[00:28]  * RoyK installed his home server on broken mirrors and just resynced them on a newly added drive - nice
[00:29] <RoyK> now, how do I install grub on the new drive?
[00:45] <RoyK> there are issues in all sorts of software, but it's only in the lands of Oracle and Microsoft where you need a handful of gods or a trillion dollars to fix them.....
[00:46]  * genii-around shakes his fist in the general direction of Oracle
[00:47]  * RoyK shakes his fist in the direction of his home server where he just added a new drive to mirror up the root and now it's noisy, the little bugger - damn it's only got like six disks......
[00:48] <twb> RoyK: go put it in the 19" rack in the basement then
[00:49] <RoyK> wish I had one
[00:51] <genii-around> I have like 3 empty server racks, 2 19" one 21" which are rusting away here in my basement
[00:52] <RoyK> 21"?
[00:52] <RoyK> never heard of those...
[00:54] <RoyK> what do you use those for? 19" equipment with sufficient amount of gaffer tape? ;)
[00:54] <genii-around> RoyK: They used to house Rogers Cable equipment, I sort of inherited them.
[00:55] <twb> It is the way of things
[00:58] <genii-around> These particular cabinets housed the battery backup system, 32 huge lead acid buggers
[00:59] <RoyK> aren't lead batteries the most used even today in UPSes?
[02:09] <qman__> I don't have a rack either, my stuff is hanging in the floor joyces by various means
[02:19] <twb> RoyK: yes, because they're stable
[02:22] <Roasted> anybody ever see Samba do this before? failed negprot: NT_STATUS_IO_TIMEOUT
[02:26] <twb> Roasted: no,
[02:26] <twb> Roasted: maybe strace it or something
[02:27] <Roasted> twb, strace?
[02:29] <twb> If you don't know what that means, I can't help you
[02:31] <Roasted> sweet
[02:31] <Roasted> great attitudes here :D
[02:32] <Roasted> anybody else ever see that error? I saw it when i ran smbtree. For some reason this laptop was unable to connect to my samba share when it had done so previously the day before without a hitch.
[02:32] <twb> I am not representative of normal IRC denizens.
[02:32] <Roasted> that's good. we'd be in trouble if you were.
[02:32] <Roasted> thanks anyway, I'll keep asking around.
[02:32] <twb> Roasted: test with smbclient on localhost first
[02:32] <twb> Otherwise problems could be unrelated to smb, like your cat5 cable is buggered
[02:32] <Roasted> wireless
[02:33] <twb> Whatever
[03:57] <twb> Heh, I had a freakout because my laptop had ifupdown 0.7
[03:57] <twb> But it's ifupdown 0.7~really0.6...
[06:29] <Takyoji> Is there a way to write a script that runs when an interface is up?
[06:29] <twb> Takyoji: /etc/network/if-up.d
[06:29] <Takyoji> /etc/network/if-up.d/ or?
[06:30] <Takyoji> ahh alright
[06:30] <twb> Takyoji: what's your use case?
[06:30] <Takyoji> Mounting an NFS share
[06:30] <twb> Because if it's firewall, iptables-persistent package is better
[06:30] <twb> Or ufw, I guess
[06:30] <Takyoji> regarding the issue of fstab trying to mount the NFS share before the network is even accessible
[06:32] <twb> Oh that shit
[06:32] <twb> One moment
[06:32] <twb> http://paste.debian.net/139962/
[06:33] <twb> That last stanza
[06:33] <twb> At at 10.04, I encountered a cyclic dependency in NFS mounting at boot, and that was the only way to fix it short of replacing mountall and upstart with stuff that actually worked
[06:34] <twb> And yes, that really does tell mountall to try to finish mount -a every tenth of a second FOREVER
[06:40] <Demosthenes> anyone else have an issue with initial shell logins taking up to 5 minutes to occur? i suspect its the fancy welcome message (disk status, users, packages, etc)
[06:41] <twb> Demosthenes: immediately after a reboot?  Yes, remove byobu
[06:41] <twb> The package responsibel is actually called something like unattended-upgrades-common and has Section: gnome
[06:41] <twb> If it continues to happen indefinitely, it's a fuckup in your nsswitch.conf settings
[06:42] <twb> (e.g. the ldap server is unreachable)
[06:42] <Demosthenes> twb: interesting.
[06:43] <Demosthenes> generally after reboot and first login per user.
[06:43] <twb> Yep
[06:43] <Demosthenes> sometimes even times out and you hve to try again
[06:43] <twb> Then it's the former
[06:43] <Demosthenes> thats as annoying as that "command suggestion" thing
[06:43] <twb> You can stop c-n-f on a per-user basis btw
[06:43] <twb> http://cyber.com.au/~twb/.shrc
[06:43] <twb> line 24
[06:44] <Demosthenes> oh no, whole box. ;]
[06:44] <twb> in that case just uninstall c-n-f
[07:04] <c0nv1ct> menu in ubuntu
[07:04] <c0nv1ct> MT
[07:05] <c0nv1ct> http://i.imgur.com/1YbXp.jpg <-- meant to paste this... any idea what is causing it? showed up after recent update
[07:05] <twb> c0nv1ct: are those characters unreadable, or is that an artifact of your photo?
[07:06] <c0nv1ct> twb: that is exactly how my tty looks at boot
[07:06] <c0nv1ct> no readable text since grub complaining about fd0
[07:06] <twb> Does hitting Alt-<left> a dozen times fix it?
[07:06] <c0nv1ct> every tty is the same garbled font
[07:07] <twb> OK, then what I would do is go in with a live CD and blacklist the framebuffer, reconfigure console-setup (and rebuild the ramdisk), or disable plymouth
[07:07] <twb> Probably all of the above
[07:07] <c0nv1ct> i can ssh in fine, framebuffer seems like the culprit since there is some blankness during boot that would imply that
[07:08] <twb> Oh, I'm assuming you're on x86 and are OK with ASCII -- if not then that's more effort
[07:08] <twb> Like if you need to see greek on the tty or something
[07:08] <twb> Yeah the problem is clearly something funky fb-wise
[07:08] <twb> I haven't seen those specific symptoms before
[07:10] <c0nv1ct> dmesg says it is using efifb... pretty sure this old AMD doesnt use EFI
[07:11] <twb> so blacklist it
[07:35] <Demosthenes> twb: it was landscape!
[07:35] <twb> Sigh
[07:35] <Demosthenes> removed landscape-common, and no more lag
[07:35] <Demosthenes> at login anyway
[07:36] <twb> That install option for landscape should say "DO NOT pick this unless you gave Canonical your VISA card" or something
[07:36] <twb> Since AFAIK it's pointless and harmful unless you actually have a support contract or whatever
[07:39] <riot_le> hello, i am searching for Help by setting the correct Group-Permissions on a Path, maybe anyone who can help?
[07:40] <Demosthenes> twb: yeah, seemed advertisy, but i can't begrudge them a little branding.
[07:41] <twb> Demosthenes: more just that I get called in to clean up when users try to be their own admins, and sometimes they pick that because it sounded neat
[07:41] <Demosthenes> twb: just frustrating that i read the user profiles, and /etc/bash.bashrc, etc, and none of it clued me in. wasn't til i found the MOTD was where they're written, and somehow autoupdated.
[07:41] <twb> Which sounds like what happened to you
[07:41] <Demosthenes> i don't recall asking for it
[07:41] <twb> Demosthenes: yeah, update-motd and friends.
[07:41] <Takyoji> riot_le: changing what group a file/folder is owned by, or read/write/execute for a group on a file?
[07:42] <Demosthenes> i tend to be a package minimalist
[07:42] <Demosthenes> and i'm not interested in their service
[07:42] <twb> Demosthenes: the other one that messes with me, is you get an MTA by default *iff* you configure software raid.
[07:42] <Takyoji> riot_le: If you're changing the group, the syntax is: chgrp (group name) (file name)
[07:42] <Takyoji> if it's a folder and you want it done recursively, add -R to the command
[07:42] <twb> Takyoji: -Rh if you're paranoid
[07:43] <Demosthenes> twb: wild ;]
[07:43] <twb> Demosthenes: it's because mdadm has Recommends m-t-a so it can send panic mail
[07:44] <riot_le> @Takyoji: I added a User to a System and gave him the Path as Homepath where he can write. When User X add a File to the Path the File has the Attributes Owner X Group X
[07:44] <Demosthenes> twb: wise mdadm
[07:44] <riot_le> but i want that the Group is still y (like any other Files in Folder)
[07:44] <Demosthenes> twb: of which i appreciate. i'm running raid1 and raid6, and already tested that ;]
[07:44] <twb> riot_le: you might want setgid on the dir
[07:44] <Demosthenes> raid1 across a pair of 16GB usb sticks for booting the OS ;]
[07:45] <twb> Demosthenes: yes, but it's not logical that you get (an unconfigured) MTA iff you configure software raid
[07:45] <twb> Demosthenes: it violates principle of least surprise
[07:45] <Demosthenes> yeah, unconfigured is a pita
[07:45] <Demosthenes> oh thank god someone else knows that principle
[07:45] <Takyoji> Also, twb, I managed to write a script that works upon the interface going up, for it to wait 3 seconds, and kill mountall.
[07:45] <twb> OK, to be fair, I *think* during install it prompts you to configure postfix
[07:45] <Takyoji> so yay, it works
[07:45] <Takyoji> Yes, it does install postfix as a requirement
[07:46] <Takyoji> Also, can you check for faulty drives via /proc/mdstat? :P
[07:46] <twb> Takyoji: you'd just better hope that 3s stays correct everywhere you put it
[07:46] <Takyoji> I know. :P
[07:46] <Demosthenes> Takyoji: mdadm monitors and emails...
[07:46] <twb> Takyoji: e.g. if you also deploy that config on an embedded box and an s/390 or something
[07:46] <Takyoji> I don't have the emailing function configured at all
[07:46] <riot_le> @twb how to setgid?
[07:47] <twb> Takyoji: btw, if you think of a better solution (that works for LTS), do let me know
[07:47] <Takyoji> alright; I'll keep it in mind.
[07:47] <twb> riot_le: uh, it's dangerous to tell you without the full explanation of how POSIX DACs work, and I don't have time
[07:47] <jamespage> morning all
[08:19] <lynxman> morning o/
[08:20] <BuenGenio> hello
[08:20] <BuenGenio> how do I check the current power consumption?
[08:20] <BuenGenio> in Watts
[08:22] <twb> BuenGenio: plug in a power meter between mains and your server
[08:22] <BuenGenio> twb, really?
[08:23] <twb> Really.
[08:23] <BuenGenio> you know you can just use the post office to send mail?
[08:23] <BuenGenio> plus receiving post cards is much nicer
[08:23] <BuenGenio> why don't you do that?
[08:23] <twb> Because I have no friends to write.
[08:23] <BuenGenio> thought so
[08:24] <twb> If you expect me to build a better solution, then go back in time and deploy it in time to get into the release of Ubuntu you're running, then you're shit outta luck.
