yeah, USE Flags are amazing
they take some getting used to, but I love them
but there were a few annoying things
like it'd keep swapping between mirrors
and one of the mirrors was always really slow
mm and things like gnome took like 6 hours to build
that is not typically the case anymore
which would hav been k6-2 350
i think?
<thinks back>
I built KDE4 pretty quickly (~3.5 hours?)
no that would have been athlon 1300
1 GB ram
one hard-disk
:/
3.5 hours is quick?
i'd asssume things build slower now with everything shifting to c++
and doing templates
and rtlo
used to take me ~8h to build a kernel
and all that jazz
used to take me 25 minutes to build a kernel
and like 7 minutes to rebuild a kernel?
:\
spoiled!
that was 2.1 days?
oh hangon
it took me like 45 minutes on my k6-2
i had stripped kernels
it actually varied a lot
i juse realised
my k6-2 was my shortest lived computer
like 2 years
1 year
i wonder what happened to it
cosi went to a p133
and athlon 1300
oh
i ditched the motherboard/cpu i think
cos my athlon 1300 had same case
mm i'm stuck in the past :/
desktop environemnts haven't really improved any time recently
mercutio: Windows 8 promises to do the opposite though.
unity is so ick it's not funny
arenlor: i'm using winows 8
up_the_irons: Was there just a heavy spike on kvr05 for disk?
arenlor: classic shell
http://classicshell.sourceforge.net/
mercutio: Sadly I support Windows, so I'll have to get it and use it.
it's not that bad
basially don't think "click on start bar"
arenlor: does not appear to be: load average: 3.74, 3.22, 2.76
think "windows key"
that looks like a load spike to me :)
iostat is way better for measuring than load average though
A few months ago my last ME client had their computer fry, I've never been so happy to have a client ignore my "always use a surge protector" advice.
hahahahaha
mercutio: a rise in load by 1 from 15 minute avg to 1 minute avg is not a spike :)
arenlor: lol
up_the_irons: hmm, it was down ~3MB/s and is slowly rising.
well a spike isn't necessarily for whole minute
it depends how much activity there is normally
that said
there's a little bit of bias in my head
where 3 is high load"
and 2 is medium
1 is light
0.50 is good
< 0.30 is hardly doing anything
mercutio: my first kernel was 0.99pl14c on my 386DX-25 ;)
load average mostly goes up by disk wait though
Now it's up to 532MB/s
it's much better at measuring disk load than cpu load
kraigu: i had a 486dlc40 running os/2
kernel sources fit on a floppy disk, but .. yeah, 8 hours wasn't unusual
but i'd played with linux ages ago
and the text output was so slow it wasn't ufnny
I miss OS/2 :(
486dlc40 is like 386dx-40
that 386 triple-booted linux,os/2,dos
i think
Heh, going through spam to make sure nothing is important, brain auto-ignores most of it, somehow sends "Cougar dating" to my concious self to analyze.
arenlor: you getting desperate? :)
os/2 with tshell was pretty fast
much fastr than linux
mercutio: Lol, must be.
mercutio: you are not considering the number of cores in the machine.  if you have 8 cores, then a load avg of 8 means 1:1 contention.  Over 8 means something is waiting, under that, all processes are runnable except ones blocked on I/O (which is generally what raise the load avg)
mercutio: you cannot say universally, "3" is high, "2" medium, etc...
it's all relative
up_the_irons: yeh i  now
i ws giving my bias
rather tha stating it as fact
a load of 3 on an 8 core machine is very different than a load of 3 on a 16 core, or 32 core machine
mercutio: roger
well yeh it depends how much you're running etc
that said if you saw load average of 3 inside vps it'd seem high
you also need multiple processes to generate load average
i suppose ic ould say that disk i/o average wait time of > 40 msec is high too
with iostat -x
Anyone else use dovecot?
but for that kind of thing to be accurate with virtualisation you need a lot of disk i/o
speaking of load average, i see a strange peak of like 13 load average
on a work host :/
it's already dissipated
doo doo odoo
it wsa never there :)
What was never there? ;)
heh
it's a mail server
i don't really care if a mail server gets a bit high in load average
it's a work pc not on arp
but like
it's all relative
like for a web proxy you want to have load average < 1 really
whereas for a mail server it should be < 20 peak
cos mail like if you get a mal storm temp
it's ok to backlog a bit
and it's better to accept than have the client back off
*/10 * * * * /etc/init.d/dovecot restart >/dev/null 2>&1
That should never be a solution.
hahaha
what's dovecot doing?
