up_the_irons: Webhostbudd_: yeah but the chassis is like $3500 plett: up_the_irons: Is the new blade chassis for dedicated servers, or will they become VPS host nodes? up_the_irons: plett: dedi plett: Cool. Those are selling well then? up_the_irons: sold out first chassis :) plett: Wow
Nice :) up_the_irons: otherwise i wouldn't be installin' a new one; it's a big hit to the bank initially
i got one pre-order for a new blade; then i'm pretty sure i sold another one yesterday (just waitin' to hear back from the guy...)
so yeah sellin' well :) plett: Indeed. Not the sort of thing you want sitting around idle without filling with paying customers up_the_irons: yeah definitely
the air isn't paying the bills
toddf: added backup space to your services list (you're the first b/c you mentioned being able to submit your key from the Portal. it is now possible :) ***: raptelan has joined #arpnetworks raptelan: anyone around/ mercutio: plett: the reason bridging often woroks better than ip routing with linux is because it doesn't store state per connection
well one reason
the other reason is that linux's fast path type support is very primitive
raptelan: i'm not from arp but i'm around :) raptelan: i'm having an issue getting my vps to boot with a custom linux install...
using the vda device
from the boot CD, `stat /dev/vda` shows device-type fc,0
but then when I go to boot my install, it's apparently fe,0, and panics.
because it can't find the root device, though it proceeds to print out all the vda partitions as fe## rather than fc##
mercutio: bit of a long shot, but do you have any ideas? mercutio: hmm
what's device type fe/fc :)
i'm running openbsd onarp
normally most linux installs auto find the disk
assuming you're not using anything like lvm
you canprobably try just root=/dev/vda1
if / is /dev/vda1
or like /dev/vda3 if / you do the boot, swap, / thing raptelan: I *am* trynig root=/dev/vda1
I'm not manually specifying the fc/fe crap
mercutio: this probably makes it clearer: http://fpaste.org/vj70/ and http://fpaste.org/SqYr/
mercutio: I got it working - I needed to put root=/dev/vda1 into the append lines in the lilo.conf, because the root=... lines get passed as a device major/minor pair, whereas append=... gets passed as a string. mercutio: weird
what dist does that?
i have used grub with linux recently raptelan: gentoo.
grub is more complicated, and wouldn't detect my disks anyways :/
also it requires IA32 emulation, whereas I preferred to build a pure 64-bit-only system.
now time for work :)
laters ***: raptelan has quit IRC (Quit: Page closed)
raptelan has joined #arpnetworks
heavysixer has joined #arpnetworks
ChanServ sets mode: +o heavysixer
dzup has joined #arpnetworks
heavysixer has quit IRC (Quit: heavysixer)
gcw|mbpro has quit IRC (Remote host closed the connection)
heavysixer has joined #arpnetworks
ChanServ sets mode: +o heavysixer
hive-mind has quit IRC (Ping timeout: 245 seconds)
gcw|mbpro has joined #arpnetworks
gcw|mbpro has quit IRC (Remote host closed the connection) brycec: shit... someone actually using Lilo these days? madness!
That's his problem right there :P (since with grub, everything is passed on the kernel line as a string to the kernel. staticsafe: yep
I don't really see a reason to stick with Lilo anymore :o brycec: it was always a pain in the ass
Maybe if I was interested in super-locking things down, I would combine lilo with a BIOS MBR block (and passwords) but that's all bullcrap anyways (better to just encrypt) staticsafe: yea ***: heavysixer has quit IRC (Quit: heavysixer)
heavysixer has joined #arpnetworks
ChanServ sets mode: +o heavysixer
lteo has joined #arpnetworks raptelan: brycec: Grub will not even recognize that there are any block devices aside from fd0 in the box.
and there's not even a floppy drive, obviosuly.
lilo is a far simpler tool for a simple job. ***: staticsafe has quit IRC (Quit: WeeChat 0.4.0)
staticsafe has joined #arpnetworks brycec: raptelan: that's a bummer. It's working fine for me (though it is technically Grub2)
(And it's presumably working fine for everyone else with a Debian image)
And Grub2 does things a little bit differently... Notably, search --no-floppy --fs-uuid --set 5ce09401-eaab-48f4-8fbe-5a56e26160be raptelan: brycec: I'm speaking of grub2 brycec: *shrug* You're happy with lilo, so I guess there's no point in debugging your grub woes :P raptelan: brycec: it would be nice if grub came out of the dark ages and actually could be compiled on a 64-bit x86 without a multilib install and run without 32-bit emulation support.
I only went to lilo because grub didn't work. :P
it was easier to fix. :P
I really don't care - I just want system to boot.
and I'm good now :)
and I'm happy that I don't have to have IA32 emulation enabled in my kernel anymore. brycec: fwiw raptelan /boot/grub/x86_64-efi/acpi.mod: ELF 64-bit LSB relocatable, x86-64, version 1 (SYSV), not stripped
(That's from my home Arch system, 64bit, EFI, Grub2) raptelan: hm
maybe the documentation I was reading is just out-of-date.
it does appear that I can install a from-source version of grub2 actually (I'm using Gentoo) toddf: please tell me you're not using a virtio disk vs an ide disk? raptelan: toddf: I'm using it via /dev/vda.
