[00:15] <phunyguy> lickalott: nobody said you were trying to be malicious.... these are friendly reminders.  :)
[00:19] <lickalott> copy
[00:19] <phunyguy> so I have a question for the server masses.  What would be the best way for 5-10 users, on about 20 machines to sync user data across, so that I can manage users and groups on server, and they appear the same on all machines?
[00:20] <phunyguy> I saw the OpenLDAP method, but was curious if there was something simpler.
[00:20] <lickalott> was gonna say LDAP
[00:20] <lickalott> prob the easiest.  Don't think they make a command line version of Active Directory....lol
[00:21] <phunyguy> yeah, I am just worried about user and group management in LDAP itself.
[00:21] <phunyguy> seems pretty difficult.
[00:22] <lickalott> only other way would be to share out the files/folders that you want them to access (i.e. nfs, smb, etc...)
[00:23] <phunyguy> well that's not what I am worried about.
[00:23] <lickalott> but if they change their password on their machine you'd have to change it on your main server for authentication
[00:23] <lickalott> what is your end game?
[00:23] <phunyguy> I just want all machines to reflect the same UID/GID as everything else.
[00:24] <phunyguy> I could do it all manually, but any time something changes, it has to be done on all of them.
[00:24] <lickalott> yeah
[00:24] <lickalott> lemme look around a bit
[00:25] <sarnold> back in the day we used to rdist /etc/shadow, /etc/passwd, /etc/groups .. probably ldap is less bad. :)
[00:25] <phunyguy> sarnold: exactly, especially since service accounts could potentially use different UID/GID
[00:25] <sarnold> phunyguy: ooof, that's probably a much larger issue today than it was way back then, too :)
[00:25] <phunyguy> indeed.
[00:26] <lickalott> http://www.linux.com/community/forums/system-management/active-directory-alternative/13905#p13905
[00:26] <phunyguy> NIS
[00:26] <phunyguy> that was it
[00:26] <lickalott> NIS or NIS+ may work for you
[00:26] <phunyguy> I used that in the past.
[00:26] <lickalott> beat me
[00:26] <phunyguy> thanks
[00:26] <phunyguy> I saw it in the link you posted :)
[00:27] <lickalott> look into CUPS for print servers as well (if you need it)
[00:27] <lickalott> phunyguy, I got one for you
[00:27] <lickalott> ever installed clonezilla to an hdd?
[00:27] <phunyguy> nope
[00:28] <phunyguy> so you are looking for a backup solution, eh?
[00:28] <lickalott> what do you use to back up your stuff?
[00:28] <phunyguy> I have a few tricks up my sleeve.
[00:28] <phunyguy> first question, what are you backing up?
[00:28] <lickalott> yeah.  I've tried dd, FSarchiver (can't configure because it pukes at teh blkid portion), and clonezilla on a usb drive it a pita because I don't want to have to reboot every fri.
[00:29] <phunyguy> so you want to back up your entire system?
[00:29] <lickalott> I have a script (cron job) that backs up my profile, but i just rebuilt and wanted a fresh image in case the server goes down again
[00:29] <lickalott> true
[00:29] <phunyguy> does it need to be CLI, or will a GUI app work?
[00:29] <lickalott> i'm running server 12.04
[00:29] <phunyguy> hrm...
[00:29] <phunyguy> duplicity is one.
[00:29] <phunyguy> if you have a GUI, you can do backintime
[00:30] <phunyguy> and an extreme centralized solution, you can use BackupPC
[00:30] <lickalott> sarnold mentioned a few.  i still need to sit down and research some.
[00:30] <phunyguy> BackupPC is rsync-based, and does incrementals.
[00:30] <lickalott> backup PC = linux?
[00:30] <lickalott> ahh
[00:30] <phunyguy> yeah, it requires a webserver and it is perl based
[00:30] <phunyguy> basically it goes out to your machines, and rsyncs them
[00:31] <phunyguy> so you put in what machines you want it to back up and it does it on the schedule you specify.  There is a bit of a learning curve though.
[00:31] <phunyguy> backintime is the same type of thing, but has a GUI, and is client to server, instead of server to client.