[08:25] <BuenGenio> if you're talking to me, I wasn't expecting /you/ to do anything. not to give smarty-ass answers, anyway.
[08:25] <twb> BuenGenio: fair enough.
[08:26] <BuenGenio> I don't know who you are and what ou do, but - shoving a power meter in my rack is not the answer I was looking for.
[08:26] <twb> it's what I did last week
[08:26] <BuenGenio> not sure if it's the sentiment I share with your (would-be) friends...
[08:26] <BuenGenio> hah!
[08:27] <BuenGenio> maybe I'm a bit stressed (and missing the obvious).
[08:27] <BuenGenio> I don't know
[08:27] <twb> Maybe if you have name-brand Sun gear or something, there is a better way, but AFAIK not for fungible whitebox crap
[08:27] <BuenGenio> maybe I need to see a shronk
[08:27] <twb> Or obviously if you have a decent UPS already, you can ask it
[08:27] <c0nv1ct> crap, traded the terminal font problem for a networking problem... the network service is stuck on stop/wait
[08:27] <BuenGenio> I have PowerBar installed on Windows - that tells me exactly the watts I'm using
[08:27] <BuenGenio> thought Linux would have figured that out long before that
[08:28] <twb> Shrug.  Maybe you can ask something in /sys/, but I'm not aware of it
[08:28] <c0nv1ct> i'm late to this convo, but powertop doesnt work for you?
[08:28] <twb> c0nv1ct: that shows IRQ wakeups and stuff, not an overall wattage number IIRC
[08:29] <twb> Current debian sid version doesn't tell me watts on my atom-based netbook, fwiw
[08:30] <twb> c0nv1ct: what did you do that borked networking?
[08:31] <c0nv1ct> not sure really, the only change i made was to add a dns record to interfaces, but i removed that line and it still doesnt work
[08:31] <twb> c0nv1ct: you didn't turn plymouth off or blacklist fb or anything?
[08:31] <twb> That symptom sounds like you didn't turn plymouth off hard enough or so
[08:32] <c0nv1ct> to fix the console font i set GRUB_TERMINAL=console in defaults/grub and did update-grub2
[08:32] <twb> That shouldn't cause that problem
[08:32] <c0nv1ct> i am seeing plymouth errors though, maybe i need to turn that off harder
[08:32] <c0nv1ct> whats the hard way to disable plymouth all together?
[08:32] <twb> Maybe post-lucid, plymouth relies on grub to set up the framebuffer?
[08:33] <twb> c0nv1ct: by "harder" I meant, like, with extreme prejudice
[08:33] <twb> Ideally you just uninstall it, but as at lucid a Depends screw up makes that impossible
[08:33] <ersi> BuenGenio: There's no totally reliable way to determin power usage from software. The closest you get, is by polling ACPI/A battery/UPS. If you want RELIABLE power metering, you should really follow twb's advice of getting a physical power meter.
[08:34] <c0nv1ct> ugh, why is ubuntu-server messing with framebuffers and crap, i'll just uninstall it and see
[08:34] <twb> Good luck
[08:34] <BuenGenio> thx
[08:34] <twb> BuenGenio: that was actually at c0nv1ct :-)
[08:34] <BuenGenio> It's just that my data center is 10,000km away
[08:34] <twb> BuenGenio: I feel you pain, man
[08:34] <BuenGenio> it's more curiosity really...
[08:34] <twb> BuenGenio: can you get a colo monkey to put one in for you?
[08:35] <twb> If it's in a datacenter they might even have that already
[08:35] <c0nv1ct> yeah, cant remove plymouth... wow
[08:35] <lynxman> normally datacenters measure consumption per power line
[08:36] <twb> c0nv1ct: ok, the two main things I've done is to dpkg-divert all the bits of plymouth from /usr/share/initramfs-tools and /etc/init, and to add blacklists for more framebuffers in /etc/modprobe.d/ -- both will require update-initramfs -u -k all, and note that if you cock it up you'll need a live CD or so to get back in to fix it.
[08:36] <ersi> BuenGenio: I'd hear with your hosting provider, if they possibly could hook you up with some monitoring. But I also, feel your pain about this subject
[08:37] <twb> lynxman: power line as in per port in the rack's power rail (thus, per device in the rack), or power line as in port in the wall
[08:37] <lynxman> twb: power line as power strip on the rack
[08:38] <twb> Right
[08:38] <lynxman> twb: that's the normal behaviour, so if you have N servers on that power strip you can guesstimate consumption
[08:38] <ersi> Guesstimation <3
[08:39] <twb> I ask because I don't wear enough ties to babysit the kind of customers that have colo hosting
[08:39] <lynxman> twb: If you have some advanced hardware the PSU will also tell you the power consumption, but I've always found those to be innacurate
[08:41] <lynxman> twb: the best solution that I've found for colo is to buy one of those APC power control strips, the modern ones have also consumption per plug
[08:44] <Rolpa> hello?
[08:45] <ersi> Hi.
[08:47] <Rolpa> I need a little help
[08:49] <archayl> i have ldap and unix user in duplicate. how to delete user by userid, not by name using terminal, i intend to keep the ldap user.
[08:52] <c0nv1ct> archayl: you are trying to remove users in bulk from a list of UIDs?
[08:54] <archayl> c0nv1ct: nope, just around 5.
[08:55] <c0nv1ct> archayl: ah, then nothing fancy needed, just use /etc/passwd to find the username
[08:56] <Omega42> hi... are there known problems with ubuntu 11.10 and libvirt hook scripts?
[08:58] <archayl> c0nv1ct: it's a duplicate username. i have 'john' in both ldap and system with different uid. it represents the same user. i want to retain the one in ldap.
[08:59] <xranby> Omega42: possibly, do any of these bugs match your issue? https://bugs.launchpad.net/ubuntu/+source/libvirt
[09:00] <c0nv1ct> archayl: then remove the user with userdel?
[09:01] <ersi> Rolpa: Ask whatever your wondering (that is related to ubuntu on serers) - and you might get an answer. No guarantees though ;) Don't ask to ask, ask instead
[09:01] <Rolpa> oh sorry lol
[09:01] <Rolpa> was afk
[09:01] <Rolpa> yea
[09:01] <Rolpa> Im having issues with vsftpd
[09:02] <Rolpa> I have it installed on my server installation
[09:03] <Rolpa> but when I try to FTP in with my client. it establishes a connection to the server but says my login is incorrect
[09:04] <archayl> c0nv1ct: does userdel delete the ldap user too?
[09:04] <pmorris> Hi, for some reason my server is unable to resolve hostnames since last night and I cannot understand how or why. Besides doing regular apt-get updates nothing has changed. I can "fix" the problem by running dhclient but every time I restart networking I have to do this again
[09:05] <c0nv1ct> archayl: no
[09:05] <Rolpa> Ive checked my config repeatedly and I dont see an issue
[09:07] <Omega42> archayl: i didn't find my problem ther
[09:07] <Omega42> the problems, i have problems using hook scripts:
[09:07] <Omega42> qemu hook script: i get the error: "internal error Child process (LC_ALL=C PATH=/usr/local/sbin:/usr/local/bin:/usr/bin:/usr/sbin:/sbin:/bin /etc/libvirt/hooks/qemu VMxy prepare begin -) status unexpected: exit status 1" and my VMs won't satrt, The script is a correct bash script and exits 0
[09:07] <Omega42> daemon hook script: that's is totally ignored
[09:07] <Omega42> versions: libvir 0.9.2 / API: QEMU 0.9.2 / Hypervisor: QEMU 0.14.1 (KVM) / ubuntu 11.10
[09:07] <Omega42> maybe it's a poblem caused by release-upgrades, i alwaysw keep old config files - i'll try with a clean install
[09:08] <ersi> Rolpa: What kind of configuration do you have? Are you allowing local_enable = YES in your config? Or are you using another authentication method for your users?
[09:09] <Rolpa> local_enable is enabled
[09:13] <Rolpa> yea
[09:13] <Rolpa> its a default setup
[09:13] <Rolpa> Status:	Connecting to 192.168.1.6:21... Status:	Connection established, waiting for welcome message... Response:	220 (vsFTPd 2.3.2) Command:	USER stefano Response:	331 Please specify the password. Command:	PASS ********* Response:	530 Login incorrect. Error:	Critical error Error:	Could not connect to server Status:	Connecting to 192.168.1.6:21... Status:	Connection established, waiting for welcome message... Response:	220 (vsFT
[09:13] <Rolpa> thats what I get when I attempt to connect
[09:26] <archayl> c0nv1ct: thank you very much ;)
[09:38] <Rolpa> hello?
[10:17] <Daviey> Anyone keen to land a patch for SRU?
[10:24] <pmorris> Can someone who understands networking shed any light on why I have to run dhclient but every time I boot or restart networking? Since last night I've been unable to resolve hostnames and since it fails at boot time all my services including SSH fail also
[10:26] <Daviey> jamespage: Does it make sense to do an update fakesync of geronimo-jms-1.1-spec?
[10:29] <toobler> Can someone give me a link to configure an SMTP in Ubuntu, it will be very helpful for me. Thanks
[10:30] <w00> toobler, https://help.ubuntu.com/community/Postfix ?!
[10:30] <toobler> Thanks
[10:34] <jamespage> Daviey: looking now
[10:37] <jamespage> Daviey: yes
[10:39] <Daviey> cool
[10:39] <Daviey> zul: I see you last merged ntp, do you want to do so again?
[10:40] <zul> sure
[10:40] <Daviey> great
[10:42] <Daviey> hallyn: Do you think we should sync open-vm-tools?
[10:42] <Daviey> zul: Were you handling lynxman's puppet merge?
[10:43] <zul> lynxman: yeah i was told to wait for a newer puppet release though
[10:43] <lynxman> zul: There's a CVE for 2.7.6 coming through
[10:43] <Daviey> zul / lynxman: 2.7.6-1 has been in sid since the weekend
[10:43] <lynxman> zul: thought we could make both? :)
[10:44] <Daviey> lynxman: ah
[10:44] <Daviey> Wasn't thatfixed yesterday?
[10:44] <lynxman> Daviey: mdeslaur was on it for the releases, not for precise though
[10:44] <zul> mdeslaur	zul: fyi don't merge puppet 2.7.5, we need 2.7.6 to fix this: http://puppetlabs.com/security/cve/cve-2011-3872/
[10:45] <Daviey> lynxman: it seems to be fixed in 2.7.6.. no?
[10:45] <lynxman> Daviey: it is
[10:45] <lynxman> Daviey: that's what 2.7.6 is for :)
[10:45] <Daviey> lynxman: so what is blocking?
[10:45] <zul> testing doesnt have it yet
[10:45] <lynxman> Daviey: so we can try and merge 2.7.6 straight
[10:45] <lynxman> Daviey: just my mind I guess :D
[10:45] <zul> ill do the puppet merge as well
[10:46] <lynxman> zul: as said, if I can help I'll be glad to :)
[10:46] <lynxman> zul: as soon as I finish this whitepaper (today)
[10:46] <jamespage> Daviey: want me to take care of that?