I don't have any IDE or SCSI support even compiled in toddf: never heard of 'vda' otherwise
'no ide' -> arpnetworks vps uses ide -> ding! raptelan: IDE = /dev/hda
SAS/SATA = /dev/sda
VirtIO = /dev/vda toddf: so then do you have a kvr# >= 27 ? raptelan: a what? toddf: your vps raptelan: what about it? toddf: look at your portal and you'll know your kvr# based on how you hit vnc
if it is on kvr26 or lower you don't have bios that presents the os with a disk if you have requested virtio
if it is on kvr27 and higher, bios presents virtio as a disk raptelan: kvr19
so what are you trying to suggest? toddf: so if you have requested your disks changed to virtio mode, you'll need to be using a bootloader that talks virtio directly not using the bios disks
otherwise, you need to go back to ide or request a transfer of your vps to a kvr27 or higher vps raptelan: I didn't request anytihng
I just signed up
umm why do I need to change anything? toddf: then why do you use virtio when virtio is not default? raptelan: it's all working fine. toddf: you're using /dev/vda ? raptelan: umm
the installed debian image the VPS came installed with used /dev/vda toddf: aaaaah raptelan: using virtio_blk driver
NOT ide toddf: ok so now I learn something new raptelan: so when I installed my own distro, I followed suite. toddf: kvr1-26 installs debian using virtio
kvr1-26 installs bsd using ide
I've only used bsd so I didn't know, apologies raptelan: hm
yeah I used bsd with arp in the past
was a bit surprised to see "QEMU virtual cpu" as I didn't think that showed up that way under bsd toddf: bsd bootloaders use bios to access disks, so if bios does not present virtio as a disk, no booting possible raptelan: but I'm not very good with bsd toddf: cpu strings are cpu strings, passed directly to dmesg raptelan: I don't understand that much about paravirtualization either
although I guess it's more efficient toddf: to some degree we are all a product of our experiences; some however find some things more logical or easy to deal with. I used linux from 1994-1999 and OpenBSD from 1998 - present. some overlap but it became more and more clear that the way I worked aligned well with OpenBSD raptelan: I liked my time with openbsd, but never really spent enough time with it to really know it as well.
I really like portage a lot, having such a degree of control over the system and nice interface to it. toddf: lets just say I tried to make a src tree on slackware back in the day to build everything, and got really frustrated because things were not organized nor repeatable to rebuild everything, 'make build' and all world is built on openbsd .. can't get simpler than that raptelan: From a purely purist standpoint, I'd probably opt for NetBSD, but I couldn't even get their install CD to boot on an arp vps
'emerge -e @world' to recompile the whole system is pretty simple.
what I don't like about openbsd is that the base O/S is a pretty large all-or-nothing thing, not broken down into separate packages I have individual control over.
(fairly minor since it's not a big bloaty crappile like most linuxes)
I also like being on a rolling version system (not for everyone or every environment, obviously)
at work our servers are all RHEL5 and I hate it :P toddf: rolling version systems are the bane of my existence. it is too hard to install one system and a month later install the exact same system, things have changed too much.
even tracking openbsd current, there was a time (I may return to this when I've got my office setup well again) where I'd literally rsync the install bits for all archs I track, install systems based on those install bits, and build packages all in a dated subdir so I know its all consistent with itself
that way I could go forward and back if I needed to to figure out when bugs occurred etc
otherwise updating the base system and continually updating packages creates an inconsistent environment where one package is built with one version of a lib and another with another version and sometimes this creates problems brycec: I'm guessing toddf avoids ports then ;)
(or packages a port and distributes that package, I suppose) raptelan: I don't have any need to install "exactly the same system as a month ago", ever. I like keeping things current.
but yeah, it depends on your needs.
there's definitely cases where I've preferred using something stable (e.g. debian)
brycec, toddf: It's also very easy to build binary packages from a master gentoo system and then deploy ONLY those tested binaries to other systems. I used to do that when I ran a hosting company and maintained 30-40 servers.
which is basically, the same as packaging and distributing a port package.
portage is inspired by ports, after all. toddf: case for installing exact same system: $client wants a system that does X, so I build and test for a month making sure it works asthey want, then I go and install it on their computer. how am I going to guarantee it works the same if there have been changes? raptelan: in fact today I rebuilt all 180-something packages on my server to both take advantage of compiler optimizations (newer gcc version, updated flags), and to build binaries of everything "just because"
like I said, cases exist.
for my personally-maintained systems, I prefer the other.
openbsd, netbsd, debian, gentoo, probably others I'm not familiar with, all make sense for various scenarios. Red Hat, on the other hand...*grumbles* ;)
(but again just being a bit silly - it clearly works for many people - I just don't jive with it at all) brycec: Red Hat makes sense for companies that want the safetynet of a "support contract" raptelan: you can buy support contracts for any of the others too, I would expect.
maybe not from the vendor brycec: Well if I were Fortune 500, I wouldn't trust anyone but the vendor :p raptelan: I'm going through a similar issue where I work...getting people weaned off of EnterpriseDB and planning to move to open-source PostgreSQL soon. brycec: I'm not saying I agree, simply pointint out the appeal of Redhat for so many. raptelan: hmm well I guess it depends how good the vendor's support actually is.
