lol what a nick for a log bot
brilliant
if i want to reinstall my VPS with a different version of OpenBSD using the bsd.rd method using the "SSH to serial" method, would i need to prepare a bsd.rd that can output to serial console? or will a stock bsd.rd work?
heh you're obviously not realizing what points things to the console of com0
boot.conf aka /boot and boot blocks in general do this for you, all kernels have the capability to do serial console
toddf: cool, thanks. so basically i can just use the stock bsd.rd and it will do the right thing
youbetcha, if you see a boot> prompt on serial, 'boot bsd.rd' will do what you want
toddf: i totally get it now. since a reboot of the VPS would get it into serial anyway, i will be able to access the boot prompt via serial, so i can boot bsd.rd from there
funny thing is, i have done this hundreds of times on pyhsical boxes, but somehow this "SSH into the VPS serial console" concept threw me off for a moment :D
toddf: thank you
lteo: y/w
Anyone experiencing connectivity issues?
I'm up via v6 ;-)
and v4..
I'm having trouble on v4.
std networking troubleshooting. ping gw. traceroute. etc.
define 'connectivity issues' /
?
hello.
more people having problems with downloading files from outside the US to their vps?
mine takes 10-20 secs on downloading webpages with curl/lynx
[20:43][ronny@vps:~]$ time lynx -dump http://www.telegraaf.nl
real    0m19.852s
user    0m0.020s
sys     0m0.022s
hi all .. I'm also having 'connectivity issues' as in cannot get to it...
19 secs for downloading a website of a newspaper :-S
Pings are timing out, can't connect on ssh
i wonder if my vps is capped
outbound is much faster than inbound...
I am seeing 30% packet loss,
:(
Packet loss near the arp end.
i dont have that yet
toddf: huge packet loss at highwinds.com.any2ix.coresite.com
maybe the other way hmm let me try
toddf: ~30%
yeah lots packetlooss now also 30%
that must be the reason
i sent gary a msg
I get 0% packetloss on my vps at arp
toddf: from Portland OR:
--- wraithan.net ping statistics ---
160 packets transmitted, 108 received, 32% packet loss, time 159320ms
wild, from my vps to your vps, 0% packet loss
i tried big newssites in the netherlands like nu.nl telegraaf.etc
about 30% loss
--- www.l.google.com ping statistics ---
21 packets transmitted, 11 packets received, 47.6% packet loss
pinging www.google.com from my vps
toddf: well the packet loss is upstream.
yes
same here
http://pastie.org/1633728 ... % column is packet loss..
Here's mine: http://pastie.org/1633729
im getting packet loss as well
My monitoring service is also having problems from multiple locations.
jazz57: I'm in .pt and my traceroute also stops on ge0-arpnet.cust.lax07.mzima.net (67.199.135.102)
Thanks, rgouveia.  There's definitely something going on with the network.
wonder if it is the upstream mzima.net ? ;-)
Here comes help1
up_the_Irons: Anything unusual happening on the network?
We're definitely seeing some packet loss.
jazz57: it appears level3 is having some problems today
I can't even ssh to my vps.
it appears kvr12 is also having some network issue as of 10 mins ago
i'm in the console, but I don't get any network connectivity
Figures.  I'm on kvr12.
who installed routeros again! </sarcasm>
Hey there
Any known issues with kvr12 ATM?
<@up_the_irons> it appears kvr12 is also having some network issue as of 10 mins ago
Any ETA?
First problems I saw were at 11:31
jazz57: What's your local time now?
12:18
jazz57: thanks :)
Sorry, I'm so SoCal centric!
hobarrera: I'm shutting down all the VMs now; will be rebooting, kernel upgrade, then bringing box back up.  i would say 30 mins at the latest
all the vms?
up_the_irons: ok, thanks for the info
fink: the whole network stack appears dead; i've seen this once before.  will be rebooting.
up_the_irons:  will you rebooot it ?
coobra: yes
no no  no :p
up_the_irons will earn this soon  =)
http://evilrouters.net/wp-content/uploads/2011/01/reboot.png
LOL
hah
:D
lol
it's funny because up_the_irons looks just like him irl
Gotta go, thanks for the info
hobarrera: np
packet loss seems to have gone away for me.
kvr12 is rebooting now; kernel upgrade already done
ah, just about to ask about kvr12
yeah, I packet loss is gone, yay!
i was able to get the networking back by restarting the networking sub-system, but i'm pretty sure that requires all VMs to be down (b/c they hold onto the tap / bridge resources)
yeah, my typing is fast now...
yay 0% packet loss
looks good on my side too
thank you
np, but I didn't do anything.  was a level 3 issue
your presence resolved the issue :)
seems to work again :)
i seem to have that effect on problems
up_the_irons: still trying to connect here ...
