[00:06] how do you guys use your monitor in the dark [00:06] do you lower the brightness? [00:06] sometimes [00:07] i guess they say to make sure the room is always lit [00:17] i pine after HMD's [00:18] and wish for HUD's [00:18] or VRD's [00:29] * up_the_irons envisions toddf as an android [00:50] *** mhoran has quit IRC (wolfe.freenode.net irc.freenode.net) [00:50] *** mhoran has joined #arpnetworks [01:58] *** visinin has quit IRC ("sleeeeep") [04:17] *** ballen has quit IRC ("Leaving...") [06:06] *** mhoran has quit IRC ("leaving") [06:07] *** mhoran has joined #arpnetworks [06:21] jeev: What about Google Chrome? [06:23] *** sroute has joined #arpnetworks [07:36] *** vtoms has joined #arpnetworks [07:38] *** ballen has joined #arpnetworks [09:04] *** ballen is now known as ballen|away [09:23] *** ballen|away is now known as ballen [09:51] *** ElectricBill has joined #arpnetworks [09:51] Wondering if I can drop an existing KVM box into an ARP host. [09:52] mike-burns, i dont like anything google does but i use them sometimes [09:52] not sure about chrome [09:52] ElectricBill, maaaybe up_the_irons would be the one helping you on it [09:53] jeev: I feel the same way about Google but was pleasantly surprised by Chrome. [09:53] Chrome on Mac is nice [09:54] ballen, ok - he on here much? [09:54] yea quite often [09:55] you can just use the contact form on the website [09:55] which will email him [09:55] or email support@ [09:55] but likely the web form would be more appropriate [09:56] or wait till he shows up here [09:56] Email is the fastest way to get his attention, but he pops in here every 10 minutes or so. [09:57] Chrome is great. Except for doing web development specific work, I use Chrome for all my windows-based browsing. If there were a freebsd port for Chromium I'd build and use it there too. [09:58] not to say there are not the odd hiccups, if you are following the developer-track of Chrome, but it has very quickly become highly usable [09:58] I wouldn't mind a GNOME-themed, Chromium-rendered browser. [10:04] damn, are old school xeons that bad? [10:04] CPU: Intel(R) Xeon(TM) CPU 2.80GHz (2800.11-MHz 686-class CPU) dual core version [10:05] vs CPU: Intel(R) Pentium(R) D CPU 3.40GHz (3391.51-MHz 686-class CPU) dual core version [10:05] and the pentium 4 got double the score.. and the xeon is scsi [10:09] abovenet messed up router config for my friend when they wanted to move his port [10:09] cost him 2 hours downtime [10:47] *** ballen is now known as ballen|away [11:08] every 10 minutes or so? [11:08] I'M ROBOTIC [11:08] ElectricBill: existing KVM VM -- what OS? what backing store do you use? [11:09] I'd need the exact existing 'kvm' command line used to runit [11:09] *run it [11:14] Ah more like an hour. [11:27] jeev: that's why you never just have one carrier. bgp anyone? mistakes will happen, you have to be prepared [11:28] jeev: interesting about the P4 vs. Xeon score [11:29] *** ballen|away is now known as ballen [11:42] up_the_irons: Have you done anything with libvirt and iSCSI? [11:42] mhoran: nope [11:42] not sure how much i like iSCSI... [11:42] seems to be getting popular though [11:42] It's practical, from a cost perspective. [11:45] yeah [11:45] mhoran: how do you run it? [11:46] up_the_irons: Apparently not the way the libvirt guys want me to. :p [11:46] hahaha [11:46] Each VM has a LUN, and the LUN is passed into the host server and connected directly to the VM. [11:47] Our iSCSI SAN provides snapshotting, etc, which is great. Snapshots are copied to a remote site every hour, and to our backup server (and into an rdiff-backup store, which is written to tape) every night. [11:47] I think passing a raw LUN directly to the VM is most practical, but the libvirt guys only use it as a store for multiple images. [11:47] Which is sad. :( [11:49] up_the_irons, I want to be able to arbitrarily load and images from my in-house systems... [11:49] s/and// [11:49] mhoran: i've noticed that about them; a lot of people using clustering filesystems on which to plop multiple VM images; instead of using block devices directly (which is how I do it) [11:49] up_the_irons: Yeah, I don't get it. [11:50] I keep them in LVs (LVM2) mostly. [11:50] mhoran: can't you do " ..." or something? [11:50] ElectricBill: you want to do it once, or on a continual basis? [11:51] without assistance [11:51] up_the_irons: Yeah, no problems with that -- but then I have to offload iSCSI login to open-iscsi (which is fine, and what I do right now). But libvirt has some