[06:05] *** pyvpx has quit IRC (Remote host closed the connection) [06:24] *** pyvpx has joined #arpnetworks [09:30] ubnt announced a pair of new little edgerouter devices - $49 and $79 respectively [09:31] link? [09:31] tiny little guys [09:31] https://www.ubnt.com/edgemax/edgerouter-x/?utm_source=Ubiquiti+Newsletter+Subscribers&utm_campaign=118f9d5b9e-EdgeRouter_X_X_SFP_03_01_15&utm_medium=email&utm_term=0_1c1b02cb37-118f9d5b9e-238482017&mc_cid=118f9d5b9e&mc_eid=d136c500ee [09:31] Dayuum [09:31] MIPS cpus, 130kpps w/64byte pkts [09:31] yeah, cool little things [09:32] i still wanna try out an ERL, might do one of these instead since i don't think i'd replace my gateway with it and it's 50% of the price of the ERL [09:32] as soon as you turn on anything, that pps sinks like a cinder block [09:32] for its intended market it's fine [09:44] if you were looking for something that has full feature support at line rate, you're looking at ASIC powered stuff...doesn't really compare to a $50 or even a $99 box at all [09:48] my wife forgot to tell me she was selling a bunch of textbooks back to amazon. i recycled most of my boxes yesterday. DOH [09:53] amazon has a buyback program? [09:53] yeah, no idea when they started it [09:54] whoa [09:54] cool [09:54] prepaid shipping as usual [09:54] if they won't accept the books you registered to sell back, they'll just ship them back to you [09:55] If the book you want to sell is found not to be in good condition, or is the wrong ISBN, your account will not be credited and the item will be returned to the addresses you select when creating your trade-in. [09:55] hmm [09:55] yeah... [09:55] that might be a free way to ship books to places... [09:55] :P [09:55] lol [09:55] yeah, these things are heavy (medical textbooks) [09:56] this is amazing [09:56] That's what she said!! [09:56] ive pretty much given up on trying to sell my textbooks since they are so old [09:58] yeah, my wife tried selling her undergrad stuff right out of college and couldn't find a bookstore that wanted them [09:59] but it seems like amazon wanted about 80% of the graduate stuff. i just hope they don't send them back, lol [10:25] Nice [10:25] I should look into that... [10:25] That's what she said!! [10:25] At least get a couple of bucks [10:25] BryceBot: no [10:25] Oh, okay... I'm sorry. 'I should look into that...' [10:43] *** staticsafe has quit IRC (Ping timeout: 272 seconds) [11:10] *** staticsafe has joined #arpnetworks [11:15] *** anisfarhana has joined #arpnetworks [12:03] *** RandalSchwartz has joined #arpnetworks [12:03] *** RandalSchwartz has quit IRC (Changing host) [12:03] *** RandalSchwartz has joined #arpnetworks [13:28] oh super [13:29] 8.8.8.8 broke here [13:29] You in Turkey? [13:29] California [13:29] Different kind of oppression, I guess. [13:29] :P [13:30] acf@vermilion:~$ nslookup google.com 8.8.8.8 [13:30] ;; connection timed out; no servers could be reached [13:30] mtr pls [13:31] http://paste.unixcube.org/k/a3c85f [13:31] strange. I didn't see packet loss in the mtr earlier [13:31] lol mike-burns [13:31] but DNS still wasn't resolving [13:31] like right now [13:32] 0% packet loss, DNS not resolving [13:32] http://paste.unixcube.org/k/7a9c17 [13:33] 8.8.4.4 is also affected [13:33] probably hitting a down anycast node [13:33] it responds to ICMP though [13:34] it's also intermittent [13:34] It's responding to me now [13:34] but a few minutes ago I also timed out [13:35] yea, it's ok for me now too [13:35] nvm, down again :P [13:37] still wfm [13:42] It works here, beautiful. [13:50] i wouuldn't use 8.8.8.8 and 8.8.4.4 as my two dns servers [13:50] you could use one of them, and one of something else. [13:50] 4.2.2.1 [13:51] personally i like the idea of 3 dns servers. [13:51] You glutton. [13:51] but normal resolvers have long timeouts. [13:54] 192.168.1.1 and 1.2 ? [14:00] it'd be nice if freebsd would shift to openntpd [14:00] Sure but NIH. [14:01] there seems to be another ntp vulnerability on the freebsd security list. [14:01] it didn't sound that bad, but hard to know [14:01] freebsd didn't make ntpd did they? [14:02] Well phk@ announced a re-write. [14:04] *** m0unds has quit IRC (Quit: WeeChat 1.1.1) [14:07] *** m0unds has joined #arpnetworks [14:08] oh? [14:08] so it's going to compete with openntpd? [14:08] http://phk.freebsd.dk/time/ https://github.com/bsdphk/Ntimed [14:08] Yeah. [14:12] ahh it seems it may be targeting to be more full featured than openntpd [14:13] In the mean time, openntpd is in FBSD ports. [14:13] Excuse me guys. [14:14] Is it normal I don't understand what both of you talking about? [14:14] It happens to me a lot, so yes. [14:14] Ok carry on. [14:14] Is there anything in particular you want to know more about? [14:15] I just thought my sickness level increasing..thats all. [14:15] Doctor confirmed that i have a disease few weeks ago. [14:17] *has confirmed [14:24] ntp is network time protocol [14:24] I know that at least. [14:24] it keeps servers clock's synchronised over the internet. some people have "good clocks" and run servers, which are then connected to by people who have good connectivity [14:24] and then users connect to those hosts with good connectivity. [14:25] I had experienced with master clock before. [14:25] the legacy ntp daemon has been around forever, and is a lot of old code. [14:25] kind of like sendmail, bind etc. [14:26] openntpd is a kind of minimal implementation with security in mind, with security being preferred over feature-set. afaik it doesn't have hardware support for good clocks to feed etc. [14:26] Why bother using ntp if we know it had problem before? [14:26] well openntpd is a little newer. [14:27] Whats wrong with single clock on your machines? [14:27] and ntp can be tweaked to be less ... open [14:27] normal clocks on computers lose time a lot [14:27] So people use ntp just because it keep clock sync over the net? [14:27] it varies a lot. [14:27] ntp is good if you want to have second-level accuracy. [14:27] so if you have two servers and look at logs on each, you can correlate times. [14:27] and NTP is useless, if you have stand alone server, with no internet connection? [14:28] if you don't use anything your clock may be 10 minutes+ out of sync eaisly. [14:28] with things like dsl modems etc that have ntp support they don't even have a hardware clock. [14:28] I see. [14:28] so the time will be wrong by months/years/etc. [14:28] I don't have any problem with the accuracy of the clock before. [14:28] ntp can work with a gps clock [14:28] I mean without NTP [14:29] http://www.ntp.org/ntpfaq/NTP-s-refclk.htm [14:29] windows uses ntp [14:29] i've had problems with clocks before. [14:30] Not if you untick the options for Internet Server in windows. [14:30] You guys are amazing [14:30] You can tell a story about clock and NTP like 24 hours non stop. [14:31] we weren't talkin gabout ntp for 24 hours [14:31] You have your fact, I have my own fact, debate, arguing, go silent,quit, come back, and another issue. [14:31] Wonderful. [14:31] Another story/issue etc etc [14:32] that's kind of what IT is like, yes, anis. [14:33] But IT is killing me. [14:33] how so? [14:33] It maybe can kill you too. [14:33] So don't tell people in future nobody telling you that IT can kill you. [14:33] life kills you [14:34] it's all part of the cycle. [14:34] Correct. But my case, IT is killing me. [14:34] If not because of IT, i probably can be better. [14:37] 5.35 AM [14:37] Sleep time. [14:37] Bye bye mercutio and mike-burns [14:37] Ciao. [14:37] It was nice talking to both of you today. [14:38] in news today, gmail let their secure certificate expire. [14:39] What? The cert I have loaded is good from 25 March through 22 June, 2015. [14:40] for smtp to smtp i think [14:40] someone was complaining on outages@ a few days ago [14:40] and then someone else said it was fine for them [14:40] (I have another tab open to gmail with 19 Mar - 16 June, 2015. Interesting) [14:40] i imagine it wasn't expired for that long. [14:40] (looks like Google rotates certs every three months) [14:41] i can't see the emailp, maybe itw asn't outages@ [14:41] oh it