Disconnecting various folders.
that sounds bizzare
i've had very few issues with dovecot
Yeah, but that little hack solved it completely.
compared to qpopper
or cucipop
Could be that I'm using imaps
the annoying parts of dovecot are mostly to do with getting the imap folders mapped
and the namespaces
but that's syntax etc
not stability
never used imaps, but used iamp
maybe you should try imap?  see if it goes away?
you could always do a ssh tunnel to your imap server if you really want encryption
mercutio: Eh, my solution is simpler.
imap clients like to leave connections open
mercutio: That's the issue I think, if I don't use a folder for a while it disconnects it.
The client thinks it's open though.
For whatever reason, restarting it keeps it from disconnecting.
oh
that could be a nat timeout
do ssh connections die randomly?
Yep.
erk :/
tcp keepalive may help
i'm assuming that it's in your router thouh
It's a DD-WRT
oh
are you doing double nat?
Nope, modem bridges to it, it runs as the only NAT.
weird
do you use bittorrent or anything else with lots of connections?
I use bittorrent, but never with a lot of connections.
mm
bittorrent can keep state on lots
even with small number
Libreoffice isn't the most popular of downloads.
oh as a seed
Currently seeding yes.
 # cat /proc/sys/net/netfilter/nf_conntrack_count
17
heh
4 peers total.
i hardly have any
# cat /proc/sys/net/netfilter/nf_conntrack_max
23400
Holy shit, it's been so long that I forgot I asked iTunes to launch, and it's just now coming up.
i wonder what my adsl modem has
oh my adsl modem has connection tracking disabled
yeah i'd say it'd be better to try and fix any nat issues other restarting dovecot every 10 minutes
but is uppose if it's good enough for you, it's good enough for you
mercutio: I am lazy ^_^
i used to have an adsl modem that's drop ssh connections all the time
now i just route a /29 through the adl modem
adsl
and everything is sweet
If writing an entry into crontab is all I have to do, then that's all I do.
and nat on a linux vps for wifi
putty has a workaround for ssh connections dying btw
I have code for websites that have /* Ugly, but works, so fuck it */ written above it.
mercutio: What's the workaround?
tcp keepalives and seconds between keep alive under connection
(under putty config)
How long should be good?
150
but you can try lower/higher
it depends what the timeout is like
basicall if your nat timeout is 3 minutes then 170 should work
if it's 2 minutes you'd need 110
well i mean 119 or 179 should work
but you never know if it'll get delayed or bad time keeping or somthing
I like leaving a safety gap.
that used to bug me to hell and back
becuase i would be not using a shell
then i'd type on it and it'd be all like connection reset by peer
Yep
but it'd never screw up when it was working
but going and making a coffee or something and coming back would pretty much always screw it :/
Yep
also some bsds and linux have keep alive in recent versions
err always keepalive
Debian?
# cat /proc/sys/net/ipv4/tcp_keepalive_intvl
75
i thik it's that
i've never done anthin with it
depends on kerenl
http://tldp.org/HOWTO/TCP-Keepalive-HOWTO/usingkeepalive.html
3.2.0-0.bpo.2-amd64
well see if those variables exist
you'd have to drop keepalive_time
as it defaults to two hours before sending any probes
Yipes.
yeh so set it to something like 120
or 150
are the proc files in there?