[00:37] <TheLordOfTime> for Apache2, where does it access its mime.types?
[00:37] <TheLordOfTime> by default.
[00:37] <TheLordOfTime> does it use /etc/mime.types or no?
[00:45] <sarnold> hallyn_,stgraber, funny "iinuxcontainers.org" on https://github.com/lxc
[00:48] <stgraber> sarnold: fixed
[00:48] <sarnold> stgraber: thanks :)
[00:48] <sarnold> hallyn_: handled ;)
[00:49] <stgraber> sarnold: I'm still amazed we get over 600 visitors/day on linuxcontainers.org considering how little there's to see :)
[00:55] <sarnold> stgraber: nice! :) I suspect people are looking for a five-minute-guide to using lxc to see if it'll solve their problems..
[00:56] <sarnold> stgraber: (at least, I've considered that it'd be nice to have full-featured containers of all our supported releases to have full apt-file caches ready -- my schroots have nothing in them, since they're used for sbuild...
[00:57] <sarnold> .. so I went to linuxcontainers.org in part to see a five-minute guide :) hehe
[00:59] <stgraber> sarnold: haha, yeah, I usually have one container per supported release and per architecture (though armel/armhf using qemu-user-static isn't nearly as reliable as I'd like...)
[02:15] <phunyguy> does anyone know if it is possible to have NIS cache login info for when laptops are not able to see the network? (for example, at boot)
[02:17] <sarnold> phunyguy: investigate https://fedorahosted.org/sssd/
[02:17] <phunyguy> oooh
[02:17] <phunyguy> ok
[02:17] <phunyguy> thanks
[02:18] <phunyguy> sarnold: are you sure that is not just for ldap?
[02:26] <sarnold> phunyguy: hrm. It might be. :/ I had the vague impression from somewhere that it was "more" but never looked into it much myself
[02:26] <phunyguy> yeah... I got all excited with NIS, until I realized that it doesn't do any caching
[02:27] <sarnold> probably in part because NIS was designed to be deployed alongside NFS; if one is down there's no point trying to use the other :)
[02:29] <phunyguy> there has to be some solution that will work for me :(
[02:29] <phunyguy> I mean I can do LDAP, but I am not sure how hard it will be to manage
[02:30] <phunyguy> about ready to just say screw it and do manual user/group membership
[02:30] <sarnold> phunyguy: if you do go that route, look into what chef, puppet, and ansible can do for you
[02:30] <phunyguy> puppet looks promising....
[02:39] <phunyguy> My brain hurts.
[05:13] <babinlonston> Any one there to help me ? facing issue with  vsftpd using ssl  , Error:	GnuTLS error -12: A TLS fatal alert has been received. ,Error:	Could not connect to server, This is the Error I'm Facing
[05:32] <sarnold> babinlonston: you might want to try connecting to the server using openssl's s_client program, it supports e.g. -tls1 -tls1_1 -tls1_2 -ssl3 and so forth switches..
[05:34] <babinlonston> sarnold: now i'm using filezilla i'm facing this issue in filezilla, even i have tried winscp too
[09:14] <yolanda> jamespage, about adding ubuntu headers in openstack projects, shall i create a blueprint here? https://blueprints.launchpad.net/oslo/
[09:15] <jamespage> yolanda, can't we just do it under a bug? or was a blueprint requested?
[09:15] <yolanda> jamespage, well, they told me to create a custom middleware under oslo
[09:15] <yolanda> i think that's the process, right?
[09:15] <jamespage> yolanda, probably - this is a better question for openstack-dev tbg
[09:15] <jamespage> tbh
[09:16] <yolanda> i'm afraid that this takes lots of time to be approved
[09:17] <lifeless> Ubuntu headers?
[09:17] <yolanda> lifeless, welll, not ubuntu, but distribution headers
[09:17] <yolanda> talking about that: https://review.openstack.org/#/c/61128/
[09:17] <lifeless> so, lets choose a channel and stick to it ;) -> #openstack-dev :)
[09:17] <yolanda> yes, raised question there
[09:32] <Darkstar1_> Morning all
[10:54] <Darkstar1_> How can I find login logs going back a few months?