[10:46] <Daviey> zul: We can still merge/sync from sid
[10:46] <Daviey> jamespage: wassat?
[10:47] <zul> can ill take a crack at it then
[10:47] <jamespage> geronimo-jms-1.1-spec
[10:47] <Daviey> jamespage: gets it off the list :)
[10:47] <jamespage> Daviey: coolio
[10:48] <Daviey> "Add build-dependency on python-support" Thanks Debian... i won't be merging that.
[10:48] <lynxman> Daviey: lol
[10:49] <Daviey> hallyn: qemu-kvm, 0.15.1+dfsg-1 - seems all good fixes.. should we grab that?
[10:52] <Daviey> whois looks like a nice easy merge, anyone want to grab it?
[10:53] <Daviey> koolhead17: ^^ ? :)
[10:55] <koolhead17> Daviey: hellos
[10:57] <koolhead17> Daviey: show me the way O Master!! :)
[10:59] <koolhead17> Daviey: last night result after removing the sqlite.so http://paste.ubuntu.com/719588/ seems issue still persists
[11:00] <Daviey> koolhead17: hmm, can i see a debdiff?
[11:01] <koolhead17> ok 2 mins
[11:01] <Daviey> debdiff *5.3.8-2ubuntu1.dsc *5.3.8-2ubuntu2.dsc | pastebinit
[11:01] <koolhead17> hmm
[11:02]  * koolhead17 starts his virtualbox instance
[11:07] <koolhead17> Daviey: http://pastebin.com/8WP5guGB
[11:11] <Daviey> koolhead17: Hmm, perhaps that isn't what creates the conf then.
[11:14] <koolhead17> Daviey: there is one  more file name modulelist
[11:14] <koolhead17> it has entry of sqlite insted sqlite3
[11:15] <koolhead17> Daviey: http://pastebin.com/9zHq9tbz
[11:15] <Daviey> koolhead17: hmm, might be worth trying that
[11:16] <koolhead17> Daviey: shall i replace it with sqllite3 or remove the whole entry of sqllit from the file
[11:16] <koolhead17> ?
[11:16] <koolhead17> :)
[11:16] <Daviey> koolhead17: 'try it' :)
[11:17] <koolhead17> ok am removing sqlite entry from there
[11:19] <koolhead17> Daviey: one more question i will remove that part and run dch -i  i should remove "precise-i386_result" directory ?
[11:20] <Daviey> koolhead17: no need to, but can do
[11:20] <koolhead17> ok removing then running :)
[11:20] <Daviey> it'll overwrite if it is the same version
[11:20] <koolhead17> but i think it should create new version 3 as 2 is what will get upgraded :)
[11:21] <Daviey> koolhead17: you can overwitre the package :)
[11:21] <Daviey> ubuntu2 is what will get uploaded.
[11:22] <koolhead17> ok. so am doing notthing. as you said it will be updated.
[11:22] <koolhead17> another 3 hrs wait
[11:25] <lynxman> koolhead17: \o/ your first merge :)
[11:25] <koolhead17> lynxman: hola!! Yeah once the issue gets fixed :D
[12:30] <andybalong> anybody can help me please ?
[12:31] <rbasak> !ask
[12:31] <andybalong> how to open port game online in squid ?
[12:33] <andybalong> how to set it inside squid
[12:36] <cwillu_at_work> andybalong, I'm not sure that squid will do that, but I could be mistaken
[12:38] <cwillu_at_work> and my quick glance at the docs and faq seems to confirm that
[12:39] <cwillu_at_work> andybalong, squid is a proxy for specific protocols, (i.e., you can think of it as acting like both a web server and client); what you're asking for is more a dumb "shovel everything from this port to this address"
[12:40] <cwillu_at_work> and of course, having said that, I find a page that seems to describe what you want, but it also seems new and kinda experimental
[12:40] <cwillu_at_work> but even then, I'm not sure :p
[12:40] <cwillu_at_work> http://wiki.squid-cache.org/ConfigExamples/MultiplePortsWithWccp2
[12:41] <cwillu_at_work> (it describes the configuration to set up a normal port forward through the linux kernel's network layer, so I don't think this is a squid thing specifically)
[12:42] <andybalong> ok, thank you i'll try
[12:44] <smoser> hallyn, around ?
[12:44] <smoser> libvirt lxc issue of the day.
[12:46] <zul> oooh...this shouldnt be pretty :)
[12:52] <smoser> comon hallyn, stop hiding. i prefixed 'lxc' with libvirt so you wouldn't complain about that.
[12:52] <hallyn> smoser, oh, hey
[12:53] <zul> morning btw
[12:53] <smoser> libvirt lxc dies for me when i try to run a cirrOS container.
[12:54] <hallyn> what is cirrOS
[12:54] <smoser> https://launchpad.net/cirros/+download get the lxc.tar.gz and then try to run it (i try with using my lxc-libvirt-root)
[12:54] <smoser> http://smoser.brickies.net/git/?p=tildabin.git;a=blob_plain;f=lxc-libvirt-root;hb=HEAD
[12:54] <smoser> it dies with can't mount /dev/pts
[12:54] <hallyn> oh right
[12:54] <smoser> hallyn, cirros is my little toy
[12:54] <hallyn> right :)
[12:54] <hallyn> zul, good morning :)
[12:55] <smoser> hallyn, is there some way i can test lxc proper ?
[12:55] <smoser> run-a-container-in-this-root ./rootd
[12:55] <smoser> thats what i want
[12:56] <hallyn> i don't know of a such a script.  You could lxc-create a basic container, then swap out the rootfs
[12:56] <hallyn> or i can whip up such a script
[12:56] <hallyn> it hasn't been useful in the past bc rootfs needed tweaking by templates anyway
[12:57] <smoser> well, hte lxc-create path sure is attractive, doing hundreds of megabytes of disk io and network io in order to create a 15 line config file :)
[12:57] <hallyn> glad you like it
[12:58] <hallyn> hold on i'm working on a config
[13:00] <hallyn> sigh, cloud-init hanging
[13:01] <smoser> ;)
[13:01] <hallyn> (for some raason my containers aren't getting a proper resolv.conf from libvirt's dnsmasq)
[13:01] <hallyn> 20/30, almost there
[13:03] <hallyn> smoser, well, http://people.canonical.com/~serge/cirroscfg has the config and fstab files you can use
[13:03] <hallyn> it's hanging for me on the fancy ascii-art logo
[13:04] <zul> is this on oneiric or pangolin
[13:04] <hallyn> smoser, wget: can't connect to remote host (169.254.169.254): No route to host
[13:04] <hallyn> , that host isn't up anyway
[13:05] <hallyn> zul, oneiric, though with pangolin's libvirt
[13:05] <zul> hallyn: ah
[13:05] <hallyn> actually its resolv.conf is fine
[13:05] <hallyn> smoser's hardcoded host is just bogus :)
[13:06] <smoser> hallyn, typo? /var/lib/lxc/o1/fstab
[13:06] <smoser> hard coded path ?
[13:07] <hallyn> oops, yeah
[13:07] <hallyn> is that in notes?
[13:07] <hallyn> obviously we can make this into a trivial script or template
[13:07] <hallyn> which reminds me,
[13:07] <hallyn> SpamapS, were you ready to get your euca tarball based lxc template into the lxc package?  :)
[13:08] <smoser> hallyn, do i need an fstab ?
[13:09] <hallyn> smoser, if you want proc mounted, i think you do
[13:09] <hallyn> well, no,
[13:09] <hallyn> you can put those entries straight into config is uppose
[13:10] <smoser> into the config ?
[13:10] <smoser> init mounts /proc
[13:10] <smoser> and /sys
[13:10] <smoser> (as it should)
[13:14] <hallyn> smoser, then you don't need it
[13:14] <hallyn> so, does it work?
[13:15] <smoser> not tried yet.
[13:17] <SpamapS> hallyn: it was pretty simplistic... I wonder if we could just make it the "OVF" template or something.
[13:17] <hallyn> yuck
[13:17] <hallyn> no reason we couldn't though
[13:17] <hallyn> maybe one night next week :)
[13:17] <hallyn> write it over a pitcher?
[13:18] <smoser> SpamapS, well the ovf references the full disk image
[13:18] <smoser> and that is not really suitable for lxc consumption. just due to the partition table.
[13:20] <SpamapS> smoser: we could loop mount and copy the stuff out of it
[13:20] <smoser> well thats just silly.
[13:20] <zul> and a waste
[13:20] <smoser> theres a partition image that what you want
[13:20] <smoser> and libvirt newer versions support using the partition image
[13:21] <SpamapS> Just thinking in terms of integrating a standard
[13:21] <zul> right that needs to be tested as well again
[13:21] <smoser> SpamapS, doing it "right" that way would really require supporting a disk that was partitioned with multiple partitions, reading itfs fstab, mounting
[13:21] <smoser> or...
[13:22] <smoser> actually, my opininion the better way to do all this is to make a loader that you run, that essentially takes a device node, mounts a root partition in it read-only, and calls /sbin/init
[13:23] <smoser> and get lxc out of the garbage of thinking it knows anything.
[13:23] <smoser> things like mounting proc and sys are just silly
[13:26] <smoser> hallyn, ok. so i booted with your config
[13:27] <lynxman> zul: I'm all yours, want me to merge 2.7.6 from sid?
[13:27] <zul> lynxman: sure...i just started to look at it, but it looks like it can be synched
[13:28] <lynxman> zul: okay, having a lookie then :)
[13:31] <smoser> hallyn, so where should i put a getty under this lxc-start world?
[13:32] <hallyn> on tty1
[13:32] <hallyn> (or tty2..tty4)
[13:32] <smoser> there is no entry in /dev for tty1
[13:32] <smoser> just try anyway?
[13:32] <smoser> in libvirt i get an entry there i think
[13:35] <smoser> if i just try i get 'can't open /dev/tty'
[13:35] <smoser> er.. /dev/tty1
[13:40] <hallyn> smoser, create the device nodes inthe rootfs
[13:40] <smoser> hallyn, ok. so i mknod /dev/tty1 and then try to run a getty on it and i just dont see anything
[13:40] <smoser> writing (echo "hello" > /dev/tty1) also does not get anything written to screen.
[13:41] <hallyn> then do lxc-console -n cirros
[13:41] <smoser> ah. there we go!
[13:41] <hallyn> the lxc-start console is probably tty0?
[13:41] <smoser> :)
[13:41] <hallyn> so actually if you jsut create tty0 you might get console on lxc-start
[13:41] <smoser> whats the escape key ?
[13:44] <jeh> When using a seed file does anyone know how to get past  "do you want to activate serial raid devices"
[13:45] <smoser> i see no affect of /dev/tty0
[13:45] <pmatulis> jeh: you probably don't want to have that queston asked at all (whether you see it or not)
[13:46] <pmatulis> jeh: it sounds like you have fake raid on your m/b
[13:46] <pmatulis> jeh: do you actually use it?
[13:47] <hallyn> smoser, dev/console?
