ahh, can't run xen on freebsd randal: you can't run the base vm on xen, but you can run subsequent ones. I guess I don't understand that yet. xen isn't like virtualbox? the first dom0 is sepcial in that it by default provides the network/disk for vm's. xen is not like virtualbox xen is like vmware esx apparently it can actually pass that stuff onto to other vm's. brycec: it's not quite like either of them really. well it's more like esxi. Yeah, ther's a term for it, I can't seem to recall something about layers so you have a hypervisor that runs a series of virtual machines, but the first one gets "direct" access. and it then has a memory window or such to pass data to other virtual machines for disk/network. Ah here we go https://en.wikipedia.org/wiki/Hypervisor i wonder if i'm making it sound more complicated. Hypervisor :: A hypervisor or virtual machine monitor (VMM) is a piece of computer software, firmware or hardware that creates and runs virtual machines. A computer on which a hypervisor is running one or more virtual machines is defined as a host machine. Each virtual machine is called a guest machine. The hypervisor presents the guest operating systems with a virtual operating platform and manages the execution of the guest operating... Yeah, dom0 is technically a vm but it's treated special. It's similar, but not identical to the way esx works with having a base OS+Hypervisor brycec: well esxi runs linux .. you can actually get a linux shell so it may actually be quite similar, i haven't really looked uunder the hood much it has very limited drivers unlike linux you can't even do software raid with esxi (grr) For the sake of simplicity and explaining things to RandalSchwartz... so if you want a test server it's kind of annoying. brycec: true. ok yeah, it's like sxi :) esxi btw: vmware is currently getting sued for violating the gpl in esxi you can actually host a virtual machine inside esxi that provides storage to esxi. ant: makes sense. virtualception people were doing solaris/opensolaris/etc with zfs inside a virtual machine, then doing nfs from that for the esxi data store. so I could run xen on my freebsd box, and use linux for the first vm, and freebsd for the other ones? using vt-d to pass through a scsi controller generally, but i don't think that part is /necessary/ it just improves performance. Probably. (I don't know much about FreeBSD Xen compatibility) randal: yeah, but you'd do a linuux install, then a xen install, then rnu a freebsd vm. (I guess so https://wiki.freebsd.org/FreeBSD/Xen) tbh i don't relaly think for most people xen is any better than kvm. Much happier with KVM and the ilk, they don't require any guest customisations it wsa better on older hardware as it had paravirtal vm's. but modern cpus have really accelerated extensions for normal vm's. and now it's faster to have non paravirtualised. If I can't run MSDOS as a virtual machine, what use is it? :P vmware took out it's paravirtal functionality. i think in esxi 5 xen now has paravirtual, hvm, and a hybrid. hvm is full virtualisation like kvm. the only real advantage i find with xen is it makes it easy for me to host kernels outside the virtual environment and boot specific kernels. oh wow that was not worth it twss Okay! twss! 'oh wow that was not worth it' there's also better support for video/pci device passthrough. i actually went win 2k8 with hyperv 2012 and it makes xen super user friendly but kvm is working to improve their stuuff, and it's mostly because xen forked qemu. but they're going back to non forked. so if you want to do video passthrough on xen, at least before, it worked better to use the "legacy" qemu option. video passthrough on this lenovo in xen was fun by better i mean "it worked" rather than "it failed". The way I see it, at least in my workloads, there are two types of useful virtualisation: full machine a la KVM, or containerised a la OpenVZ. Everything else outside that scope is just more complicated than it's worth. heh brycec: i actually like xen windows hyperv is fast, but i am seriously disliking it freebsd guests are still faster in xen tho grody: were you doing paravirtual or full virtual for xen in freebsd? even more so when you PV fbsd i've tried freebsd with xen and vmware xen has this annoying problem with mtu's above 1.5k firstly hvm then pv grody: you should hvm. it actually works better. the pvhvm thing should be better still not hit that yet, doung pfsense with tagged vlans in xen and it seems fine but yeah, there's a hvm kernel config. grody: this was real larger mtu's maybe vlan doesn't add enough ah there's a hard coded limit in some driver. it's fixable yea im still legacy 10/100 here i still couldn't do over 2k but slightly larger worked. it always erks me when i seem to hit lots of "weird" problems, and other people don't. haha and you google high and low only to find your question asked but never answered grody: that's what i say haha grody: and then ytou find out it's been happening for years. i remember when i heard