EDB's support is crap.
and because they've hacked up the open-source version and made it closed-source with undocumented changes, that makes it extra fun.
I've been waiting about 6 weeks now for a decent response to a "why does this particular query, run against this particular set of data, cause load on the server to exceed 40?" brycec: seems like it should be a simple task for them to track down, at least to get a basic idea
"because it keeps hitting the table store, from disk" for instance
*shrug* raptelan: well, I know why, because it greedily eats up all memory that the server has and linux's stupid memory overcommit is contributing to the problem as the box doesn't have swap, but still. It doesn't do that on stock PG.
reminds me I need to talk to our admins about changing the overcommit setting.
not fun when all sorts of random stuff just starts getting killed :P
I believe *bsd is free of such nonsense. ***: heavysixer has quit IRC (Quit: heavysixer)
heavysixer has joined #arpnetworks
ChanServ sets mode: +o heavysixer
heavysixer has quit IRC (Quit: heavysixer)
jamie_ has joined #arpnetworks jamie_: hey there! having a problem connecting to our VPS with you from our office. Checked with our ISP and it's getting stuck after it leaves them. Can you help? meingtsla: can you pastebin a traceroute? jamie_: https://gist.github.com/drlube/546b684c5c6f9bc57576 ***: heavysixer has joined #arpnetworks
ChanServ sets mode: +o heavysixer jamie_: actually… that shows that it's stopping at our ISP. Let me call them.
again :( brycec: jamie_: I can reach you (or an IP close to you) http://pastie.org/6423518
(I just picked the first public IP with a rdns in your traceroute jamie_: can you reach our static IP? 96.53.33.22 staticsafe: yep jamie_: tbanks staticsafe: actually hmm
jamie_: seems unreacable from arp ***: jamie___ has joined #arpnetworks
jamie_ has quit IRC (Ping timeout: 256 seconds)
hive-mind has joined #arpnetworks brycec: jamie___: No, but I can't reach it from Chunkhost either. ARP: http://pastie.org/private/0uirywcd452btgyfsma Chunkhost: http://pastie.org/private/955gtmhmqcncwqlkecssa
though I do hit the same edge Shaw router on both jamie___: cool.. thanks… on hold with our ISP. brycec: so if I had to guess, I'd say there's something effed with Shaw's routing toddf: seems there are other issues routing through shaw as well ***: jamie_ has joined #arpnetworks
jamie___ has quit IRC (Ping timeout: 250 seconds) jamie_: on with Shaw tech support
will let y'all know what happens… they seem to think it's on their end. toddf: 'preciate! brycec: always nice when the ISP believes it's their fault
jamie_: Is this a business connection? (i.e. Do you at least get to talk to "business support"?) jamie_: yeah… talking to them brycec: I love the disparity between residential and business support with my cable ISP. jamie_: thanks for sending the reverse trace :) brycec: happy to help jamie_: ok, they're going to get back to me. No ETA, but they're working on it. Frank at Shaw Biz Tech Support is the man… his voice sounded like he could totally be a radio announcer. brycec: ha nice ***: toorop has quit IRC (Ping timeout: 276 seconds)
toorop has joined #arpnetworks
lteo has quit IRC (Changing host)
lteo has joined #arpnetworks
heavysixer has quit IRC (Quit: heavysixer)
heavysixer has joined #arpnetworks
ChanServ sets mode: +o heavysixer
jamie_ has quit IRC (Quit: jamie_)
jamie_ has joined #arpnetworks
jamie_ has quit IRC (Ping timeout: 248 seconds)
jamie_ has joined #arpnetworks
heavysixer has quit IRC (Quit: heavysixer)
heavysixer has joined #arpnetworks
ChanServ sets mode: +o heavysixer
heavysixer has quit IRC (Quit: heavysixer)
brycec is now known as YoMomma
YoMomma is now known as brycec
heavysixer has joined #arpnetworks
ChanServ sets mode: +o heavysixer
heavysixer has quit IRC (Quit: heavysixer)
heavysixer has joined #arpnetworks
ChanServ sets mode: +o heavysixer
heavysixer has quit IRC (Quit: heavysixer)
perun__ has quit IRC (Ping timeout: 248 seconds)
heavysixer has joined #arpnetworks
ChanServ sets mode: +o heavysixer jamie_: still no word from Shaw… will post back when there's a resolution. But they say that it's probably related to other issues they've been having.
ciao for nao ***: jamie_ has quit IRC (Quit: jamie_)
heavysixer has quit IRC (Quit: heavysixer)
dzup has quit IRC (Ping timeout: 245 seconds)
heavysixer has joined #arpnetworks
ChanServ sets mode: +o heavysixer
heavysixer has quit IRC (Quit: heavysixer)
brycec has quit IRC (Remote host closed the connection)
brycec has joined #arpnetworks
mnathani has joined #arpnetworks