Yep
cool :)
ok keepalive_time is the first probe
keepalive_intvl is how often to probe after that
keepalive_probes is how many non reponses before dropping the connection
Nice, thanks.
so after keep_alive time seconds it'll send keepalive_probes number of probes spaced keepalive_intvl apart
Yep
so you want keepalive_intvl*keeepaalive_probes long enough that you can have a minute or so of outage
without your connection dying :)
well i would cos i'm on dsl
and would want it to cover a resync
I'm on Cable, takes a few minutes to resync here.
yeh so make it high enough to buffer against that :)
the other thing that can do othr than keep nat connections up is it means if connections don't receive any traffic they'll still reset if the other end goes away
without it you can have ssh sessions stay around for ages in the background because they've not send anything to realise the other end isn't there
like ssh into host, reboot windows pc
Ah lol, yeah, at least 2 hours.
ssh into host see the old ssh
look again hours later and it's stll there
Remember that keepalive support, even if configured in the kernel, is not the default behavior in Linux. Programs must request keepalive control for their sockets using the setsockopt interface. There are relatively few programs implementing keepalive, but you can easily add keepalive support for most of them following the instructions explained later in this document.
The only proc I've used before is /proc/sys/vm/drop_caches
there's some always keep alive somewhre
Meh, only keepalive I care about is ssh really
Dovecot takes <1sec to restart, and there's no harm in doing so.
sysctl net.inet.tcp.always_keepaliv
in freebsd
ok well set the socket option on ssh
unless i can find out linux keepalive always
net.inet.tcp.always_keepalive  .. hmm it's the same in freebsd an openbsd
KeepAlive yes
ServerAliveInterval 60
in sshd_config
i dunno if that uses tcp/ip keepalive or not
some submitted a patch for linux in 1999
saying i was in the bsd's
I have a TCPKeepAlive yes in there already.
oh cool
Just didn't have putty configured for it
Which would explain why my Linux partition never disconnected.
ssh has keepalive defaulted.
the defaults are like 2 hours
for tcp keepalive
these nat issues are going to get worse
with the ipv4 starvation
mercutio: Yep. Even with Comcast I can't get real IPv6
heh
This promises to be hilarious: https://twitter.com/#!/search/%23NoteToMyEx
Loading Tweets seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.
it loaded the first page
It is funny, especially the teenage girls
teenage boys place a close second though
i juts saw stuff like i wish i never met you
Yeah, there's a mix of boring in there.
The one dude made a rhyme up about his ex not shutting up.
Heh, my monitor and speakers will be here Friday, my tower gets here Monday.
"Your a great Guy & I'm glad to see you happy, although you cheated & broke my heart a lot of times! I've learned & im moving on!"
how's he a great guy?
and why's she glad to see him happy?
she's stupid as hell
"perhaps next time you won't have sex with my bestfriend on my birthday"
maybe she's not your best friend then...
Told you it promises to be funny.
The "next time" in that sentence is promising.
Stupidity is great.
heh
so is t.co where all the spam is?
Eh?
bit.ly seems to have less spam
err x.co?
I've not heard of x.co
err x.co redirects to t.co
#NoteToMyEx you realize this functions ? http://x.co/n62u
stuff like that
t.co (and I'd suspect x.co) is Twitter's URL shortener/tracker
ahh
x.co appears to be godaddy
http://www.telegraph.co.uk/culture/art/9491391/Elderly-woman-destroys-19th-century-fresco-with-DIY-restoration.html
Is there anything more recommended than memtest86+?
jpalmer:
 │20:21:44 up_the_irons | really?  it has 768mb.  i was able to use the text installer (i simply prefer text anyway) to not do lvm
oh, u mean the other vm?
jpalmer: ok, gave the other one 768
hard reboot it
bbl
my first 1year uptime....
ariel@arp:~$ uptime
 01:02:25 up 365 days, 34 min,  1 user,  load average: 0.08, 0.02, 0.01
ariel@arp:~$
thans to ARP!!  hurra!
thanks*
wooo
comcast with ipv6 =p
:o
so yea... i got this new apartment that comes with internet built in
figured because they capped it after a certain point i would get comcast
whoa
the ENTIRE internet in your new apartment
thats awesome
but really i can't complain about 30Mb down / 30 Mb up
didn't think they provided symmetric, that is pretty hard to find these days
its rarer to find asymetric with upload much much higher
:P
lol
apparently thats what usc has
legit
i might have to peer off of this just for upload....
peer to a single host?
actually, i could setup a multi wan on my pfsense box and plug in the network speeds, and have it schedule packets properly
`ariel: cool!