was nanog [14:44] (Looks like smtp.gmail.com serves just a single certificate, valid 18 Feb to 31 December 2015) [14:45] (not the same as the reported issue above, I know, just observing different certificate practices) [14:46] it was smtp.gmail.com on nanog [14:46] these certificates have weird times on it [14:46] orly? huh [14:46] err dates [14:46] like why 18 feb to 31 dec [14:46] i wonder if they're backdating them a bit for people with off clocks [14:47] I feel like it's more likely they had that cert all along and it may just not have gotten pushed [14:47] having lots of ssl errors in general is often due to wrong fclock. [14:47] for 9.5 months? [14:47] why would that be "for 9.5 months"? [14:47] it's still a little strange. [14:47] @date Feb 18 2015 [14:47] 18 feb to 31 dec [14:47] 6 weeks, 6 days, 13 hours, 46 minutes, 25 seconds ago. [Interpreted date: Wed, 18 Feb 2015 00:00:00 -0800] [14:47] oh that's more than 9.5 months [14:48] Right. I'm suggesting that they forgot to install the certificate for 6 weeks [14:48] that's 10.5 months. [14:48] But the duration is odd [14:48] oh i mean duration rather than when it was there [14:51] Yeah, i was responding to your remark about backdating. [14:52] oh right [15:00] damn my level3 test site shifted to ntt :( [15:08] performance decreased a few days ago. [15:08] so it probably shifted then. [15:27] was that what caused their outage a couple days ago w/smtp? [15:42] m0unds: yeah would have been [15:42] although i think it was 3 days ago [15:55] *** anisfarhana has left "Leaving" [16:01] * m0unds shrugs [16:01] i don't use smtp, i just got alerts from google about an outage some number of days ago [16:03] s/don't/didn't [16:03] i didn't use smtp, i just got alerts from google about an outage some number of days ago [16:05] smtp as in port 587 / 465 for sending mail ? [16:06] yes, i don't use a mail client [16:06] neither do I [16:06] desktop mail client, rather [16:06] all web [16:06] for me [16:07] but in this case, their smtp server was returning an invalid cert - i was able to send mail via gmail interface w/out delays or errors, so i guess it only impacted people sending via smtp.gmail.com or whatever [16:07] http://venturebeat.com/2015/04/04/gmails-ssl-certificate-for-smtp-appears-to-have-expired/ [16:07] yeah [16:09] what the fork [17:12] I am trying to follow http://evilrouters.net/2009/08/21/getting-bgp-routes-into-dynamips-with-video/ [17:13] with current routing table [17:13] but I keep getting: Segmentation fault (core dumped) [17:13] command is : zcat bview.20150407.1600.gz.1 | bgpdump -m - > myroutes-april.1 [17:14] sounds like bad code :) [17:14] i dunno why dynamips is [17:14] oh are you just trying to get a full route table? [17:20] yea [17:20] full table [17:21] works with code from 2009 [17:21] bview.20090820.2359.gz [17:21] I have that file, but wanted more recent routes [17:30] maybe you should just get a bgp session? [17:32] https://github.com/YuZhang/bgpdump-zy/blob/master/ChangeLog [17:32] apparently there were segfaults with empty as paths. [17:32] sometimes these things aren't kept current in distributions [17:49] A+ to up_the_irons :) [17:50] * brycec is finally connected to his company's ARP Metal box :D [17:51] Not exactly blown away by SuperMicro's GUI, but meh, as long as it gets the job done. [17:55] brycec: use ipmi :) [17:55] Error: Unable to establish IPMI v2 / RMCP+ session [17:55] uhh it should work [17:55] are you using the vpn? [17:55] Yep [17:55] (as evidenced by the fact I'm on the web gui :p) [17:55] true :) [17:56] try -L USER -I lanplus [17:56] ipmitool -I lanplus -H IP -U myuser lan print [17:56] was what I tried [17:56] hmm [17:56] (with IP and myuser replaced) [17:57] I'm also getting "You don't have permission" popups on certain sections of the web gui, which seems odd. [17:57] eg: Maintenance -> System Event Log [17:58] Set Session Privilege Level to ADMINISTRATOR failed: Unknown (0x80) [17:58] Now I get that ^ from ipmitool [18:00] oh gah [18:01] * brycec plays with -L Force session privilege level [18:01] yeah