Hi. How can I contact the Sales guys? dominicl: i would try sales@arpnetworks.com any Tech Support here? I need my account be upgraded vin\: send an e-mail to support@ are there any staff here at all? it seems arpnetworks is a one man team too..is that true? it is the topic says so even ah ok, so support is slow sometimes? indeed, also up_the_irons is in PST timezone so mind that hm, wondering whether to sign up or not i haven't had to contact support yet for any reason ;) ah right I contacted support once, to add more VPS's :) nice dominicl: IMO: if you are good with admin, the support thing isn't much of an issue. if you occasionally goof up and need "hands on" then.. ARP may not be the best choice for ya. just my opinion of course. hmm thats why its called unmanaged service pretty stable, 6:42AM up 647 days he's been gone awhile... dun dun DUUUN i'm still wondering when ntt link will come up jeez, i was netsplit for an entire day; had to reset weechat :P ohai up_the_irons welcome back! mike-burns is still in the vortex... tsk tsk he was the last remaining nick in the split channel oh there was an entirely separate channel created in the split? Pretty intense... usually the netsplit server will either rejoin or kick everyone Must've been pretty bad to have a split-brain come up certain servers were isolated from the network Freenode was getting DDoSed yesterday yep, but they usually figure out they're disconnected and boot everyone to force 'em back to a live server yeah the channel never rejoined Well you're back, and responding to tickets *nudge nudge* so that's all that matters :D brycec: i actually *am* replying to yours at the moment... yay! brycec: can u give your vps a hard reboot now? yep, gimme two shakes to pull up vnc... (vnc to enter the encryption key) k going down... powered off... booting... bios... neat grub... decrypted... booting... oooh eth1 haha it's very quiet on eth1... up_the_irons: how's the ntt circuit coming along? mercutio: in provisioning stage brycec: should be extremely quiet i think only 1 other person has a direct link brycec: i just sent the ticket reply Thanks up_the_irons ! fwiw NFS has generally been rock solid for me. I swear by it. brycec: np, does it work? :) brycec: which version of NFS? brycec: with linux? and on which platform? linux/freebsd/... most people swear at it. hahhhahahahah up_the_irons: nfsv3 (v4 was a nightmare..) on linux, freebsd, and osx stale nfs handles (though osx has some weirdness...) ok, i'll note the v3 About 479,000 results (0.24 seconds) i think the last i tried was v4 v4 has bells and whistles... but is really complicated compared to v3 i c that said most of my linux nfs problems were years ago And v4 wasn't supported by the centos 5.x install environment, so I had to roll bacjk and mostly to do with reboots.. i used to use nfs mounted / I still do my testbench is 100% netboot (because fuck drives) cool lol i used to do remote X too heh, X... what a b/w whore. thin clients kind of fell out of I recently used NX (NoMachine) and that is very, very slick i've lost the word fashion? use? yeh that'll do i just used it on lan I tried over DSL it's pretty bad. actually xterm is ok heh yeah... if it's a small window gtk applications are terrible gnome-terminal is probably terrible brycec: let me know if you have ping6 that link local address i gave ya up_the_irons: From fe80::5054:ff:fe2b:8374 icmp_seq=1 Destination unreachable: Address unreachable on that note, has anyone had much luck running X on windows? brycec: yep, i missed up.. another hard reboot plz :) boooo okay brycec: for the ping, you just qualify the v6 address with the interface, like: ping6 fe80::5054:ff:fe89:c122%vlan71 going down.... in that case, the int name is vlan71, but you would do like "%eth1" Yeah % should work too going up... I appreciate that the BIOS gives me a fighting chance to hit F12.... but sometimes it's annoying i set it for 15 seconds... i didn't want to make it too long, cuz then all boot times are higher heh up_the_irons: no-go still So just to be clear - this vlan is shared among other hosts (potentially),r ight? (ie. I should firewall it accordingly) brycec: yes, firewall it if need be brycec: what's your link-local side addr? 21:45:13 < brycec> up_the_irons: From fe80::5054:ff:fe2b:8374 icmp_seq=1 Destination unreachable: Address unreachable fe80::5054:ff:fe2b:8374 (and no, I don't have any firewalling active) (yet) OH i know ifconfig vlan71 up :p brycec: try now :) hey, you're pinging woo 64 bytes from fe80::5054:ff:fe2b:8374: icmp_seq=410 ttl=64 time=0.595 ms 64 bytes from fe80::5054:ff:fe2b:8374: icmp_seq=411 ttl=64 time=0.625 ms 64 bytes from fe80::5054:ff:fe2b:8374: icmp_seq=412 ttl=64 time=0.583 ms woo fast! I was about to complain that I was still seeing 100mbps speeds, but I decided to retry with a ramdisk as the destination instead damnit... cache sent 244559550 bytes received 31 bytes 12541516.97 bytes/sec still 100mbps (to a tee) fuck ignore me... lol i have our ISO volumes replicated over that link, so i know it is f'in fast :) heh sshfs is being a little bitch with the whole %eth1 bit hah hush you, mr "let's be progressive here" :p "If host is a numeric IPv6 address, it needs to be enclosed in square brackets." -.