about the windows uptime bug it crashed after 48 days of uptime or something with a wrap around. i'd never hit that buug :/ (i did use that dirty windows for a bit) xen 4.1 says my mobo doesnt do iommu, but 4.5 is ok with it, couldnt find that one anywhere i went evil buying a new tablet i got the linx8 win8.1 .. now running arch grody: it's probably because new motherboard and old scipset. also 5520 is blackedlisted. getting a vkeyboard up was fun, had to ssh most of it err blacklisted. and so vt-d is strictly possible, but things disable it not that you're using that chipset mercutio, no idea it is a pain of a lappy lenovo s205 uefi is skiwiff too i was doing vt-d on z77 for my gpu pass through not all motherboards supported it. oh amd you like amd don't you grody :) wel for a fraction of pruce to get eq. its the pepsi of cpus i have ex-lease laptop, it was old when i got it and a chromebook i am thinking of sticking that special boot thing on it linuxbios? err coreboot it seems now it has the google one, but if it runs low on battery then it reverts to only booting chromeos. and it's booting in legacy mode and you have to press ctrl-l as you boot i kind of wish coreboot was used more by now. i kind of hoped it'd take off. it is interesting that google are using it though uefi isn't really that wonderful. That's what she said!! and it's huge and complicated. really the boot loader should just be doing the init system, then providing resources to the initial os. err hardware init i'm ok with things like a little gui to overclock, and update bios from the gui etc. but in the normal course of boot it should juust do the minimum, and not have a hugely massive api http://www.uefi.org/sites/default/files/resources/2_4_Errata_B.pdf that was fun what was? This turned out to be pretty entertaining http://www.therestartpage.com/ (click things, move windows, fiddlw with stuff. It's interactive.) heh... that's cute I'm booting openstep Steve's Soul powers these machines. :) ok i hate windows hyperv im going back to debian xen it lags like hell running freebsd guests linux is fast as hell mind network lag? is anyone else seeing ping spikes to 4.2.2.2 but fine to 4.2.2.1. i assume 4.2.2.2 is probably getting ddos'ed or something, it's just curious and my usual "test ip" I don't monitor it... But I'm not surprised 4.2.2.2 is the ip "everybody" knows and hits 4.2.2.1 is far less-known mercutio: 4.2.2.2 has always had terrible ping latencies for me http://kremvax.acfsys.net/smokeping.cgi?target=Remote.l3dns acf: it's been mostly fine for me before i don't monitor to it i just check internet stuff with it at the time wow it's really bad for you since like the end of june but before that it was fine is that on a vm? no, that's ARP Metal interesting yea, it's super strange i was just lookig athe localhost ping time i suppose it's pretty stable, it jut seems high mine sits around 10u on an i3-2100 http://paste.unixcube.org/k/479429 yeah i'd already tried from arp native well vm on arp yea, it's not going through anything strange afiact but i did manage to test from one location that seemed better it was better when it hit san jose 4.2.2.2 rather than los angeles 4.2.2.2 ah right, that's anycast yeah, but los angeles seems worse than san jose like way worse maybe I should start monitoring 4.2.2.1 i haven't got historic data, and you do :) i wondered if it was some new ddos or something i had weird slow gtt behave earlier today then i saw someone reporting about gtt issues on outages is *that* what that was oh you noticed weird gtt shit? it was pretty chronically bad. I didn't have much time to debug at the time but something was clearly not right i still don't know who gtt is compromised of properly. but their network seems a badly managed hodge podge atm i think it's gtt, tinet, nlayer, mzima, and someone else something like that I guess gtt just bought everyone up what and didn't really try to integrate anything try goign to www.gtt.net umm it's not loading, and saying "checking your browser before accessing gtt.net" yea and to allow up to 5 seconds it's been way more than 5 seconds. appears to be some Cloudflare anti-ddos thing yeah so gtt had massive ddos attacks today i imagine i thought gtt were bigger than cloudflare it's kind of scary when gtt are doing anti-ddos protection with cloudflare and their main web site doesn't even work I think cloudflare are super experts at anti-DDOS though what's that web host tracker thingy for HTTP anyway the one that usedd to say about apache vs iis netcraft? that's it oh www.gtt.net loads now it did for me, after > 5 seconds :P http://toolbar.netcraft.com/site_report?url=http://www.gtt.net so it looks like it was self hosted 5th feb 2015 huh yea just clicked refresh there That's what she said!! then the CloudFlare one showed up it only recently changed from windows to linux too https://www.staminus.net/gtt-stops-offering-flowspec-what-this-means-to-ddos/ cloudflare use ntt at least what i've noticed for communication to origin web sites