that's happening to me now too [18:01] but -L USER fixes it [18:01] heh (I swear I didn't break YOUR Metal box :p) [18:02] haha [18:02] the gah was mostly about having a copy of my old key [18:02] but yeah it works for me [18:02] i just need the -L USER [18:02] so try ipmitool -L USER -I lanplus -H IP -U myuser lan print [18:02] Can confirm, -L USER gets me no error [18:03] Also, "lan print" is blank :/ [18:03] same here [18:03] try sensor [18:03] (as is "lan print 1") [18:03] s/lan print/sensor/ [18:03] (as is "sensor 1") [18:03] well not that hah [18:03] sol works :D [18:03] yeah [18:03] sweet [18:04] Which is really the most critical one :p [18:04] it's better than using web [18:04] well that and power/reboot [18:04] sensor lists several sensors with no values. [18:04] hmm it works fine for me [18:05] "sensor get ..." seems to work, slowly. [18:05] At least gives me info about the sensor [18:05] sensor is slow here too [18:05] it does a rtt for every line i think [18:06] "sensor reading FAN" has no reply though [18:06] (but sensor get FAN" dumped info about it.) [18:07] yeah i wouldn't worry about sensor normally [18:07] i just find it an easy thing to check ipmi is working [18:08] Ditto. Though I would like to be able configure alerts to notify me on failure... [18:08] youu can do that inside the machine [18:08] (that's one of those "You don't have permissions" popups in the web gui) [18:08] That's what she said!! [18:08] BryceBot: no [18:08] Oh, okay... I'm sorry. 'youu can do that inside the machine' [18:10] brycec: did you go for ssd's? [18:10] Pretty sure it's just the standard 1TB [18:10] ahh ok [18:10] that's what the portal says [18:10] just 1? [18:10] (I inherited the system) [18:10] Looks like the base specs [18:10] ahh ok [18:10] damn i'm paranoid about not having raid :) [18:10] Yep based on price, it's the "Starter" level [18:32] mercutio: Tip for the future, -L OPERATOR [18:32] how's that different from -L USER ? [18:32] USER doesn't have permission to "power reset" ;) [18:32] oh [18:32] (It's a notch above USER on the scale) [18:33] i can't say i tried power reset :) [18:33] And corresponds with what's listed under "user list" [18:33] well sensor still worked with it, thanks. [18:33] :P 18:02:55 ⤷ | well that and power/reboot [18:33] yeah i've tried it with ipmi in general, just not on arp [18:34] * brycec goes back to breaking into his own server [18:34] That's what she said!! [18:34] init=/bin/bash [18:39] OpenBSD ;) [18:39] boot -s [18:39] hehe [18:39] i assume everyone uses linux these days for some reason [18:40] brycec: are you enjoying the gigabit connectivity on your METAL box? [18:40] Not yet [18:40] since I have no login on the box :p [18:40] heh [18:41] http://www.openbsd.org/faq/faq8.html#LostPW [18:41] will there be virtualization on that box? [18:42] openbsd doesn't do virtualisation. [18:42] lol mercutio thanks, I'm familiar ;) [18:42] oh ok [18:42] *will* there be? Maybe, I might reinstall with FreeBSD or Linux for Qemu/kvm [18:42] ctrl-alt-del not working by default is a bit irritating. [18:42] oh right. [18:46] sonofagoddamnmotherfucking... I wish Java would stop trying to use Firefox's proxy settings. *sigh* [18:49] (Sorry, that was out of line. Just frustrates me.) [18:50] it does seem strange [18:50] i thought things didn't generally use firefox's proxy settings. [18:50] Linux desktop running Awesome, so there's no DE for Java to scrape the proxy settings from, so it falls back to using Firefox's for some reason. [18:51] I should rephrase that - there's no DE capable of providing proxy settings for Java. [18:51] reading http_proxy is the standard [18:51] that's what chrome, wget, etc use for proxy [18:51] I only use Firefox when I have to bounce over a SOCKS/SSH proxy/tunnel because FF's proxy settings are quick and easy to twiddle. (Chrome is a command-line switch only, or environment, either way requires restarting Chrome) [18:51] But, meh [18:52] yeah i've done similar. [18:52] it's kind of annoying that you can't change proxy easily in chrome [18:52] i used to use some