- rtfm ftw okay okay okay, real performance now... LOL copying from the backup server to a ramdisk, just over 100mbps sent 244559550 bytes received 31 bytes 13219436.81 bytes/sec yeah, you get used to the square brackets if you do enough ipv6. apache does the same thing. copying to the backup server, sub-100mbps. sent 244559555 bytes received 31 bytes 9228663.62 bytes/sec brycec: just do a straight scp of some huge file, you should see like 500 mbps (i haven't found a single stream go faster than that, it might be an mtu thing or w/e...) this might be a dumb question, but can I set the mtu to 9000 on the iface? i think u can i would be used to [], except all my ipv6 is config'd by slaac (or manually, which is sans-[]), using hostnames, or ssh (or ping6 or nmap) which don't mind it sans-[] yep, mtu set to 9000 and ping still works... avg speed still 100mbps. The first 111MB went at 40MB/s, then it stalled for 5 seconds, and it was never as quick seems like I starved your cache or something s/starved/exhausted/ (that was vps->backup) s/vps/vps-ramdisk/ backup->vps-ramdisk was much, much quicker (probably still in the server's cache), 40MB/s brycec: 40MB/s =~ 320 Mbps brycec: i just up'd the mtu on both sides on my end, btw brycec: so give it another try (oh yeah, both ends... heh) hmm something broke ssh is hanging mid-negotiation (they connect, but die during handshaking and such) hmm.. ping still works I've seen this before when the MTU was too high o'rly? packets started dropping over a certain size probably the mtu it's google'able http://www.snailbook.com/faq/mtu-mismatch.auto.html (tcpdump shows it constantly retrying) (feel free to abuse ssh and try logging it) hmm.. i wonder if the taps and br's also need to be changes damn lots of changes heh, could be brycec: give it another try, just to see if it doesn't stall (changed your tap mtu to 9000) hangin' brycec: one more time just went through yep, connects now hurray! oh hey nixbag , how's missouri? vps-ramdisk -> backup server 23.3MB/s (186mbps) up_the_irons: i'm back in LA now. i wonder if kvm networking is just hitting some cap nixbag: that was 2 weeks?? man time flies oh hell... up_the_irons: my ssh has frozen, and vnc "Faield to connect to server" (this happened as I started a backup->vps scp) brycec: hmm (and needless to say, no response to pings) brycec: your vm appears stopped i guess it crashed :o well I didn't stop it... i just started it for ya hmm.. musta died then thx (I hit Boot too) haven't seen that very much, but then again we're experimenting with new stuff... Let's try this again! ow. i think i have arthritis :/ up_the_irons: we cut the trip short because of incoming snow storms. up_the_irons: I'm seeing 40MB/s both directions pretty consistently :D that is nice and fast brycec: oh sweet brycec: i need to find a way to get those MTUs persistent on your tap / br cuz when you reboot the interface is created a-new openbsd is fun. oh goody... "Just don't reboot" brycec: i made an adjustment, can u plz retest? brycec: pfff.. weak sauce $ ssh kvr06 uptime 21:56:59 up 1154 days, 18:50, 1 user, load average: 1.16, 1.26, 1.83 $ i hope it doesn't now crash tonight... up_the_irons: well I'm sorry I only recently signed up :P you jinxed it. brycec: lol backup->vps(ramdisk) 46.6MB/s, vps(ramdisk)->backup 38.9MB/s so still good? I'm happy cool, appears only the tap needs the 9000 mtu no need to mess with the bridge interfaces Do I see an option in the portal to configure the mtu in the future? :D (or just set everyone to 9000... or just do it by hand/on request) i think it's simpler to just try 9000 everywhere if it's stable, then it's all good might cause confusion with fresh installs that attempt 1500 by default (don't get me wrong, i'd love 9000 everywhere) well, if the tap is 9000 on the host, but only 1500 on the VM side, that shouldn't mess with anything the mtu is 9216 on all my "backplace" traffic ports, but the front-end NICs are still 1500 (cuz i'm lazy and get enough speed anyway) brycec: can u test one more time? (sorry, didn't think we'd get into all this mtu stuff now :) up_the_irons: still good backup->vps(ramdisk) 46.6MB/s, vps(ramdisk)->backup 38.9MB/s (again) brycec: ok nice seems like the vlan on the backup server doesn't need the 9000 mtu either. like, only "physical" interfaces gonna reboot the backup server to make sure the mtu sticks noooo woops sshfs does NOT like yuo right now, up_the_irons in my testing and "real" use: (rx/tx) max 391.09 Mbit/s | 339.03 Mbit/s looks like we're back up mtu stuck, yay great btw, i added the backup space to your Portal profile, officially you'll see more options now options where? in portal oh cool! click "Backup Space" i'll spoil the surprise: just let's you submit new ssh keys I love that df tells me I can write out another 2.7TB... in a 20GB quota :p *2.5 **2.6 up_the_irons: that's not all - also mentions the hostname brycec: is this over sshfs? even though the hostname doesn't match the ipv4 address you gave out up_the_irons: Yes cuz 'df' is not a binary within the chroot, so you can't see it while logged in lol sorry for the confusion yes, df against the sshfs mountpoint brycec: interesting. dunno if there's anything i can do about that probably not unless you can fuck with sftp on the server yeah true up_the_irons: Could I ask for a bump in my quota? interesting, the openbsd implementation of virtio networking can only do a mtu of 2024 not 9000 toddf_: whoa weird