[10:55] <Darkstar1_> last only carries me as far as the last 10
[12:00] <yolanda> jamespage, did an initial approach for active/active rabbitmq: https://code.launchpad.net/~yolanda.robla/charms/precise/cinder/ha/+merge/199271, https://code.launchpad.net/~yolanda.robla/charms/precise/rabbitmq-server/ha/+merge/198768
[12:03] <jamespage> yolanda, looks like a good start - how does it test?
[12:03] <jamespage> can you rip out rabbit nodes and still manage the deployment?
[12:03] <yolanda> jamespage, i looked at cinder.conf, verified that rabbit_hosts were there, and refreshed when adding new unit
[12:04] <jamespage> yolanda, i'd probably do a bit more testing than that
[12:04] <yolanda> and checked in logfile that cinder connects correctly - it's always grabbing first node anyway
[12:04] <jamespage> to ensure that its does actually fail over as well
[12:04] <yolanda> jamespage, sure, what do you suggest?
[12:04] <jamespage> so poweroff the first node
[12:04] <jamespage> and see if cinder switches OK
[12:04] <yolanda> ok
[12:07] <yolanda> jamespage, magic! http://paste.ubuntu.com/6588536/
[12:07] <jamespage> yolanda, nice
[12:08] <jamespage> yolanda, do we need to land a charmhelpers branch first?
[12:08] <yolanda> according to doc, there are some config vars to define the delay, retries, etc... do you think it makes sense to add them on cinder charm and in config?
[12:08] <yolanda> jamespage, and yes, it was just a sample on cinder, but I need to update charmhelpers
[12:08] <jamespage> yolanda, I'd hope that there are sane defaults for that sort of stuff
[12:09] <jamespage> yolanda, OK - propose the charm-helpers merge and then re-sync
[12:09] <yolanda> yes, they are defaults
[12:09] <jamespage> yolanda, but looks good
[12:09] <jamespage> nice work btw
[12:09] <jamespage> adam_g, active/active rabbitmq with cinder ^^
[12:09] <yolanda> jamespage, it was partially done on rabbitmq, but i found that it was always sending rabbitmq_hosts var. I just send it now if vip is not present
[12:10] <jamespage> yolanda, makes sense
[12:11] <yolanda> great, i'll add it into charmhelpers, and then it's as easy as to refresh templates in other charms: ceilometer, nova-compute and neutron. The others don't support it
[12:15] <jamespage> yolanda, do we not need to set some of the bit here:
[12:15] <jamespage> http://docs.openstack.org/high-availability-guide/content/_configure_openstack_services_to_use_rabbitmq.html
[12:15] <jamespage> durable queues and ha policy?
[12:15] <jamespage> for 12.04 users they still get old rabbitmq
[12:15] <jamespage> pre 3.0
[12:17] <yolanda> jamespage, i'm not sure about it. I added the HA policy flag to rabbit>=3.0 , but this means that rabbit>=3.0 needs rabbit_ha_queues=true added?
[12:18] <yolanda> and for rabbitmq_durable_queues, i think it doesn't make any different between pre 3.0 and post 3.0 , it's just another config setting
[12:20] <jamespage> yolanda, hmm - I just pulled your branch into my environment
[12:20] <jamespage> after killing one of the nodes, and then trying to create a volume it sticks in creating
[12:20] <yolanda> it doesn't connect to new node?
[12:20] <jamespage> the scheduler and volume services have not noticed that rabbitmq #1 has disappeared
[12:20] <jamespage> yolanda, the api service has as its tried to use it
[12:20] <jamespage> and noticed
[12:21] <yolanda> i did the test and it reconnected myself, which version are you using?
[12:21] <jamespage> but the scheduler and volume service as just waiting for notifications from the powered off
[12:21] <jamespage> yolanda, grizzly
[12:21] <jamespage> yolanda, I see the multiple node config in the cinder charm OK
[12:21] <yolanda> i'm with cloud:precise-havana/proposed
[12:21] <yolanda> let me try with grizzly then
[12:22] <jamespage> yolanda, have all the cinder services noticed the drop?