extension to do it for me [18:52] although in my case it's usually to disable proxy [18:53] https://chrome.google.com/webstore/detail/proxy-switchysharp/dpplabbmogkhghncfbfdeeokoefdjegm [18:53] that's what i used, iirc [18:54] this looks handy [18:54] pretty painless, you can set up profiles for different proxies [18:54] i used an ssh proxy to evade web filtering at work because it was obnoxious and getting anything whitelisted was awful [18:54] Thanks for the link [18:54] heh [18:55] so i set one up for direct, one for one box, one for another and could just switch it by clicking on the icon and selecting a profile [18:55] pretty handy [18:55] That is handy. [18:55] Personally, I'm going to stick to using a separate browser. Minimizes my disruption. [18:58] m0unds: can that proxy switcher work on a per-tab basis? [18:58] Oh man, going through some old authlogs and seeing attempts for "xbmc" and "xbian" really has me worried that people are connecting their RPi's directly to their Internet connection :/ [18:59] m0unds: if you didn't notice, "SwitchyOmega is the 2.x version of switchysharp and should be used instead." [18:59] (Not sure if you still use the extension since you changed jobs) [19:00] (That's from the github) [19:00] mnathani_: looks like Omega can switch by host/url [19:01] *** nesta_ has joined #arpnetworks [19:01] https://chrome.google.com/webstore/detail/proxy-switchyomega/padekgcemlokbadohgkifijomclgjgif third slideshow image [19:02] "You can tell SwitchyOmega to switch between proxies automatically through the mighty Switch Profile." (from the quickstart guide) [19:02] I think I have lost count of the number of new things learnt from this channel [19:03] *** nesta has quit IRC (Ping timeout: 265 seconds) [19:03] *** nesta_ is now known as nesta [19:03] *** m0unds has quit IRC (Ping timeout: 265 seconds) [19:03] lol [19:04] w00t, can confirm it does automatic switching [19:04] So it's potentially better than per-tab [19:05] you mean I can watch netflix now without using Firefox [19:05] *** m0unds has joined #arpnetworks [19:05] awesome! [19:05] Yes [19:05] But I'm sure any proxy-changing extension would have allowed that :p [19:06] http://www.open-xchange.com/dovecot [19:06] has anyone heard of this? [19:06] Heard of Dovecot itself? yes of course. [19:06] nah open-xchange [19:07] apparently powerdns and dovecot are now part of some "open-xchange" company thingy? [19:07] I think I heard of it awhile back trying to build an Exchange-compatible alternative, but I might just be confusing very-similar names [19:07] their website is too template like for my liking [19:08] heh [19:08] i was always impressed by dovecot [19:08] it seems on a similar level to postfix to me [19:08] dovecot is an mta, or imap provider? [19:08] it's a imap provider that has a mda [19:08] mda [19:08] (what mercutio said) [19:09] (pop too I think) [19:09] yeh pop too. you don't have to use their mda too. [19:09] so you can use postfix/procmail/dovecot as the mda bit [19:09] like any of them. [19:09] procmail is the easiest known example of a mda [19:09] postfix/dovecot do other things as well, [19:10] basically dovecot adds indexes on top of normal mail, and if you let it deliver mail for you it'll update the indexes automatically [19:10] otherwise it'll update indexes when you check your mail. [19:10] but it's also got some filtering capabilities with less arcance syntax than procmail. [19:10] personally i still use procmail [19:12] hmm dovecot has 1000 euro security bounty too [19:26] brb reboot [19:27] *** mnathani_ has quit IRC () [19:39] *** mnathani_ has joined #arpnetworks [19:39] brycec: didn't notice, but hadn't used it since sept. thx for heads up though because i did still have it on chrome, just disabled [19:40] https://twitter.com/davidu/status/585632424626905088 [19:40] TWITTER: Re: The Google DNS attack, we see backscatter in our telescope for requests for shifen[.]com. This graph should be 0. http://t.co/Pcm4dTuN8p (Wed Apr 08 02:37:05 +0000 2015) [19:45] I added it (Omega) and disabled it, just to have it