pyvpx: :p mercutio: why's netbsd not working? pyvpx: it does not like the IPMI Virtual CDROM Drive mercutio: can it boot off floppy?
as archaic as that sounds :) pyvpx: didn't bother to try as I assumed it was the same driver/code path.
also, that means I'd have to make one :p mercutio: i doubt it is.
virtual floppy i meant pyvpx: or install ancient netbsd and slowly upgrade? haha mercutio: i've used virtual floppy on hp systems before. pyvpx: I thought I'd be smart and install OpenBSD on one drive, then manually shoehorn netbsd on the other...but only one drive is bootable/recognizable in the system bios mercutio: damn pyvpx: then I tried a FreeBSD liveCD, but could not get networking to work at all?? didnt look into it futher.
brought up interface, added default route...no nothin'...even to the gateway address. ARP entry...but no icmp replies. I'm guessing that's a sysctl thing but...gah
I'm trying to move and I've already wasted a month (I'm happy to give up_the_irons "free" money though :P) mercutio: maybe a problem reading the mac address?
and trying to use 00:00:00:00 or something
but i'm stabbing in the dark pyvpx: nah, it got the right one. just no packets passing. tcpdump just shows STP frames and nothing else. mercutio: did you try in promisc mode? pyvpx: so I'm further guessing it somes weird liveCD thing I dont know or care to investigate futher. by now, if I had any kernel hacking skillz, I'd have patched netbsd and had it installed :p mercutio: heh
what's netbsd actually doing?
does it show the cd? pyvpx: I could install a bootloader on the first drive that points to an install on the second drive and try my luck that way mercutio: or you can't even load the kernel? pyvpx: kernel loads, but then it can't mount the root filesystem as, according to the kernel, sd0 is offline and unable to be opened
"sd at scsibus0 target 0 lun 0: <IPMI, Virtual Disk, 3000> disk removable mercutio: hmm pyvpx: cd0 at atapibus0 drive 0: <IPMI, Virtual CDROM, YSOJ> cdrom removable
sd: drive offline
sd: unable to open device, error = 19
etc etc
boot device: cd0
then it errors out and asks where to find the root filesyste :p
I had this problem with another cheap provider
when I needed a box in london for some things
wasted 99EUR and a month
was convinced it was their shotty iDRAC
nope, turns out NetBSD just has a bug that no one hits or cares to fix?
ich wiess es nicht mercutio: that sounds like a damn annoying bug pyvpx: granted had I posted to the mailing list months ago, perhaps I'd have gotten some response or traction. rather than where I am now, politely bothering people in #netbsd and scratching my head
trying weird OS gymnastics -: pyvpx really just wants a shiny new netbsd/xen install, or three mercutio: netbsd dom0? pyvpx: ja mercutio: i have a kind of crazy idea to get an install going
setup an install, save it as a file, use a linux live cd and dd it onto the drive.
err save the raw dd data
but it'd be nicer if could get it to work normally, as what happens if it goes wrong etc.
although it's probably easier to try to get it going if you can boot normally :)
hmm
you could also try pxe if you ahve another computer on the network pyvpx: nope, no other ones. and yeah, that was my idea with the freebsd liveCD. only I just wanted to network access instead of asking for yet another iso (or .img) added to the ISO library.
I think I'll just ask as best I can on the netbsd mailing list if someone will hold my hand through debugging and patching this.
it *can't* be too difficult. it works flawlessly in open and free....
famous last words, of course. m0unds: oof, that sounds like a headache ***: fink has joined #arpnetworks
mkb has quit IRC (Remote host closed the connection)
mkb has joined #arpnetworks mercutio: oh this is vps brycec: Why would a VPS have IPMI? mercutio: ipmi made me think dedicated. brycec: ^ mercutio: cos otherwise there's not iso library is there? brycec: actualyl mercutio it can be, of course
(VPS CD comes up as: [ 271.622784] scsi 0:0:1:0: CD-ROM QEMU QEMU DVD-ROM 1.0 PQ: 0 ANSI: 5)
*actually mercutio: true, i didn't know if netbsd was doing something weird
if it's dedicated it's probably yet another intel revision of ethernet breaking freebsd
which would make sense brycec: Don't dedicate boxes have a VPN-only address (and address space)? Or is that just the IPMI controller?