rgouveia: mine just started up
gonna check if I'm on kvr12...
what happened?
mine is up too :-)
Pings look good, too.
my coworker upgraded the VPS to the March 1 openbsd/amd64 snap and got "nvram: invalid checksum". are only certain openbsd versions supported?
lteo: all are supported, you just need to note 'disable mpbios' in big red flashing letters across your upgrade procedural documentation
http://support.arpnetworks.com/kb/vps/openbsd-hangs-at-setting-tty-flags-when-i-use-a-custom-kernel
toddf: thank you :) sorry, we just got the vps yesterday and we're still trying to feel our way around
lteo: ... and if you fuck it up, you can fix it via vnc by disabling mpbios from the boot loader.
jdoe: thanks, i'll add to my notes in case we ever need to
thank him, I think the how-to-fix link is to his site ;)
man, if i added all my customers to my linkedin profile, i'd have  massive connections
up_the_irons: let's connect
fink: my username is "garrydolley"
i am too drunk to know wtf linkedin is, but i will go create an account there and add you up_the_irons
just so you can have massive connectinos
azmarco: i like your attitude ;)
:)
some dumbasses actually put where they work (i work there too)
up_the_irons:  i can't find you on there man
azmarco: search for "Garry Dolley"
boom
accepted!
holy shit
you know people at booz allen?
i work w some contractors that work for them
azmarco: booz allen?
booz allen hamilton
hmm, i don't recall
they are a decent sized contactor where i work
why does slicehost suck so bad?
fink: their STL site has had massive network problems today
fink: they're expensive for what they offer, and their disk is slow (but everyone's disk is slow, shared disk is always slow)
another packet loss spike
saw that
now I know why
huge traffic going to one vps on kvr12.  same vps that had a huge spike earlier (and probably why kvr12 losts its networking stack)
we're getting /.ed?
who?
Anyone have best practice experience with ZFS RAIDZ?
Define best practice, but I have production experience with it.
comparing 8x 4 Drive RAADZ vs. 4x 8 Drive RAIDZ2
in terms of risk, performance, rebuild time
Assumming it's the same 32 drives in either case, rebuild time would be shorter with 8x4.
yep same drives
across 5 controllers
Performance is going to be about the same, since you have to calculate the same amount of parity.
8 drives per controller except for the 5th that'll have 4
wouldn't IOP be double on 8x RAIDZ
4x8 Z2 will provide n+2 for any two disks. 8x4 Z1 is only partially n+2.
as a raid set can only do one operation at a time
That's 36 disks, not 32.
two os drives, two spare
I'll also have two 128GB MLC SSD for L2ARC
and two 32GB Intel X25-E mirrored for ZIL
I'm pretty sure ZFS can do parallel operations on a single vdev.
If it were me, I would configure it as 4x8-disk Z2, making sure to put two disks on each controller.
hmm, once I get the hardware I'll have to benchmark it
yep
I think the other factors are a a wash at that point and that configuration gets you minimum n+2.
right
With 32 spindles, n+2 is would be more important to me. :)
s/ is//
Which OS, btw?
Nexenta Core
I'll install Solaris 11 to see if its anywhere
any better*
We have a file server at work running that.
It's wonky.
so n+2 with RAIDZ2 is good, but its not doubling risk to goto RAIDZ in this case
due to faster rebuild time
and > number of raid sets
how so wonky?
Maybe, but IMO n+2 and t(rebuild) = 2m >> n+1 and t(rebuild) = m
well depends on what the actual difference in rebuild time is
If you just do things how Nexenta wants you to do them, it works ok, but we tend to demand more than that from the servers.
We have to drop to their "shell" to do most of the volume and iSCSI, NFS and CIFS server configuration bits
yea I'm not going to use their appliance OS
but core
Ah.
which doesn't have their CLI
or gui
nor the TB limit on the free version
Ours isn't the free version. :)
indeed
Yea the plan it to play with Gluster on top of this
and see how well it would work to create a unified name space of multiple units
The NexentaStor box has been problematic.  We're close to saying to hell with it and reloading the thing straight-up Solaris.
Has Oracle change the licensing on Solaris
changed*
Just some pesky bits about the support contract being voided.
at some point it was you could only trial it for 90 days on non-Sun hardware without a Solaris support contract
the support.arpnetworks.com site is basically the same as e-mail support@, correct?
emailing*
last time i emailed, garry opened a ticket for all further correspondance
hm.. cuz last time (been a little while), I couldn't get an e-mail to him due to some configuration error on my e-mail provider's part
meh, I'll just open a ticket :)
mattx86: it's the same, yes
up_the_irons: ah, thanks :)
np