[13:50] <smoser>  /dev/console seems ot have an affect
[13:51] <smoser> hallyn, how do i detach from an lxc-console
[13:51] <hallyn> ctrl-a q
[13:52] <hallyn> do you not launch a getty on /dev/console?
[13:53] <smoser> i do not.
[13:53] <smoser> most things do not
[13:53] <smoser> they launch something on tty1
[13:53] <smoser> or ttyS0
[13:54] <smoser> hallyn, you have any idea what is going wrong in libvirt lxc ?
[13:55] <hallyn> no
[13:55] <hallyn> what do i do with that script you gave me
[13:55] <smoser> did you try it? it dies with the pts issue.
[13:55] <smoser> ./libvirt-lxc-root ./rootdir
[13:58] <hallyn> odd, that goes through the cloud-init loop much slower than lxc does
[13:58] <smoser> you get to that loop ?
[13:58] <smoser> it goes through slower because it has a network.
[13:58] <hallyn> smoser, yes, adding a consoel entry to inittab gives a console lxc-start window, fwiw.  but presumably you'll want to do 'lxc-start -d -n cirros' anyway, and not use that
[13:58] <smoser> under lxc the dhcp is not getting a ip address
[13:58] <hallyn> yes i get that loop in libvirt
[13:59] <hallyn> and, got a prompt
[13:59] <smoser> i hate you
[13:59] <hallyn> now, i'm running 0.9.6
[13:59] <hallyn> not sure if something else has changed
[13:59] <hallyn> smoser, i assume you extracted the rootfs as root?
[14:00] <smoser> hallyn, no. i used the -lxc.tar.gz
[14:00] <smoser> what did you use ?
[14:08] <jeh> pmatuli: correct, it's on the MB, even though it's disabled in the bios it still prompts me for this
[14:10] <hallyn> smoser, i did sudo tar xvf and pointed your script to the dir
[14:10] <smoser> which file did you download ?
[14:10] <hallyn> your lxc.tar.gz
[14:10] <jeh> pmatulis: I have it, don't use it, and don't want to be prompted for it
[14:10] <smoser> x86_64 ?
[14:10] <smoser> strange.
[14:11] <hallyn> smoser, regarding a new lxc template, yes, what i'd like is to just point the template to an existing, untarrted rootfs
[14:11] <hallyn> smoser, yes, x86-64
[14:11] <jeh> pmatulis: I tried "disk-detect     disk-detect/dmraid/enable       boolean false" but that didn't work
[14:11] <hallyn> i asked about untarring as root bc of course i first tried as me, but couldn't create the devices in it
[14:29] <pmorris> Since last night I've been unable to resolve hostnames and since it fails at boot time all my services including SSH fail also. Manually running `dhclient` makes hostnames resolve again, but restarting networking (or rebooting) recycles the problem. Any idea why this could have suddenly started happening?
[14:33] <zoopster> pmorris: something installed/changed/updated prior?
[14:33] <zoopster> pmorris: any errors in the logs?
[14:41] <cerberos> how do I find out which debian version ubuntu 10.10 uses? google is not my friend
[14:42] <_ruben> whatever was in debian testing at that time
[14:52] <Ursinha> Daviey, hey
[14:54] <smoser> hallyn, so would you expect a getty on a 'lxc-start -n <name>' output ?
[14:55] <hallyn> ?
[14:55] <hallyn> if you add a console entry to /etc/inittab, then you get a getty there
[14:55] <hallyn> worked for me anyway
[14:56] <smoser> yes
[14:56] <smoser> but would you expect one there?
[14:56] <smoser> adding one gets local echo on it.
[14:56] <hallyn> oh.  yes.
[14:56] <hallyn> what do you mean?
[14:56] <smoser> http://paste.ubuntu.com/719725/
[14:57] <hallyn> i always just do 'lxc-start -n x' and expect the console there.  but a lot of people prefer to do -d
[14:57] <hallyn> jikes
[14:57] <hallyn> i don't recall having that.  lemme retry
[14:57] <zul> Daviey: ntp uploaded btw
[14:57] <hallyn> i'll go read up more on augeas while waiting for cloudinit
[14:58] <smoser> hallyn, you can drop the cloud-init stuff.
[14:58] <smoser> just edit /var/lib/cloud/config.sh
[14:58] <smoser> NOCLOUD=1
[14:59] <smoser> err.. IS_NOCLOUD=1
[14:59] <hallyn> yeah, i see the echo.
[15:01] <hallyn> in ubuntu containers we do 'getty  -8 38400'
[15:01] <hallyn> but -8 is not supported in your getty
[15:02] <hallyn> odd.
[15:03] <smoser> i've got lxc in a func now
[15:03] <smoser> $ sudo lxc-start -n cirros
[15:03] <smoser> lxc-start: Permission denied - failed to create pty #0
[15:03] <smoser> lxc-start: failed to create the ttys
[15:03] <smoser> lxc-start: failed to initialize the container
[15:03] <hallyn> you've got me in a func
[15:04] <hallyn> does 'lxc-start -l debug -o outout -n cirros' give any more info?
[15:04] <hallyn> (in outout)
[15:04] <smoser> same thing i pasted
[15:04] <hallyn> no in file outout
[15:05] <smoser> same thing
[15:05] <smoser> strace shows: chown("/dev/pts/2", 0, 5)               = -1 EROFS (Read-only file system)
[15:06] <pmatulis> jeh: so go into your BIOS, remove all traces of fakeraid configuration (remove meta-data from you disks), and then disable fake-raid altogether
[15:07] <hallyn> smoser, you might want to check dmesg on the host for an oops
[15:07] <smoser> no.
[15:07] <smoser> its very clean
[15:07] <jeh> pmatulis: thanks. The solution was that, and zeroing the MBR
[15:11] <pmatulis> jeh: good stuff
[15:35] <smoser> hallyn, ok. so something ended up mounting /dev/pts ro
[15:35] <smoser> but i have no idea what
[15:37] <mari00> hi all, i'm struggling with a manual install of tomcat6.
[15:37] <hallyn> smoser, has it stopped?
[15:37] <mari00> i want to install it under a user (tomcat6) that uses the nologin script.
[15:38] <mari00> but i can't quite get it to work.
[15:38] <hallyn> smoser, i don't know why busybox-getty is doing the echo;
[15:38] <mari00> anyone here think they could help me?
[15:38] <hallyn> agetty does not
[15:48] <Ursinha> Daviey, are you still there? alive?
[15:48] <RoAkSoAx> Ursinha: he's a vampire... he never dies :)
[15:49] <Ursinha> RoAkSoAx, :)
[16:07] <Daviey> Ursinha: hey
[16:08] <koolhead17> hola RoAkSoAx
[16:12] <zul> Daviey: when you get a second can you have a look at http://people.canonical.com/~chucks/keystone
[16:14] <Daviey> zul: debdiff for the lazy? :)
[16:14] <Daviey> don't worry
[16:15] <zul> Daviey: aaaaaaaaa
[16:16] <zul> Daviey: http://people.canonical.com/~chucks/keystone/debdiff
[16:18] <Daviey> zul: waaaat is this:
[16:18] <Daviey> -        help="specifies port for Admin API to listen on (default is 5001)")
[16:18] <Daviey> +        help="specifies port for Admin API to listen on (default is 35357)")
[16:18] <Daviey> They changed the default port, seriously?
[16:18] <zul> Daviey: welcome to keystone country....stay for the flavor
[16:20] <koolhead17> zul: :P
[16:20] <koolhead17> Daviey: hahahaha
[16:20] <zul> Daviey: they still havent cut a release but that tarball matches the git tag but its renamed to something sensible
[16:21] <Daviey> zul: it's really much larger than i expected.
[16:21] <Daviey> It's a universe package, and still makes sense IMO.
[16:21] <zul> Daviey: yeah
[16:22] <Daviey> If keystone is currently as broken as people say
[16:23] <Daviey> zul:I would raise a discussion with ~ubuntu-mir, and concurrently we need to verify it works before uploading to -proposed i think.
[16:23] <zul> Daviey: i was going to upload to a ppa and ask people for testing
[16:23] <Daviey> zul: That is a great idea
[16:23] <koolhead17> zul: let me know once its done :)
[16:23] <zul> koolhead17: k
[16:24] <zul> Daviey: why ubuntu-mir though?
[16:24] <zul> i would say ubuntu-sru myself
[16:24] <Daviey> zul: gah, yes.
[16:24]  * zul gets SpamapS drunk
[16:25] <zul> anyways i need to find something to eat
[16:48] <koolhead17> Daviey: no luck even after removing the option :(
[16:51] <koolhead17> the debdiff http://paste.ubuntu.com/719822/
[16:57] <Daviey> koolhead17: crikey
[16:57] <Daviey> koolhead17: wait, you didn't remove the --with-pdo-sqlite thing?
[16:58] <koolhead17> Daviey: i think i did :(
[16:58] <koolhead17> lemme check
[17:00] <koolhead17> Daviey: lemme paste you the source file
[17:00] <koolhead17> *rules
[17:03] <koolhead17> Daviey: http://paste.ubuntu.com/719834/
[17:05] <koolhead17> Daviey: i did mistake actully there is 2ubuntu3.dsc created and i did debdiff with 1 & 2
[17:06] <koolhead17> http://paste.ubuntu.com/719838/ this has correct debdif :)
[17:09] <xibalba> hello everyone, i'm trying to use scponlyc and was wondering if i could get some assistance.
[17:10] <xibalba> here is some output i have on the issue, http://paste.ubuntu.com/719841/
[17:13] <Daviey> koolhead17: and that didn't fix it?
[17:14] <RoyK> [offtopic] Anyone that knows C well around? I'm doing a little fork() test here, I see text output is repeated (the Forking n times line) on this fork test if I don't fflush(stdout) - any idea why? http://paste.ubuntu.com/719846/
[17:15] <koolhead17> Daviey: no. but i am assumung am doing something wrong http://pastebin.com/mzYdG0k8  <-- apt-cache policy
[17:16] <Daviey> koolhead17: nah, that indicates you have a local package installed
[17:16] <Daviey> koolhead17: we'll try and track this down tomorrow in a shared screen session?
[17:16] <RoyK> pastebin.com is rather ugly stuff
[17:16] <RoyK> !paste
[17:16] <koolhead17> i my php5_5.3.8-2ubuntu3.dsc  then the packages inside  precise-i386_result should have 3 as extension
[17:16] <koolhead17>  
[17:16] <Daviey> ah yes
[17:16] <Daviey> true
[17:16] <koolhead17> Daviey: okey. :)
[17:16] <xibalba> hey fellas, anyone using scponlyc? i think i'm missing something really dumb here
[17:17] <koolhead17> Daviey: so am doing sumthing wrong sumwer
[17:17] <koolhead17> :)
[17:17] <Daviey> koolhead17: probably :)
[17:18] <koolhead17> Daviey: and the mistake is pbuilder-dist precise i386 build *ubuntu2.dsc  :P
[17:18] <koolhead17> so another 4 hrs now it seems
[17:18] <koolhead17> it has to be pbuilder-dist precise i386 build *ubuntu3.dsc IMHO
[17:19] <koolhead17> :P
[17:19] <zul> holy crap 4 hours to build php?