handy config options which will automagically log into luns, which is nice. I guess I could write some udev magic. [11:51] mhoran: the snapshotting and copying to remote site is pretty cool, but doesn't it spike up the load on your SAN? [11:52] up_the_irons: Yup, but we have five of them, clustered. :) [11:52] ElectricBill: "without assistance" part is definitely not supported; would be a big deal to implement that. It is on the road map, but far out [11:52] up_the_irons, can one carve space into partitions and dynamically designate which is boot volume, with other volumes being attached to live image? [11:53] if so, I can do it myself, see? [11:53] mhoran: the reason I don't / can't snapshot my LV's and back them up off-site is b/c it would put FAR too much load on the host machine (degrade performance for VMs). If your SAN is hardware based and is able to do this w/o killing the box, that is quite cool [11:54] up_the_irons: Yeah, it's a combination of hardware/software. Was LeftHand Networks, now owned by HP. It's a nifty device. [11:54] mhoran: so instead, I use DRBD to sync the block devices to another machine, in real time [11:54] Basically targeted at virtualization, and exactly what we're doing. [11:54] up_the_irons, I snapshot LVs for subsequent copy, but I have control and responsibility for all VMs in the hosts. [11:54] mhoran: but a snapshot feature would be cool, b/c you can get images that are "back in time" [11:55] up_the_irons: Yeah, I looked at using DRBD for some replication stuff over here, since the SAN snapshots are hourly and we wanted up to the second sync of some data. However, one of our consultants had some horror stories, and we ended up not going that route. [11:55] ElectricBill: the "subsequent copy" part really taxes the host box, I've found, which is why i don't like it [11:55] up_the_irons: Yeah, exactly. [11:55] up_the_irons: I use it for testing all the time. It's quite helpful. [11:55] ElectricBill: "dynamically designate boot volume", etc... I'm not sure I understand your question there [11:55] up_the_irons, agreed, but if scheduled carefully, not too bad. [11:55] up_the_irons: The shared storage is great, too bad live migrations don't work in our Xen installation. :( [11:56] However, I think I may have the go-ahead to start playing with KVM. We'll see. :) First step is getting our Xen configs migrated to libvirt. [11:56] mhoran: too bad, I know you've been fighting that for a while [11:56] We're ordering that new E5550 system, and i told them that it won't work with the Xen 2.6.19 kernel we're running. So, we'll see. [11:56] up_the_irons, for example, at linode.com, I can allocate multiple volumes from an allocation, and designate some of them as "profiles", which are essentially boot images.... [11:57] up_the_irons, So I can treat one profile as raw data when I've booted from another, thus using my own VM to load another VM image [11:57] mhoran: ooO, E5550 [11:58] Hmmm... but... [11:58] I can't write the MBR, now that I think of it. [11:58] Yet it doesn't seem like a big deal. [11:59] ElectricBill: Linode features are far more advanced than what I've got here; [11:59] ElectricBill: you can boot your VM from a CD / DVD, and then do a custom install [11:59] *** visinin has joined #arpnetworks [11:59] ElectricBill: all VMs come with their base OS install CD "inserted" into the CD-ROM drive, so they can be booted from, if a re-install is required; accessing the VM in this case is over VNC [11:59] up_the_irons, OK, but it doesn't let me throw the images around from host to host as I do now in-house. [12:00] ElectricBill: right, no tossing images around :) [12:00] you could break a window, sheesh [12:00] ;) [12:00] I'd be shattered [12:01] wah wah wah... [12:02] up_the_irons, so then, a space allocation on arpnet is monolithic? [12:03] ElectricBill: you order a 20GB VM, you get a 20GB disk for that VM; that's about all there is to that [12:03] up_the_irons, is it partitioned? Is the MBR part of the alloc? [12:04] ElectricBill: it comes parititioned by default, but you could do the whole install over again, and partition how you like (over VNC). Imagine the 20GB being a 20GB physical hard drive. So yes, you get the MBR and everything else [12:06] I've edited disklabels, MBRs, partitions, you name it, on VMs [12:06] up_the_irons, I'm getting it (slowly). [12:06] (things that are also accessible to the customer) [12:07] up_the_irons, but no way to have, say, a 16GB /dev/vda and a 4GB /dev/vdb instead. [12:07] but it does