in case I ever need it :p [19:56] works for me, haha [20:02] *** mjp has quit IRC (Remote host closed the connection) [20:04] *** mjp has joined #arpnetworks [20:19] Aw my Metal's gbps connection seems to max out at 100mbps, v4 :/ [20:19] that seems strange [20:19] to where? [20:20] Link is GbE, but pulling http://www.azurespeed.com/Azure/Download cruised at 11.98MB/s [20:20] try curl http://la.meh.net.nz:24/10m > /dev/null [20:20] * brycec finds other files [20:20] i reckon it should go 800 megabit [20:20] Ooh, CDN is faster, 18MB/s [20:20] http://cachefly.cachefly.net/speedtest/ [20:21] Next to impossible to gauge GbE speed with a 10M file, but it clocked at 44MB/s [20:21] brycec: it's not impossible :) [20:21] there's testfile.zip [20:21] for 200mb [20:21] 15MB from cachefly [20:21] 44mb/sec is nice. [20:22] 200MB file at 26MB/s [20:22] oh you're using openbsd [20:22] I guess my initial test was just crap [20:22] openbsd's network static isn't great for fast speeds. [20:22] Meh, it's good enough. And now that I've proven I can exceed 100mbps, I'm happy [20:22] Thanks mercutio and m0unds [20:23] if you don't have tcp timestamps enabled on the server, it won't increase it's receive window above 16k [20:23] sho 'nuff [20:23] and it caps window size at 1mb by default in the source iirc [20:24] it also increase the window size a bit slow for cubic [20:24] fortunately for sending it works pretty good [20:24] Looks like OpenBSD is topping out at 75MB/s to mirrors.arp [20:25] mirrors can do more than that [20:25] aka 600mbps [20:25] try downloading the same file again [20:25] 88.11MB/s [20:25] yeah that's more like it [20:25] (good thinking) [20:25] 704mbps [20:25] nice [20:25] i've had 100 million bytes/sec from it before [20:26] (i think it was divide by 1000 rather than divide by 1024) [20:26] aka 100 mebibytes/s [20:26] isn't megabyte/1024 ? [20:26] hm, shoot you've got me second-guessing myself now :p [20:26] actually it is /1000 [20:26] megabyte = 1000, mebibyte = 1024 [20:26] hangon that's weird [20:27] mebi = binary = 2^X power [20:27] but a kilobyte is 1024 bytes/. [20:27] should be 1000 technically [20:27] @wa 1 kilobyte to kibibytes [20:27] convert 1 kB (kilobyte) to kibibytes;0.9766 KiB (kibibytes);1000 bytes;8 kb (kilobits);8000 bits;7.813 Kib (kibibits);0.007629 Mib (mebibits);information;[information];Words of data on an 8-bit system:, ->1000 words, ->500 longwords;Words of data on a 16-bit system:, ->500 words, ->250 longwords;Words of data on a 32-bit system:, ->250 words, ->125 longwords;Words of data on a 64-bit system:, ->125 words, ->62 longwords [20:27] 1000 ^^ [20:29] but 16 megabytes of ram is 16777216 bytes [20:30] @wa 16 megabytes to bytes [20:30] Sorry, I couldn't reach the backend API. [20:30] @wa 16 megabytes to bytes [20:30] BryceBot: TRY HARDER [20:30] Sorry, I couldn't reach the backend API. [20:31] you broek it? [20:31] I dunno, not my backend API :p [20:31] The lazy can just https://www.wolframalpha.com/input/?i=16+megabytes+to+bytes [20:31] Answer: it's 1.6*10^7 [20:31] so, 16000000000000 [20:31] (I lost count of 0's) [20:32] Hmmm what to install on this box... Vmware? Proxmox? XenServer? FreeBSD? So many choices... [20:33] has it got more than 8gb of ram? [20:33] No [20:33] what's it going to be used for? [20:33] It has no purpose [20:33] But the boss doesn't want to give it up just in case we need it [20:33] (silly!) [20:33] haha [20:33] (we have a spare VPS for that reason too) [20:33] linux/kvm then [20:34] aka Proxmox [20:34] it means if you do want to use it [20:34] you can just copy an image over etc [20:34] I personally <3 Proxmox, but I'm open-minded [20:34] haven't used it [20:34] meh I can do the same with Vmware or FreeBSD (to greater or lesser extends) [20:34] i use xen [20:34] *extents [20:35] vmware on 8gb ram? [20:35] * brycec shrugs [20:35] I'm sure it would install >.