[17:20] <koolhead17> ;-)
[17:21] <koolhead17> Daviey: shall i redo it all again, i think this is the mistake :)
[17:21] <lynxman> zul: takes less than openoffice still
[17:21] <koolhead17> lynxman: hahaha. :P
[17:22]  * koolhead17 giving it a try
[17:22] <xibalba> here guys, check this out if you dont mind http://paste.ubuntu.com/719857/
[17:23]  * koolhead17 takes 15 mins break
[17:38] <RoyK> koolhead17: hepp! break's over!
[17:41]  * koolhead17 is back
[17:41] <koolhead17> RoyK: its always +- 10 mins
[17:41] <koolhead17> :)
[17:41] <RoyK> :)
[17:42] <RoyK> - A DBA walks into a NoSQL bar, but leaves because he can't find a table
[17:43] <koolhead17> RoyK: hahaha :)
[17:43] <just-a-visitor> =-O
[17:43] <koolhead17> Daviey: some success. got a nice error :D http://paste.ubuntu.com/719879/
[17:44] <koolhead17> just-a-visitor: howdy
[17:44] <just-a-visitor> koolhead17: Hi! It was a good joke.
[17:45] <koolhead17> indeed
[17:45] <koolhead17> RoyK: :)
[17:47] <Daviey> koolhead17: nice
[17:47] <koolhead17> Daviey: now wondering what next :)
[17:47] <koolhead17> ahaha
[17:48] <xibalba> hmm, do i need to be a +v to talk or something?
[17:48] <xibalba> or am i not following correct protocol for help
[17:49] <RoyK> xibalba: on some channels, you need to be identified - /msg nickserv help
[17:49] <xibalba> oh am i not? let me od that right now, whoops
[17:50] <xibalba> ok i'm identified now
[17:50] <RoyK> I've been able to read you fine all the time.....
[17:50] <xibalba> i'm just looking for a little guidance on scponlyc, the chroot'd scponly shell
[17:50]  * RoyK has no idea how that works...
[17:51] <xibalba> drats
[17:51] <koolhead17> nor even me :(
[17:52] <xibalba> double drats
[17:52] <RoyK> 'even' :D
[17:53] <Daviey> koolhead17: My brain is baked at the moment.
[17:54] <koolhead17> Daviey: will look into it tomorrow. cheers!! :)
[17:54] <Daviey> rocking
[17:54] <xibalba> thanks guys, see ya later
[17:55]  * RoyK likes backblaze.com - $5 per month for backup of up to ∞GB
[18:03] <cemc> o/ I'm using LVM for my KVM guests. if I create a new LV manually (with lvcreate etc), it doesn't show up in virt-manager in the storage pool. is there a way to refresh that list?
[18:05] <pmatulis> cemc: you should be able to browse to the device
[18:05] <koolhead17> RoyK: am paying allready a lot to my VPS provider like 125$ monthly
[18:05] <koolhead17> :P
[18:06] <RoyK> koolhead17: for what?
[18:06] <cemc> pmatulis: I don't see it in the list
[18:07] <koolhead17> hell no  133$ yearly for 512 RAM and 40 GB hdd
[18:07] <RoyK> koolhead17: HAHA
[18:07]  * RoyK has some private servers for such use
[18:08] <koolhead17> RoyK: am running few wordpress blog and galler2 thats it
[18:08] <koolhead17> :)
[18:09] <RoyK> erm - not $125 monthly?
[18:09] <RoyK> $133 a year isn't so bad
[18:16] <RoyK> koolhead17: do they provide backup on that system as well?
[18:17] <koolhead17> RoyK: nopes. no guarntee of tht
[18:18] <RoyK> that sucks rather badly.....
[18:18] <koolhead17> RoyK: i know
[18:22] <kirkland> SpamapS: ping
[18:22] <arrrghhh> hey all.  anyone use/know how to use powernap?
[18:23] <lifeless> quick, hurry up and sleep!
[18:23] <arrrghhh> lol basically yes
[18:23] <arrrghhh> then quick, hurry wake up!
[18:23] <lifeless> you install it
[18:23] <arrrghhh> it is installed
[18:23] <lifeless> thats all thats needed AIUI, it hooks into the existing policies
[18:23] <lifeless> to turn off cpus when they are idle and htings like that
[18:23] <arrrghhh> hrm
[18:23] <kirkland> arrrghhh: serveral of us, yes.  RoAkSoAx is the current maintainer, I'm the original author.  what are you trying to do?
[18:24] <arrrghhh> well i guess i just wanted to know more about it
[18:24] <arrrghhh> ideally i'd like it to shutdown/hibernate/suspend when not in use
[18:24] <arrrghhh> and the tricky part to me seems to be waking it back up...
[18:24] <arrrghhh> do i have to do anything to configure that?  how does it work?
[18:24] <kirkland> arrrghhh: there's a good series of articles at http://blog.dustinkirkland.com/search/label/PowerNap
[18:25] <arrrghhh> excellent
[18:25] <arrrghhh> thank you
[18:25] <kirkland> arrrghhh: including several slideshows and presentations
[18:25] <kirkland> arrrghhh: you can use powerwake (wake-on-lan) to wake it back up
[18:25] <kirkland> arrrghhh: if you need to wake it remotely
[18:25] <kirkland> arrrghhh: or just press the power button to wake it locally
[18:25] <arrrghhh> heh
[18:26] <arrrghhh> well i was reading something about it detecting traffic?
[18:26] <arrrghhh> so when it detects a ssh attempt or a upnp attempt, etc it wakes?
[18:26] <arrrghhh> seems from a shutdown state WoL is the only option.
[18:28] <kirkland> arrrghhh: RoAkSoAx has been working on an arp monitor, where you'd have one server watching for arp traffic to a bunch of others that might be powernapping;  and when traffic to their addresses shows up, it would powerwake them
[18:28] <kirkland> arrrghhh: that bit is still proof of concept stage
[18:28] <kirkland> arrrghhh: currently, though, you can come out of powersave mode on ssh connection attempts, that's trivial
[18:29] <arrrghhh> hrm ok
[18:29] <kirkland> arrrghhh: but you can't yet come out of suspend/hibernate/poweroff without a WoL packet first
[18:29] <arrrghhh> i see
[18:29] <arrrghhh> i'll do some more reading on powernap.  perhaps there's some other aspects of it that i can use that will help reduce power consumption
[18:29] <kirkland> arrrghhh: i have used an ssh alias, sshwol, that powerwakes a server, and then sits there in a loop ssh'ing to it until it's back online
[18:29] <kirkland> arrrghhh: cool, good luck
[18:30] <arrrghhh> my buddy was talking about some WHS plugin, and made me jealous.  i figured there was some way to do it in linux... :D
[18:30] <arrrghhh> lights out is the name of the plugin if you're interested
[18:30] <arrrghhh> seems like it depends on WoL as well tho....
[18:31] <arrrghhh> and the developer just built some client-based app that sends that magic packet...
[18:31] <panfist> i'm trying to run an ubuntu server vm on my desktop, but something changed in my environment causing it to be unable to mount a network share and the boot is stuck complaining that mountall terminated with status 32
[18:32] <panfist> short of pointing the virtual machine to a live rescue disk, is there a hotkey to press to get to rescue mode while it's booting?
[18:32] <arrrghhh> well if you hit esc to get the grub menu
[18:32] <arrrghhh> you could try getting into rescue mode...
[18:33] <panfist> i'm not sure why my other machines show a menu to choose what to boot into but this one does not, it just goes straight through from the virtual bios screen to the boot output
[18:33] <panfist> even if i hit esc repeatedly while it's booting i can't catch a grub menu
[18:33] <arrrghhh> hrm
[18:33] <panfist> this is 10.04 by the way
[18:33] <arrrghhh> you're sure the VM is configured correctly?
[18:33] <arrrghhh> ok
[18:34] <panfist> i'm pretty sure the vm is configured correctly, it's trying to mount an NFS share that's unavailable
[18:34] <panfist> my coworker is absent and the share is on a machine that's shut down in his locked office
[18:34] <arrrghhh> the NFS share is dependent on the VM booting?
[18:34] <arrrghhh> uhhh
[18:35] <arrrghhh> unless the share is required for your VM to boot
[18:35] <arrrghhh> i don't see why it would _prevent_ it from booting....
[18:35] <panfist> it's saying mount error(113): no route to host; refer to mount man page...; mountall: mount /path/to/mountpoint [485] terminated with status 32;
[18:35] <panfist> and it's just sitting there
[18:36] <arrrghhh> is the share required for the VM to function?
[18:36] <panfist> no, the share just has some cgi scripts that the vm crunches, but i don't see how they are crucial to the boot process for the vm
[18:37] <arrrghhh> hrm
[18:37] <panfist> if it were ubuntu desktop, it would say something like, "press s to skip or m for manual recovery"
[18:37] <arrrghhh> this NFS mount is in fstab i presume?
[18:37] <panfist> yes
[18:37] <arrrghhh> interesting.
[18:38] <arrrghhh> well
[18:38] <arrrghhh> you could point a livecd ISO at it
[18:39] <arrrghhh> and go the long way, fix the fstab entry
[18:39] <arrrghhh> i don't know why it would halt on a failed NFS mount....
[18:39] <arrrghhh> unless it was for something silly like /boot :P
[18:39] <panfist> yeah, i've done that before, but this isn't the first time it's happened and it's annoying to have to boot a live iso to fix this problem when i should be able to get a proper rescue environment locally if i had grub menu
[18:40] <arrrghhh> i agree
[18:40] <arrrghhh> and *normally* you'd just hit esc.
[18:40] <arrrghhh> that's odd that it isn't working.  i'd blame the VM software.... but i don't know for sure.
[18:40] <arrrghhh> i'm not an expert on the topic, i've only run ubuntu server in vbox a handful of times.
[18:40] <arrrghhh> the rest are all physical machines...
[18:42] <panfist> oh well, thanks, back to the old solution of using a live cd i guess
[18:42] <panfist> i guess grub must be configured to have a 0 timeout or something and i should be able to change that to another value after i get back in
[18:43] <arrrghhh> yea but even still
[18:43] <arrrghhh> with 0 timeout, i thought there was still a (albeit short) period where you can hit esc
[18:43] <arrrghhh> i might just be blowing smoke tho :P
[18:43] <panfist> like you said, probably blame it on the vm software
[18:44] <arrrghhh> the simplest thing i can do now heh
[19:10] <Bilge> Why might hostname lookup fail every time I restart networking or reboot?
[19:10] <Bilge> Even though I can run dhclient manually and everything appears to be fine
[19:11] <xranby> Bilge: have you checked your routing tables?
[19:11] <xranby> Bilge: perhaps dhclient are somehow not run on bootup for you?