[12:22] <yolanda> yes
[12:22] <jamespage> yolanda, well it might be that in grizzly the messaging layer is not as good
[12:22] <yolanda> jamespage, well, in cinder-api log i don't see any try to connect to rabbit
[12:23] <jamespage> yolanda, can you create volumes?
[12:23] <yolanda> that is only logged for scheduler and volume in my case
[12:23] <yolanda> let me try to create some
[12:24] <yolanda> i haven't related with keystone, i'll deploy full environment and try
[12:26] <jamespage> yolanda, +1 this needs a good test to validate it works how we think it does
[12:26] <yolanda> sure
[12:26] <yolanda> first wanted to check the approach
[12:27] <jamespage> I know adam_g had some concerns re subscribers noticing rabbitmq had gone away
[12:33] <jamespage> zul, can you make sure the lab-local changes for openstack-ubuntu-testing get back to branch on LP please
[12:38] <jamespage> yolanda, hmm - I'm not seeing the scheduler and volume services notice the failed rabbitmq node
[12:39] <yolanda> i'm deploying with grizzly now to check it
[12:39] <yolanda> on havana worked perfectly
[12:39] <yolanda> maybe it needs some config settings?
[12:39] <jamespage> yolanda, that is on havana - I re-deployed
[12:39] <jamespage> yolanda, how did you turn off your rabbitmq-server node
[12:39] <jamespage> I just powered mine done
[12:40] <yolanda> jamespage, it stopped service
[12:40] <yolanda> I did a rabbitmq-server stop
[12:41] <jamespage> yolanda, well that's not quite the same thing - that probably correctly terminates client connections
[12:41] <jamespage> in a failure that is not guaranteed
[12:41] <yolanda> jamespage, what did you do to test it?
[12:41] <jamespage> yolanda, I'm testing on openstack - so I did a 'nova stop <uuid>' on the first unit
[12:41] <yolanda> jamespage, and rabbit was down?
[12:42] <jamespage> effectively making it disappear like it was unplugged
[12:42] <yolanda> ok, i'll try in the same way
[13:10] <zul> jamespage:  ack
[13:15] <yolanda> jamespage, are you following headers discussion on openstack-dev mailing list? what do you think about it?
[13:15] <sda> hi all, I would like to be able to connect to my virtual machine desktop with another device (another desktop, laptop, tablet) and be able to choose what resolution use, and be able to connect multiple user at the same time. VNC is not good for do that, any idea?
[13:48] <plm> Hi all
[13:49] <plm> people, my sata HD on my server show in dmesg error reading and timeout to reading. I tryed smartctl and I have segmentation fault. well, my hd I think has badblocks. I need to do a entire copy of it. What is better/secure solution?
[13:50] <plm> using dd?
[13:50] <jamespage> yolanda, I am - can't say I'm surprised
[13:51] <yolanda> jamespage, yes, this feedback is the one i expected when i filed the bug
[13:51] <plm> or clonezilla?
[13:51] <jamespage> as a system admin the first thing I would do is endeavour to disclose as little information as possible about what my deployment is running on
[13:52] <jamespage> plm, dd is probably OK_ but you will probably have issues reading the backblocks so it might not succeed
[13:52] <yolanda> jamespage, and the --version call doesn't help us for this case
[13:52] <jamespage> yolanda, nope
[13:53] <plm> jamespage: and the clonezilla? I cant lost data
[13:53] <yolanda> so do you think is better to keep that as patches? or just discard the feature for openstack?
[13:53] <jamespage> plm, never used clonezilla
[13:53] <jamespage> yolanda, I'm concerned about patch cost
[13:53] <yolanda> maintenance...
[13:53] <jamespage> most patches need refreshing from time-to-time especially with the CI approach we take with openstack packages
[13:54] <yolanda> it could be a bit hard to maintain, as it needs to be on all the components
[13:54]  * jamespage thinks about this a bit more
[13:54] <plm> jamespage: what you recomend?