> [20:35] well youu need a raid controller if you want to do raid with it [20:35] But I'm fairly inexperienced when it comes to vmware [20:35] or you need to do the virtual machine that acts as storage [20:36] yeah it'd install [20:36] I'm sure I can't sell the boss on this either, but perhaps I could free up the VPS and move them all over [20:36] * brycec thinks it's silly to pay for 3 VPS *and* Metal [20:36] yeah shift to 1 vps + metal maybe [20:37] What reason is there to keep any vps at that point? just reducing spof? [20:37] yeah [20:37] well the metal doesn't even have raid :) [20:37] depends how much you're storing, but restore time in the case of hard-disk failure could be quite long [20:38] Good points [20:38] so for anymthing critical you may want to have a copy locally [20:38] or be able to use arp's backup stfuf [20:38] and pull it back [20:38] Indeed [20:39] what about if youu got rid of 2 vps's, made the 3rd vps bigger, and got another hard-disk? [20:39] it's probably about the same cost [20:39] Perhaps, though we're pretty good on VPS sizing now (no need to upgrade) [20:39] I am looking at upgrades for metal (ssd, add'l drives, etc) [20:39] maybe you can ugprade to 16gb at the same time? [20:39] ahh ok [20:40] i thought you may want to be able to run all 3 vps's worth of stuff on one vps. [20:40] Would be nice to have 4*256GB SSD, but I can't justify doubling (actually, more than) the monthly cost [20:40] i don't thiink you can do more than 2 hard-drives [20:40] unless you get one of the 1u servers [20:40] https://www.arpnetworks.com/dedicated says "Max 4" [20:40] yeah [20:41] but i don't think you can upgrade if you're on a blade past 2. [20:41] that's for new orders.. [20:41] (One VPS is completely unused right now, so I only have to worry about 2 VPS, and they could easily be consolidated) [20:41] Hm, good to know, something to keep in mind and ask Garry later [20:41] i have 3 disks on arp metal [20:42] but it was provisioned that way from the start. [20:42] it's a bit weird. [20:42] I'm still not a fan of ARP's drive pricing. In a few months, I've paid enough to have bought the hdd in the first place :/ [20:42] i imagine that this may change soon [20:42] hard-disks use lots of power too [20:42] well about 7 watts [20:42] constantly [20:43] damn where's a quiuck calculator for that [20:43] looks like power costs around 22c kwh there [20:43] which seems cheap [20:43] SSD too, although it's a bit less egregious :p takes about 6mos to pay off one of those. [20:44] well cehaper than i thoughjt it'd be [20:44] it's actually the same everywhere you know :) [20:44] you make your money on the upgrades [20:44] you could go dedicated. [20:45] ok it looks like $1/month [20:45] so yeah power isn't that bad [20:45] but i forgot to margin in power supply loss [20:46] also hard-disks can fail, and that means someone replacing the hard-disk. [20:47] I've heard that SSD and hard disk are now similar in MTBF [20:48] hard-disk failure rates aren't really going down [20:49] the higher density seems to be puushing failure rates up a little even [20:49] and they seem to fail quicker when they fail now :( [20:50] like if you get a single sector not able to be read you should replace a drive asap [20:50] whereas it used to be that it'd work for a few months probably [20:50] I've seen the same - about equal. SSD can last a bit longer too. [20:50] a lot of that's to do with ecc though [20:50] as you only get to see errors when it goes above a certain point. [20:51] i've seen more "bugs" with ssd's. [20:51] but i've had more failures with hard-disks [20:51] like ssd's that'll just stop reading, or go really slow or such [20:52] Ditto. [20:55] one thing i've noticed is more and more of a shift to 2.5" hard-disks. [20:56] and 300gb sas drives vs 256gb ssd's, at 2.5" makes ssd's seem ilke quite a nice idea. [20:56] and laptop drives aren't really that great. [20:57] you can do about twice as many 2.5" drives as 3.5" drives in the same footprint. [20:57] and even 2.5" sas drives seem tiny [20:58] they weigh a lot more than ssd's though [21:58] http://www.anandtech.com/show/9145/samsung-sm951-pcie-ssd-now-available [21:58] the 128mb ssd is quite affordable. especially when you consider it does 2000mb/sec read speed.