[19:11] <Bilge> I don't know much about network or the first thing about routing tables
[19:11] <Bilge> My server is leased from a company
[19:12] <Bilge> Without changing anything on the box suddenly this problem started ocurring since yesterday
[19:12] <Bilge> So maybe it's something wrong with their network but I guess not if I can just run dhclient and everything is fine?
[19:13] <Bilge> Is it _meant_ to be run on startup?
[19:13] <Bilge> The only strange error I get on boot is this one:
[19:13] <Bilge> removed hw address from /etc/udev/rules.d/70-persistent-net.rules (eth0)
[19:13] <Bilge> But network dependent services like SSH have already failed to start before that point
[19:13] <xranby> Bilge: have you switched network cards?
[19:14] <Bilge> Not to my knowledge
[19:14] <Bilge> I would imagine that would result in downtime and I'm not aware of any, as I said, the server is leased from a company and they should inform me if they were to make any changes to the hardware
[19:15] <Bilge> The only thing that has changed to my knowledge is occasionally running apt-get update
[19:15] <xranby> because if they had switched network cards then the mac address would change and the new card would get allocated to   eth1 instead of eth0
[19:15] <xranby> and thus the default network configuration would stop working
[19:15] <Bilge> Is there some way I could check that?
[19:15] <xranby> you can run ifconfig
[19:15] <xranby> it will display the mac address of your network card
[19:16] <xranby> then you can check the /etc/udev/rules.d/70-persistent-net.rules and check that the same mac adress are used for eth0
[19:16] <Bilge> There are no entries in that file
[19:16] <xranby> if they do not match
[19:16] <xranby> hmm
[19:16] <Bilge> Just a load of commented out stuff
[19:16] <xranby> look at the bottom
[19:17] <xranby> i would have expected you to find something like
[19:17] <xranby> SUBSYSTEM=="net", ACTION=="add", DRIVERS=="?*", ATTR{address}=="00:19:99:6d:80:8f", ATTR{dev_id}=="0x0", ATTR{type}=="1", KERNEL=="eth*", NAME="eth0"
[19:17] <Bilge> There's nothing in that file at all
[19:17] <xranby> because the purpose of that file are to make sure that the eth names stays the same for each reboot
[19:18] <Bilge> If I run grep '^[^#]' on it
[19:18] <Bilge> i.e. filter all lines starting with #
[19:18] <Bilge> I get nothing
[19:18] <xranby> # Entries are automatically added by the 75-persistent-net-generator.rules
[19:19] <Bilge> That file does not exist
[19:20] <Bilge> Mine says it is generated by /lib/udev/write_net_rules
[19:20] <xranby> Bilge: which ubuntu-server release are you running?
[19:20] <Bilge> The latest LTS
[19:20] <Bilge> 10.04.3 LTS
[19:21] <xranby> Bilge: if ifconfig indicate that your network card gets names something else then eth0 then
[19:21] <xranby> try add a similar line manually
[19:22] <Bilge> ifconfig doesn't seem to make any reference to ethN
[19:22] <Bilge> Actually yeah it does
[19:22] <Bilge> I have eth0, eth0:0, eth0:1 and eth0:2
[19:22] <Bilge> For my additional IPs
[19:23] <Bilge> They all have the same HWaddr
[19:24] <Bilge> They're all set up manually in /etc/network/interfaces
[19:24] <xranby> if they are setup manually then why use dhcp?
[19:25] <Bilge> Because running that was the only way I could get it to work
[19:25] <Bilge> Maybe DHCP is picking up something other than what is configured in that file?
[19:25] <Bilge> It specifies "network", "broadcast" and "gateway" in there
[19:25] <Bilge> Maybe they changed without notice?
[19:25] <xranby> dhcp sets default gateway in the routing tables as well
[19:26] <xranby> oh right .. you had gateway in there..
[19:26] <xranby> if you run route
[19:26] <xranby> you should be able to see a default gw
[19:27] <xranby> if you can ping 8.8.8.8  (google dns)
[19:27] <xranby> then you might be simply missing a valid dns config
[19:27] <xranby> your dns server perhaps have switched ip?
[19:27] <xranby> Bilge: look inside /etc/resolv.conf
[19:27] <xranby> and check that the nameserver listed here still work
[19:29] <xranby> to summarize 1. check that the network card got the right eth number 2. check that it have been assigned an ip using ifconfig 3. check the routing table using route 4. check the dns nameserver in /etc/resolv.conf
[19:29] <xranby> if all those pass then you should have network acess
[19:29] <Bilge> Going to be about 15 mins to reboot the kVKM since while trying to get it to print a # character it decided to reboot itself :/
[19:30] <xranby> all these steps gets set automatically when you run dhclient
[19:30] <Bilge> ifconfig looks normal
[19:31] <Bilge> Is there a way to tell if a nameserver is behaving properly?
[19:31] <Bilge> Like an acid test?
[19:31] <Bilge> Also how do I use `route`?
[19:31] <xranby> if ping google.com   pings google then the nameserver test pass
[19:32] <xranby> you can also use the cooand dig
[19:32] <xranby> command dig
[19:32] <xranby> dig google.com
[19:32] <xranby> will list where it obtained the knowlege about googles ip
[19:32] <RoyK> dig soa your.tld
[19:32] <RoyK> dig a
[19:32] <RoyK> etc
[19:33] <xranby> you can use dig to test a specific nameserver   dig @nameserver google.com
[19:34] <xranby> RoyK: simply type route
[19:34] <xranby> will list active routes
[19:34] <RoyK> dig soa @hole.somewhere.tld spade.somewhereelse.tld
[19:34] <RoyK> xranby: ?
[19:35] <xranby> RoyK: example if your router acts as a nameserver
[19:35] <xranby> dig @192.168.1.1  google.com
[19:35] <xranby> then dig will ask the nameserver at 192.168.1.1 where the ip are for google.com
[19:35] <RoyK> xranby: I know those tools quite well, thanks ;)
[19:36] <xranby> (21.31.07) Bilge: Is there a way to tell if a nameserver is behaving properly?
[19:36] <xranby> ok
[19:36] <RoyK> xranby: RoyK != Bilge
[19:36] <Bilge> :3
[19:36] <xranby> .. ah
[19:36] <xranby> my fault
[19:37] <pmatulis> Bilge: and that smiley means what exactly?
[19:37] <Bilge> Kitty
[19:37] <kirkland> lifeless: have a second to help smoser with a squid question?
[19:38] <RoyK> what about the squid?
[19:38] <lifeless> course
[19:38] <smoser> not a big deal...
[19:38] <smoser> twice in a row i do: http_proxy=http://nelson:3128/ wget http://smoser.brickies.net/scratch/cirros-0.3.0-i386-disk.vhd
[19:38] <smoser> both times i see:
[19:38] <smoser> 1319657440.372  14495 192.168.1.101 TCP_MISS/200 27272579 GET http://smoser.brickies.net/scratch/cirros-0.3.0-i386-disk.vhd - DIRECT/69.163.204.191 text/plain
[19:39] <smoser> squid.conf at http://paste.ubuntu.com/719984/
[19:39] <smoser> image is 27272192 (~26M)
[19:40] <RoyK> smoser: isn't there a max cachable object size default somewhere?
[19:40] <lifeless> smoser: don't you mean 27272579 ? :)
[19:40] <lifeless> and yes, your object size cap needs raising
[19:41] <smoser> i suspected that bug didn't see it explicitly anywhere.
[19:41] <RoyK> #Default:
[19:41] <RoyK> # maximum_object_size 4096 KB
[19:41] <smoser> oh goodness. thats too small. :)
[19:42] <lifeless> smoser: if you're doing live boots off that you may want to raise maximum_object_size_in_memory too
[19:42] <maxtmahem> Question: Best practice to put my changes to suoders in a new file in sudoers.d?
[19:42] <smoser> maxtmahem, yes.
[19:42] <lifeless> maxtmahem: I wouldn't, because its not sanity checked and you can wedge your system
[19:42] <lifeless> maxtmahem: I always use visudo
[19:42] <smoser> lifeless, what is the default value for maximum_object_size ?
[19:42] <RoyK> smoser: see my comment above
[19:43] <lifeless> 4M still (though I haven't pulled trunk in a while :P)
[19:43] <smoser> RoyK, yeah, i saw that, but:
[19:43] <smoser> my /usr/share/doc/squid/examples/squid.conf says:
[19:43] <smoser> #Default:
[19:43] <smoser> # maximum_object_size 20480 KB
[19:43] <maxtmahem> lifeless: I use visudo -f newfile name, should do it's sanity checking for then right?
[19:43] <lifeless> maxtmahem: hmm, I'm not sure tbh :)
[19:43] <smoser> either way, kirkland it would seem to me that if the primary purpose is a deb cache, then you want to bump that a bit in the orchestra config.
[19:43] <RoyK> smoser: my /etc/squid/squid.conf says 4M
[19:44] <lifeless> smoser: also - http://redbot.org/?uri=http%3A%2F%2Fsmoser.brickies.net%2Fscratch%2Fcirros-0.3.0-i386-disk.vhd
[19:44] <RoyK> smoser: anyway - just change it...
[19:44] <RoyK> lifeless: now that'll be a tough job to cache....
[19:44] <smoser> lifeless, RoyK htank you.
[19:45] <kirkland> smoser: yeah, no doubt!
[19:45] <smoser> i tihnk it must be 20M in the ubuntu build , unless the doc there is wrong.
[19:45] <lifeless> smoser: 4M
[19:45] <lifeless> http://bazaar.launchpad.net/~squid/squid/3-trunk/view/head:/src/cf.data.pre#L2897
[19:45] <Bilge> My nameserver is set to 10.0.2.3 heh
[19:45] <smoser> but still, 20m wont cache your kernel.
[19:45] <Bilge> dig fails
[19:45] <smoser> lifeless, would it not be possible that that is changed in ubuntu build ?
[19:46] <lifeless> its possible, but I'd hope that the ubuntu/debian maintainer would at least post such a patch upstream
[19:46]  * RoyK spent a couple of hours yesterday trying to describe to this programmer that his HTTP headers said 'don't cache this', and no, I can't force squid to cache that when the server says no - fscking .net noobs...
[19:46] <lifeless> and I don't recall discussion about such a patch
[19:46] <lifeless> RoyK: well, you can force squid to do it, may not be a good idea.
[19:46] <RoyK> lifeless: heh - how can I do that?
[19:47] <maxtmahem> lifeless: Well at least I didn't break it doing like that.
[19:47] <RoyK> I know it's stupid, but it might shut them off if I tell them squid is working, but their site is bogus....
[19:47] <hallyn> smoser, why for you want to use vpc!
[19:47] <lifeless> smoser: also you probably want to set a cache ttl on the resource, because its changed recently the default ttl will be fairly short
[19:47] <smoser> hallyn, vpc is what all the cool kids are using.
[19:47] <smoser> kvm is so old.
[19:47] <lifeless> RoyK: refresh_pattern is the primary method to override stuff
[19:47] <lifeless> RoyK: you may need a custom build because http-rule-breaking-options are off by default
[19:47]  * hallyn scratches his head
[19:48] <hallyn> time to retire
[19:48] <RoyK> lifeless: thanks
[19:48] <smoser> hallyn i was playing with something from cloudstack and they want a vhd format in.