And as I recall, they are dual-NIC, so one would have to be sure they're configuring the correct interface. mercutio: both interfaces are on the same network
so either should work brycec: True. i wonder, if a switch is configured for [static] port bonding, would the other end also need to be configured as such? Or would simply operating a single interface work? Probably depends on the type of bonding... mercutio: you should be able to alternate what ports you stick packets out on
but they may get reordered, as it's separate switches
and it's gigabit anyway so not really necessary.
there's something on the wiki somewhere about ways to set it up so it can failover
actually that reminds me, i dunno which switch the ip/kvm is on
i've been thinking about playing with bonding at home
but you kind of need more than gigabit from two cards. BryceBot: That's what she said!! mercutio: so i thought it'd be kind of cool if it could route one of them over infiniband
i cap under port speed anyway
if you've got gigabit ports, then if something local sends at full gigabit another host can get impacted slightly
so i just cap most ports at 900 megabit
then stick fq_codel on it ***: fink has quit IRC (Quit: fink) m0unds: staticsafe: are you using dhcpv6 on your mikrotik box? staticsafe: m0unds: nope m0unds: comcast enabled ipv6 for biz subs in my market, but dhcpv6 is lame and the implementation on my srx is buggy
figured i might try my rb450, but there's a chance it's buggy on that thing too, haha mercutio: is it dynamic ipv6? m0unds: they dynamically assign a prefix
you use IA_PD "hints" to request larger allocs
comcast will provide /64 by default, will allow resi customers to request a /60 and will allow biz to request /56
unfortunately, for whatever stupid reason, juniper's IA_PD support is hardcoded to request a /48 and there's a dev request being completed to allow that to be specified in the cli
so it requests a /48 and gets a /56 and ignores it staticsafe: m0unds: i don't use it but there are probably people in #ipv6 who use DHCPv6-PD on RouterOS, IIRC it works fine
i had someone IRC ask me for help for it earlier and they had comcast
it worked fine iirc ***: KDE_Perr1 is now known as KDE_Perry
xales1 has quit IRC (Quit: Derp) m0unds: ah, ok
staticsafe: have you seen the fun stuff you can do w/the ammo pack + aircraft in ps2? staticsafe: yeah m0unds: i had no idea
someone put a pack on the nose of my mosquito and i did a rapid vertical take off and did 3 flips and blew up. it was amazing. staticsafe: oh i recorded something today that was amusing
lemme see if i can cut it up
need some decent video editing software hm mercutio: can you just forward it to a linux box or osmething?
oh i was scrolled up a bit ***: mkb has quit IRC (Ping timeout: 244 seconds)
Seji has quit IRC (Ping timeout: 250 seconds) m0unds: haha, enabling dhcpv6 on my rb crashes it
updated to 6.23, and it crashes when i load/unload qos queues and also when i enable dhcpv6 brycec: You know what doesn't crash? pfSense :P m0unds: lots of other things don't crash either
i figured i'd see if routeros 6 still sucks on the 450g, and it does brycec: Bummer m0unds: still crash-happy, over a year since i last tried it brycec: (I know, I'm not helpful :p) m0unds: srx> show system uptime
Current time: 2014-12-05 23:51:05 MST
System booted: 2013-12-26 15:27:44 MST (49w1d 08:23 ago)
^ there's something that doesn't ever crash
haha
only reason it was rebooted was for a firmware update (flow support for ipv6) mercutio: 8:30PM up 1296 days, 9:43, 1 user, load averages: 1.25, 1.25, 1.24