[13:55] <jamespage> plm, tbh the only time I ever had data important enough to have to recover a bad disk, I sent it to a specialized company todo that.
[13:55] <jamespage> its notoriously difficult to get it right
[13:55] <jamespage> plm, sorry - no backups I guess?
[14:02] <plm> jamespage: yes, has backup, but it old..
[14:02] <plm> *is
[14:07] <pmatulis> for dm-multipath, is there a standard method of being notified of a path failure?
[14:07] <jamespage> yolanda, I don't think you will get this upstream
[14:17] <yolanda> jamespage,yes, i don't expect it
[14:21] <jamespage> yolanda, still thinking about it
[14:24] <jamespage> yolanda, blimey I'm split on this one
[14:24] <yolanda> jamespage, i'm thinking if adding this feature will discourage people using our packages?
[14:28] <jamespage> yolanda, I think so long as its minimal overhead and configurable no
[14:28] <jamespage> think Apache ServerToken/ServerSignature
[14:29] <jamespage> only displays on server generated error pages, completely configurable in terms of disabling it
[14:29] <yolanda> jamespage, for our case, it's just simple to set a None in the key
[14:30] <yolanda> but shipping that as patches will mean to maintain a patch forever
[15:06] <jamespage> zul, there is still something wonky with the precise chroots in the lab
[15:06] <jamespage> swift is failing todo semaphore locking in its unit tests
[15:07] <zul> jamespage:  yeah ill take a look
[15:08] <zul> jamespage:  its next on my list
[15:08] <jamespage> zul, goody
[15:08] <jamespage> I fixed that python-jsonpatch issue btw
[15:08] <jamespage> zigo has a fix from me :-)
[15:08] <zul> jamespage:  goody
[15:09] <zul> just packaging  savanna yippe skipped
[15:25] <jamespage> adam_g, if you have time the ceilometer charm branches are ready for re-review - I made the changes we discussed in the agent charm to selectively pick data from the relation
[15:44] <jamespage> zul, can I get a +1 on python-jsonpath-rw for backporting for icehouse please - ceilometer needs it for trunk
[15:44] <jamespage> its in universe so we will need to MIR as well I guess
[15:45] <zul> jamespage:  +1
[15:46] <jamespage> zul, ta
[15:49] <xevwork> I've created a preseed file that I'm using with 12.04.3 and I'm trying to add RAID configuration to it. I just want an extremely basic RAID setup: root and swap mirrored on sda and sdb. During installation, it successfully creates md0, but fails while creating md1. It's creating swap partitions that I'm not asking for and they're interfering with the creation of the second RAID device.
[15:49] <xevwork> Does anyone have any tips for getting preseeded RAID to work?
[16:01] <smoser> rbasak, ping for ubuntu-meeting
[16:07] <zul> Daviey:  can you have a look at python-taskflow its blocking the opentack-ci stuff
[16:13] <Daviey> zul: ok
[16:28] <Daviey> zul: Accepted.. python-babel is depended on twice and there is an extra leading space on the last line of d/control.. but nothing to block on :)
[16:29] <zul> Daviey:  cool thanks
[16:34] <xevwork> I'm trying to preseed a RAID configuration. When the installer partitions the drives, it successfully creates the first array but fails on the second. If I check the logs, it says that it couldn't write the array's metadata. Looking further into it, there are swap partitions that I did not ask for!
[16:34] <xevwork> Here's my configuration: http://pastebin.com/MmfgzBE5
[16:34] <xevwork> It's creating /dev/sda5 and /dev/sdb5 as swap partitions and activating them.
[16:37] <jrwren> i'm using nova-compute, havana, from cloud archive, following the guide and having an issue when I start an instance. the libvirt driver creates the disk image, but its owned by libvirt-qemu, then nova-compute tries to use libguestfs to inject the keys and it fails because nova-compute is running as nova user and can't write to the file.
[16:50] <jrwren> nevermind. wrong perms on /boot/vmlinuz*
[17:03] <sm_> Hello, Is there somebody that does know even something about mail-stack-delivery?