[19:48] <kirkland> smoser: i'm making that change now to orchestra
[19:48] <kirkland> smoser: i wonder what it should be ...
[19:48] <lifeless> kirkland: what what should be ?
[19:49] <kirkland> lifeless: the default maximum_object_size in Orchestra's squid
[19:49] <smoser> lifeless, for a squid proxy who's primay pourpose is to serve debs
[19:49] <lifeless> do you do net booting ?
[19:49] <kirkland> lifeless: yes
[19:49] <smoser> that pastebin i sent is what kirkland has now in orchestra
[19:49] <lifeless> whats the largest asset you'll be getting - across all of deb/kernel/initramfs
[19:50] <kirkland> lifeless: hmm, ISOs maybe?
[19:50] <kirkland> lifeless: i was thinking 200MB or even 1GB
[19:51] <lifeless> kirkland: if folk are getting those as part of the bootstrap of a cluster, caching would be good ;)
[19:51] <lifeless> kirkland: so, I suggest you do this:
[19:51] <kirkland> lifeless: really, it's mostly just deb's
[19:51] <kirkland> lifeless: so i'd need to find the largest debs we install
[19:51] <kirkland> lifeless: at least the kernel, for sure
[19:51] <smoser> openoffice has to win that.
[19:51] <smoser> had i not rm -Rf'd my local archive recently i could tell you
[19:52] <lifeless> kirkland: not caching packages/releases/ etc will quite negatively impact your bootstrap performance
[19:52] <kirkland> i'm going to set it to 750MB, which should cover standard ISOs
[19:52] <lifeless> kirkland: you should instead purge them when you want to force a change
[19:52] <kirkland> lifeless: of course
[19:52] <lifeless> anyhow, to cache big objects do this:
[19:52] <lifeless> cache_dir aufs /var/spool/squid 40000 16 256
[19:53] <lifeless> bah
[19:53] <Bilge> xranby: dig doesn't seem to work no matter what is specified as the name server
[19:53] <lifeless> cache_dir aufs /var/spool/squid-small 40000 16 256 max-size=40M
[19:53] <lifeless> cache_dir aufs /var/spool/squid-big 40000 16 256
[19:54] <lifeless> this will put anything smaller than 40M in the first cache dir until its full, and then overflow into the second
[19:54] <xranby> Bilge: was the routing table correct?
[19:54] <lifeless> big things will only go into the second
[19:54] <kirkland> lifeless: interesting
[19:54] <lifeless> this stops big things evicting a tonne of old-but-valid debs :)
[19:54] <kirkland> lifeless: i like it
[19:54] <lifeless> (they can evict them from the second dir of course)
[19:56] <lifeless> or you can switch to a different repl algorithm like dual-frequency-size, but two dirs is easier to understand :)
[19:56] <Bilge> xranby: I don't know
[19:56] <Bilge> I don't know how to read the output I'm getting
[19:57] <kirkland> lifeless: can you proof read http://paste.ubuntu.com/720010/
[19:57] <kirkland> lifeless: just the last 3 lines i changed per this discussion
[19:57] <lifeless> yeah, that should do
[19:57] <lifeless> try it :)
[19:57] <kirkland> lifeless: k ...
[19:58] <lifeless> I suggest adding a couple of comments
[19:58] <smoser> kirkland, did you double the size of your cache though?
[19:58] <smoser> potentially
[19:58] <lifeless> like 'netbooting isos is better if they get cached' and 'reserve small object space so big objects can't use it all up'
[19:59] <smoser> lifeless, comments are for weenies
[19:59] <smoser> :)
[19:59] <lifeless> smoser: I'm a weeny!
[19:59] <smoser> oh, and btw, thank you very much for your help.
[19:59] <kirkland> lifeless: yeah, thanks
[19:59] <lifeless> anytime
[20:02] <allegrem_> Hi ! I've a little problem with Samba between a Kubuntu (as a server) and a Windows XP (as a client). My workgroup 'Samba' is shown in the workgroups list on Windows, but when I click on it, an error occurs "Nom du réseau introuvable" ( ~~ network name not found). Any idea ?
[20:03] <Bilge> xranby: dhclient changes the routes
[20:04] <kirkland> lifeless: hmm, not quite there yet, looks like i need to manually create some directories, 2011/10/26 15:04:19| /var/spool/squid/small/00: (2) No such file or directory
[20:05] <lifeless> kirkland: squid -z of course ;)
[20:06] <kirkland> lifeless: tried that, not working
[20:06] <lifeless> kirkland: orly ? pastebin
[20:06] <lifeless> (you ran it as squid right ?)
[20:06] <kirkland> lifeless: ah, EACCES          13      /* Permission denied */
[20:06] <lifeless> :P
[20:07] <kirkland> lifeless: 'proxy'
[20:07] <lifeless> 'meh' @ overloaded names
[20:08] <kirkland> lifeless: bingo
[20:08] <kirkland> lifeless: smoser: working like a champ
[20:08] <lifeless> sudo squid -z should have worked too, I think.
[20:09] <kirkland> lifeless: well, i created those directories, but they were owned by root:root;  changed that to proxy:proxy and squid -z worked fine
[20:09] <lifeless> kirkland: ok, so you foot-gunned :P
[20:09] <lifeless> kirkland: never make swap dirs by hand
[20:09] <kirkland> i'll get the install bits cleaned up in in orchestra and we're good to go
[20:09] <kirkland> lifeless: :-)
[20:10] <kirkland> lifeless: sure thing, mate
[20:10] <smoser> i did the same here.
[20:10] <RoyK> anyone that knows if it's possible to use squid for both a front and reverse proxy? I have a reverse proxy at work, but since it's mostly idle, it'll be nice to use it as a proxy for the internal clients as well
[20:10] <smoser> and now i'm getting 403 forbiddenon anythihg through it.
[20:10] <lifeless> RoyK: yes, setup two listening ports
[20:10] <lifeless> smoser: thats going to be unrelated
[20:10] <Bilge> xranby: is it possible to specify the IP for eth0 but still pick up the other setting such as gateway, network and broadcast via dhcp instead?
[20:10] <lifeless> smoser: if a cache dir can't take an object, the request doesn't barf, it just doesn't cache
[20:10] <RoyK> lifeless: you mean two squid instances, or just two ports?
[20:10] <allegrem_> Any idea about Samba ?
[20:11] <lifeless> RoyK: there are some global things we should make non-global, but its mostly ok
[20:11] <lifeless> RoyK: http_port 80 accel\nhttp_port 8080
[20:11] <RoyK> lifeless: thanks
[20:11] <lifeless> RoyK: you can use myport in acls to differentiate the two
[20:12] <RoyK> lifeless: but then, using it on port 80 for both uses won't be very easy?
[20:12] <lifeless> RoyK: not one port, two ports - 80 for your reverse proxy, 8080 for your forward proxy
[20:12] <RoyK> ok
[20:12] <lifeless> yes, it requires a little care
[20:13] <lifeless> I'm not endorsing it :) just answering your question :)
[20:13] <RoyK> :)
[20:13] <lifeless> you also need to make sure your cache_peers are properly acled
[20:13] <lifeless> so that forward requests don't get sent to them
[20:14]  * RoyK tried to setup the reverse proxy as a VM on hyper-v and hyper-v just cut off the network from it in minutes... hyper-v sucks rather badly at running linux guests :P
[20:14] <lifeless> oh, and make sure you explicitly prevent the accel port from making direct requests - it must always use a cache peer
[20:15] <lifeless> otherwise you turn your cache into an open relay
[20:15] <RoyK> I guess I'll use a separate box, then - makes life easier ....
[20:15] <lifeless> its harder to mess up:)
[20:15] <RoyK> yeah
[20:36] <allegrem_> Hi ! I've a little problem with Samba between a Kubuntu (as a server) and a Windows XP (as a client). My workgroup 'Samba' is shown in the workgroups list on Windows, but when I click on it, an error occurs "Nom du réseau introuvable" ( ~~ network name not found). Any idea ?
[20:38] <RoyK> !samba
[20:41] <allegrem_> ubottu: are you a real bot ?
[20:41] <allegrem_> awesome !
[20:48] <RoyK> is there a way to natively share a filesystem with kvm vm?
[20:48] <adam_g> nfs?
[20:48] <RoyK> not a unix vm, windows, and preferably without samba
[20:49] <allegrem_> has anyone ever met the "Can't become connected user!" error with samba ??
[20:49] <RoyK> trying to work my way around the limitations of backblaze backup
[20:49] <RoyK> allegrem_: it's probably just a configuration issue.....
[20:50] <RoyK> allegrem_: so you need to be a bit more specific before someone can help you...
[20:50] <allegrem_> I think windows is trying to log as guest, but he can't. How could I force him to ask login and passwd ?
[20:51] <RoyK> in win7?
[20:51] <allegrem_> I forgot to say that there is an line in my samba logs saying "Can't become connected user!"
[20:51] <allegrem_> windows xp
[20:51] <RoyK> it should ask for a username and password
[20:52] <RoyK> default setup requires you to create an smb user IIRC
[20:52] <RoyK> man smbpasswd
[20:53] <allegrem_> I already created two test users (isabelle, carole) on the kubuntu PC, and I run smbpasswd for each of them
[20:53] <SpamapS> kirkland: pong, sup?
[20:54] <SpamapS> kirkland: (sorry, sprinting, haven't been watching IRC)
[20:54] <allegrem_> but the user on the windows xp is "simone". Is it important ?
[20:54] <RoyK> check the samba logs
[20:55] <allegrem_> it says "Can't become connected user!"
[20:55] <RoyK> allegrem_: if you're just on a home network, you might not need authentication
[20:55] <RoyK> if so, set security = share in smb.conf
[20:56] <allegrem_> i'm just testing a samba install on my home network, but the goal is to deploy it at my mother's office
[20:56] <allegrem_> so I do need security = users :p
[20:56] <RoyK> or user
[20:57] <RoyK> allegrem_: unless they have a windows server with AD or something, which makes the picture look rather different
[20:57] <allegrem_> no they haven't
[20:58] <allegrem_> there are just 3 users to set up, I can do it manually
[20:59] <RoyK> ok
[20:59] <RoyK> then security = user should do well
[20:59] <RoyK> make sure all those users have unix accounts, though
[20:59] <RoyK> useradd -m thisuser .....
[21:00] <allegrem_> I'm testing with only two users
[21:00] <allegrem_> (isabelle, carole)
[21:01] <allegrem_> I created unix account, set up smbpasswd
[21:02] <RoyK> try creating user simone
[21:02] <RoyK> see if that works any better
[21:02] <RoyK> since that's apparently the user logged into the windoze machine
[21:09] <maxtmahem> if you have AD you can try out likewise open to add the machine to the domain real easy.