require prior knowledge of how to do that stuff; there is no control panel for it. Some guys have booted their VM into single-user mode, or even from CD, and fiddled with fdisk, disklabel, etc... [12:08] ElectricBill: I could do that if you want, but it is a "manual" set up process, you wouldn't be able to change it on a whim [12:08] ElectricBill: like, I'd give your VM a 16GB 1st disk and 4GB 2nd disk, and you take it from there [12:09] The only thing I would want to change on a whim is the designated boot drive, sort of like the way it can be done with virt-manager. [12:09] speaking of vda... some other guys in here would be interested in knowing: yesterday, one of my customers, who is a Gentoo developer, got the para-virt disk going in his VM, with relative ease. Too bad it only works on Linux though.. [12:09] Oops - I'm wrong... [12:10] i should change all my Linux VMs over to para-virt.. [12:10] virt-manager only allows selection of disk vs. CD. [12:10] but you know what I mean? [12:10] ElectricBill: you want to be able to boot from any disk you like? [12:11] yeah. then I could use a small sys to manage the bigger volume when it isn't running. [12:11] ElectricBill: you can do this; you fire up VNC, reboot your VM, and hit 'F12' at the "BIOS" screen. It then let's you select the boot device [12:11] up_the_irons, duh. Of course! [12:11] :) [12:11] I think we're cooking now. [12:11] hehe [12:12] I have to run out to get meds for my son (home sick from school) ... [12:12] I will try to hook up later on this. [12:12] Thanks for the brainstorm ...or drizzle? [12:13] ElectricBill: note, however, there is no "reboot" option in the control panel (Portal) yet, so reboots would be done within your VM. If you need a hard reboot, you file a ticket. I *just* got a prototype working last night, of a system, that will allow customer controlled "start [12:13] ", "shutdown", "hard shutdown", etc... of VMsw [12:13] *VMs [12:13] ouch. [12:13] but it isn't released yet [12:13] That won't work for me, but I assume it's on your short list. [12:14] yeah, it's on the short list [12:14] ok - thanks again - later. [12:14] ElectricBill: yeah, that's why I pointed it out, won't work for everyone and I don't want to disappoint later [12:14] ElectricBill: later [12:15] cd $office [12:50] I wondered where bacon donuts came from. Now I know: http://baconbaconbacon.tumblr.com/post/63922776/bacon-donut-this-is-the-most-tasty-bacon-donut [13:35] *** vtoms has quit IRC ("Leaving.") [13:35] *** vtoms has joined #arpnetworks [14:46] *** ballen is now known as ballen|away [14:57] *** vtoms has quit IRC ("Leaving.") [15:39] oh man [15:39] i want in n out [15:39] eh, it's at the nickel diner [15:39] that place visinin likes [15:39] either way, that bacon donut should be illegal [15:53] it's too delicious to be illegal [15:57] lol really [15:57] how oily is it [16:14] not oily at all [16:14] actually [16:47] heart attack on a plate [16:47] They should stuff one with deep fried butter for good effect. [16:59] heh [16:59] i need to learn fax over ip [17:01] *** ballen|away is now known as ballen [17:01] *** ballen has quit IRC () [18:04] up_the_irons, what happened with your powerconnect [18:04] jeev: it's up and running; works pretty good [18:05] jeev: can't graph vlan's, but luckily I don't need it to do that. the fact that it won't do that means I could never use it as a customer access switch, but works great for I/O backplane (backups,DRBD, etc...) [18:05] why can't you graph [18:05] oh, vlan's as in [18:05] multiple port vlans ? [18:05] it only graphs the ports and port-channels [18:05] ah [18:06] but if I make a VLAN, say 105, i won't get a graph for 105 [18:06] i duno how to get it to sync time [18:06] but on Cisco, it'll graph all that stuff :) b/c a Vlan becomes a bonafide interface [20:31] *** heavysixer has quit IRC () [20:51] hi [20:52] hi [23:06] *** Rada has quit IRC (Read error: 60 (Operation timed out)) [23:17] *** visinin has quit IRC ("sleep") [23:20] opteron 2220 decent ? [23:27] still in warranty [23:27] http://cgi.ebay.com/HP-proliant-DL365-G1-2X-Dual-Core-2220-2-8GHz-8GB-RAID_W0QQitemZ330364662943QQcmdZViewItemQQptZCOMP_EN_Servers?hash=item4ceb44b49f [23:27] nov 2010 [23:27] sort of worth it, i'd offer $500 [23:28] HP's are usually decent [23:29] unfortunately the drives are expensive [23:41] SAS? [23:41] or plain SCSI? [23:46] sas [23:47] 72-74 gig whatever it is [23:47] is 150 bux [23:47] those are the drives i have now [23:48] SAS is crazy expensive, just like SCSI