hmm.. kvr14 looks pretty normal to me
up_the_irons: hmmm, weird. having massive issues with my minecraft server.
Hien: i don't remember tbh.
Hien: was quite a while ago
whitefang: the load on the server is normal
how is CPU time allocated?
it seems a bit better now, but only have 3 of the 5 people that were on it from earlier. up until about 20 minutes ago with 5 people it was brutally painful all night.
it does seem much better now.
CPU time is scheduled by the standard Linux scheduler
standard. that's deadline, right?
or is it cfq nowadays
yah now its really bad once the 4th person connections
up_the_irons: kvr14 still normal cpu?
whitefang: yes
jbergstroem: i think it's deadline
Hien: making your vps now
up_the_irons: nice, just recieved billing email ;)
:)
i just ssh'd
:)
nice!
Hien: prepare for the best experience of your geek life
geek? this is Real Men Work
jbergstroem: what are you still doing here then? :)
nesta: jbergstroem LOL
jbergstroem / up_the_irons: cfq and deadline are disk I/O schedulers and have nothing to do with CPU scheduling.
brycec: doh
brycec: well it's a *something* scheduler for the CPU!
mercutio: how does your routing look?
lol
console: connect(): 3109@kvr14.arpnetworks.com: Connection refused
can anyone on kvr14 hit their serial consoles?
up_the_irons: well it worked before, haven't look at it since
disk performance seems to have gone up from when last talking
not nearly as much of a hit writing to disk rather than /dev/null
i just posted a question in #cisco, in case anyone else is in there and has some insight
but for anyone not already in #cisco... what's the question?
crap, it was long ;)
s/it was long/i'm too lazy to copy and paste/ :P
up_the_irons: related to the ntt issues?
mercutio: yes
brycec: on this crappy 3g i'm on, yes, i'm too lazy to copy & paste ;)
up_the_irons: mosh is your friend :)
hah yeah
still need to try that
you will be surprised, I work a lot on the train, and mosh makes a huge difference there
mosh is phenomenal... but its predictive stuff gets entertaining when you're scrolling or in a text editor (like when the line wraps)
brycec: but doesn't the schedulers tell the cpu how to prioritise different workloads of io?
up_the_irons: <3
incase it matters (since fbi is here logging it might show up in search)
QinQ (802.1AD) slvan(4) works great .. if you set the right mtu
1500 mtu on em0 .. 1496 mtu on svlan5 .. and stuff like nfs mounts of -r=16384,-r=16384 "just work(tm)"
between arpnetworks vps'es of course
mosh is just awesome when you have anything above 300ms latency
ehhhh. that's weird. one of my processes was "Killed" but not by me.
errr nevermind.
toddf: if I had the time, i'd pick your brain on that and write up a full blog post.  definitely interesting stuff and probably not a whole lot has already been written on it.
dr_jkl: <3 :)
mercutio: how's your routing doing?
up_the_irons: in theory the mtu of em0 should permit svlan to go 1500 by adding 4 more bytes, but I'm guessing there is somewhere outside kvm that is blocking it from being larger than 1500, but at least I figured out why nfs wasn't working
$ cat /etc/hostname.svlan5
vlandev em0 -autoconfprivacy mtu 1496
10.1.4.1/24
$ grep " /u " /etc/fstab
10.1.4.2:/u /u nfs rw,nodev,nosuid,intr,-3,-r=16384,-w=16384,-b 0 0
we are announcing 174.136.96.0/20 out of NTT again.  I'll be around if there are any issues.  I've tested one of our own /24 for over 24 hours and things seem good.
toddf: yeah there's MTU stuff all over.  if i had the patience, we could probably make it work.