[21:11] <allegrem_> it still doesn't work :( I'm trying to reboot the windows pc
[21:15] <tjaalton> kirkland: hey, ideas why 'adduser --encrypt-home foo' leaves the home dir with 500 perms on oneiric?
[21:15] <kirkland> tjaalton: those perms are correct, when $HOME is not mounted
[21:15] <kirkland> tjaalton: upon being mounted, they'll be 700
[21:16] <tjaalton> kirkland: ok, well at least the preseeded installation leaves the ecryptfs setup somehow broken
[21:16] <kirkland> tjaalton: the 500 is to keep users from inadvertently writing data in clear text into their $HOME when it's unmounted
[21:16] <kirkland> tjaalton: ie, "leaking" clear text data to disk
[21:17] <kirkland> tjaalton: hmm, interesting
[21:17] <kirkland> tjaalton: oh, hmm, i have an idea
[21:18] <kirkland> tjaalton: so you used literally 'adduser --encrypt-home foo' in the late-command or something?
[21:18] <kirkland> tjaalton: or did you preseed the user to encrypt-their-home in d-i?
[21:19] <tjaalton> kirkland: no I preseeded "d-i user-setup/encrypt-home boolean true" like on the doc, wondered why the mount fails and started digging further
[21:19] <kirkland> tjaalton: interesting
[21:19] <kirkland> tjaalton: okay, do you have a $HOME/.ecryptfs/wrapped-passphrase file?
[21:20] <kirkland> tjaalton: can you ls -alF $HOME/.ecryptfs/
[21:20] <tjaalton> kirkland: nope, don't have that
[21:20] <kirkland> tjaalton: and pastebin that for me?
[21:20] <kirkland> tjaalton: ah, that's the problem
[21:20] <kirkland> tjaalton: hmm, i wonder what went wrong there
[21:21] <tjaalton> I'll pastebin the snippet from the installation syslog
[21:21] <tjaalton> http://pastebin.com/LBBLbfnb
[21:21] <tjaalton> see the error from stty
[21:29] <kirkland> tjaalton: how odd;  this is 11.10?
[21:29] <tjaalton> kirkland: yep
[21:29] <tjaalton> maybe I'll do another run with full debconf debug
[21:30] <kirkland> tjaalton: and you did preseed your password in as well?
[21:30] <tjaalton> kirkland: yes. I can login from the vt's just fine
[21:31] <kirkland> tjaalton: you've already rebooted after the installation, of course?
[21:31] <tjaalton> kirkland: yep
[21:32] <kirkland> tjaalton: okay, well, if you don't have that wrapped-passphrase file, you'll never be able to get to your new home/skeleton data in this installation
[21:32] <kirkland> tjaalton: so you'll need to kill this user, add another one, etc. if you want to manually salvage this installation
[21:32] <kirkland> tjaalton: i'm trying to track down the bug right now though
[21:33] <kirkland> tjaalton: can you pastebin ls -alF $HOME/.ecryptfs/ ?
[21:34] <tjaalton> kirkland: nah it's a throwaway installation, testing stuff to install OOTB all the way
[21:34] <kirkland> tjaalton: okay
[21:36] <tjaalton> kirkland: http://paste.ubuntu.com/720118
[21:37] <kirkland> tjaalton: hmm, yeah, dang, everything else is there except for wrapped-passphrase
[21:37] <tjaalton> kirkland: ok, so I'll reinstall it with DEBCONF_DEBUG=developer
[21:37] <kirkland> tjaalton: thank you!
[21:39] <TheEvilPhoenix> anyone got a spare maverick server they can confirm a bug with?
[21:40] <kirkland> tjaalton: one more question
[21:40] <kirkland> tjaalton: have you wiped this install yet?
[21:40] <tjaalton> btw, rsyslog.d/99-orchestra.conf pushes the installation logging to the server /var/log/syslog as well.. don't think that's intended
[21:40] <tjaalton> kirkland: partitioning..
[21:40] <kirkland> tjaalton: :-)
[21:41] <tjaalton> it takes 10min to install :)
[21:41] <kirkland> tjaalton: re: rsyslog ... can you find a way to separate out the installation syslogging?
[21:41] <tjaalton> hmm, could have dropped the desktop task
[21:41] <kirkland> tjaalton: i couldn't in time to fix it for 11.10
[21:41] <kirkland> tjaalton: the installation rsyslogging must happen over udp
[21:42] <TheEvilPhoenix> can anyone confirm this in maverick?  https://bugs.launchpad.net/ubuntu/+source/php5/+bug/882291
[21:42] <tjaalton> kirkland: haven't looked at that too much, just noticed it by accident :)
[21:42] <kirkland> tjaalton: yeah
[21:42] <TheEvilPhoenix> i've confirmed it *fixed* in natty and later
[21:43] <kirkland> tjaalton: could you get a shell on that box during installation, at the very end, before reboot?
[21:44] <tjaalton> kirkland: it doesn't stop :/
[21:44] <kirkland> tjaalton: heh :-)
[21:44] <kirkland> tjaalton: can you get on it now?
[21:44] <tjaalton> i should comment out the line that does that
[21:44] <tjaalton> yes it's configuring packages
[21:44] <kirkland> tjaalton: has the user been created yet?
[21:44] <tjaalton> but the crypt setup is in the final phase
[21:44] <kirkland> yeah
[21:44] <tjaalton> nope
[21:44] <kirkland> dang
[21:45] <tjaalton> I'll just reboot and do that
[21:46] <tjaalton> there, installing without ubuntu-desktop this time
[21:47] <SpamapS> TheEvilPhoenix: marked as fix released, opened a maverick task
[21:47] <kirkland> tjaalton: okay, i think i have a fix
[21:47] <tjaalton> kirkland: cool :)
[21:47] <Daviey> hallyn: the lxc bug, i targetted it because it seemed to be a case of checking and applying the patch.
[21:47] <SpamapS> TheEvilPhoenix: unlikely it will get fixed in Maverick ... but if it affects lucid.. maybe
[21:47] <Daviey> hallyn: seems it is now more complicated.
[21:47] <kirkland> tjaalton: http://paste.ubuntu.com/720131/
[21:48] <Daviey> hallyn: It would be quite good if we could SRU a fix for Oneiric.
[21:48] <TheEvilPhoenix> SpamapS:  :)
[21:48] <kirkland> tjaalton: i think what's happening is that we're landing in the then rather than the else part of that if block
[21:48] <tjaalton> kirkland: ha, I'll hot patch it in
[21:48] <kirkland> tjaalton: that would be cool, if you can jam that into the ecryptfs-setup-private in an install and confirm it, i'll commit/upload/release
[21:49] <TheEvilPhoenix> SpamapS:  i dont have Lucid around atm
[21:49] <TheEvilPhoenix> got any lucid servers you can test on?
[21:49] <TheEvilPhoenix> :/
[21:51] <hallyn> Daviey, uh, i think you caught me weirdly between switching clients from laptop to phone.  I only got the bits that got fwded to my gchat (i.e. 'hallyn:')
[21:51] <hallyn> Daviey, his template still doesn't give me a container that i can actually start up
[21:51] <hallyn> so i'm not in a hurry
[21:52] <kirkland> hallyn: you can forward irc to gchat?
[21:52]  * kirkland is intrigued
[21:52] <hallyn> heh, i blogged the script awhile ago
[21:52] <kirkland> wowsers
[21:52] <Daviey> hallyn: I started all 3 lines with hallyn:
[21:52] <tjaalton> kirkland: damn, too slow :)
[21:53] <hallyn> kirkland, http://s3hh.wordpress.com/2011/04/29/irc-pms-to-google-chat/
[21:53] <kirkland> tjaalton: quick, ls -alF /dev/shm /target/dev/shm
[21:53] <hallyn> Daviey, ok then i got them all :)
[21:53] <allegrem_> ok I finally managed to configure Samba !! Is anyone interested in the solution or can I go to bed ?
[21:53] <kirkland> neat
[21:53] <tjaalton> kirkland: empty already
[21:53] <hallyn> Daviey, you see creating fedora containers as a high importance thing in oneiric?
[21:53] <kirkland> tjaalton: rebooted?
[21:54] <hallyn> i mean, i'm happy to spend some time on making it work, but it seems hard to justify
[21:54] <hallyn> justify the time vs. other obligations, that is
[21:54] <Daviey> hallyn: Something that should work, it would be good if Ubuntu was a good host for everything :)
[21:54] <hallyn> yes it would
[21:54] <tjaalton> kirkland: no, it did the ecryptfs-setup dance already, before I managed to save the file
[21:55] <kirkland> tjaalton: okay, and /target/$HOME/.ecryptfs/* ?
[21:55] <kirkland> tjaalton: did wrapped-passphrase make it in there this time
[21:55] <hallyn> Daviey, there's also an untested prolly broken suse template :)
[21:55] <tjaalton> kirkland: same as before
[21:55] <kirkland> tjaalton: okay, mind doing one more install and hotpatching that bit in?
[21:55] <tjaalton> I'll hack in a hook to stop the installation before finish-install.d/06user-setup
[21:55] <tjaalton> that should do it
[21:55] <kirkland> tjaalton: cool
[21:55] <Daviey> kirkland: fwiw, forwarding irc messages to your phone is a dangerous game.
[21:55] <kirkland> tjaalton: thanks for your help
[21:56] <RoyK> damn - setup a winxp VM under KVM on this box just to have a windoze machine available if needed - so far, it's used >3 hours for updating....
[21:56] <Daviey> kirkland: 3:00am messages without context saying "<smoser> Daviey: It's in a REALLY bad way" .. close to release.. makes you pay attention, even if in bed.
[21:56] <kirkland> Daviey: LoL
[21:57] <kirkland> Daviey: good point
[21:57] <Daviey> I need to backlist smoser for this reason
[21:57] <RoyK> snf I'm on a 60Mbps link - windows sucks so hard I don't beleive it.....
[22:09] <kirkland> tjaalton: any luck?
[22:12] <tjaalton> kirkland: no :/
[22:12] <tjaalton> I'll check the log
[22:12] <kirkland> tjaalton: okay, hang on ... don't reboot that installation
[22:16] <kirkland> tjaalton: okay, i gotta run
[22:16] <kirkland> tjaalton: please file a bug on this
[22:16] <Daviey> jamespage: tomorrow, would you be able to sniff bug 881504 please?
[22:16] <tjaalton> kirkland: sure thing
[22:16] <kirkland> tjaalton: attach an install's /var/log/syslog, ideally with debconf debugging on
[22:16] <kirkland> tjaalton: also maybe set -x in the ecryptfs-setup-private
[22:16] <kirkland> tjaalton: and grab that in the log too
[22:17] <kirkland> tjaalton: something is going on around the creation of that wrapped-passphrase
[22:17] <kirkland> tjaalton: we have a bad conditional around that, somewhere
[22:17] <kirkland> tjaalton: i think the stty error is just noise
[22:17] <kirkland> tjaalton: as that script is not set -e
[22:18] <tjaalton> kirkland: yeah, I'll do another installation and attach the logs, thanks
[22:19] <kirkland> tjaalton: cheers mate