[17:06] <wiehan> How does one set up a system wide proxy (something like https://www.privateinternetaccess.com/) on ubuntu server 12.04.3 through cli?
[17:08] <patdk-wk_> atleast 50 different ways
[17:09] <wiehan> patdk-wk_, simplest would be?
[17:09] <patdk-wk_> way too hard, if your asking
[17:10] <patdk-wk_> first you need the machine your protecting
[17:10] <wiehan> easiest then?
[17:10] <patdk-wk_> then you need a machine you buy, somewhere on the internet
[17:11] <wiehan> it's my 12.04.3 torrent box in the garage. have access to it via ssh
[17:11] <patdk-wk_> what good is that?
[17:11] <wiehan> patdk-wk_, can track down a paid for server easily
[17:11] <patdk-wk_> it's still in your house
[17:12] <wiehan> the idea is to still protect it (make it invisible) to torrent police or whatnot
[17:12] <sm_> Something like PeerBlock?
[17:13] <keithzg> Is anyone around using Mosh? I've loved it so far but I'm kindof missing the MOTD info (you know, with current loads, if there are pending updates, etc), which doesn't get displayed when logging in via Mosh.  Does anyone know how I might enable that?
[17:13] <patdk-wk_> it is so much easier to just not *download* *illegal* things
[17:13] <wiehan> patdk-wk_, thank you captain obvious
[17:13] <wiehan> sm_, I guess, I'll have to investigate peerblock?
[17:13] <sm_> wiehan, Are you trying to do something like PeerBlock, but for Ubuntu Srv?
[17:14] <wiehan> sm_, I just want my IP hidden when I download stuff. We just had a landmark case in our country
[17:15] <wiehan> What I use it for actually is quite irrelevant, as no one wants them to be snooped on
[17:15] <patdk-wk_> what you use it for is ver relevent, to if I bother to help :)
[17:16] <sm_> wiehan, What about torrenting trough VPN?
[17:17] <keithzg> Hrmm. I see in Mosh's changelog for 1.1.94-1 (which was many versions ago, currently 1.2.4a) that it  "Now prints message of the day (motd)". Is Ubuntu's MOTD somehow non-standard or otherwise wonky then, or is it the client-side implementation of Mosh that's at fault? Hmm.
[17:17] <wiehan> sm_, I have the server box set up already with a lot invested in it. I run a transmission torrent server and can access it through any device in the house, I also run a media server on the same machine. It was a big investment
[17:18] <wiehan> sm_, that works great
[17:18] <sm_> wiehan, May I ask how much did you spend on it?
[17:18] <sm_> In Euros please.
[17:18] <wiehan> sm_, I would just like to add a little anonymity  to it
[17:19] <wiehan> sm_, well, the project itself was a lot of fun to do. I have a server cabinet set up with fans and a micro atx server build that I did from scratch with PoE switches and an adsl router and and and... I guess with my IP camera investment - around 1000 euro
[17:20] <sm_> wiehan, That's fine. I was just curious. I dont really thing that there are really any other solusions than VPM or something else. That's a big investment indeed.
[17:20] <sm_> *Think
[17:21] <wiehan> sm_, I'll pay for a vpn service gladly. I just want to know how to set it up on ubuntu server
[17:21] <sm_> You are using Transmission?
[17:21] <wiehan> sm_, I mean something like these http://thebestproxyserver.com/
[17:21] <wiehan> sm_, Transmission indeed
[17:23] <sm_> Okey. You might be able to set it up with this guide: http://www.cyberciti.biz/tips/howto-configure-ubuntu-fedora-linux-pptp-client.html
[17:24] <wiehan> sm_, ty will investigate
[17:25] <sm_> wiehan, This might be useful 2: https://forum.transmissionbt.com/viewtopic.php?f=1&t=10732
[18:45] <sda> Hi all! I would like to access my desktop from another device. I would also like to be able to choose the resolution at connection like RDP. Also I would like to be able to connect with different users. Do you have any Idea? VNC doesn't seems fit my need!
[18:49] <TheLordOfTime> sda, "connect with different users" is gonna be hard without something like TeamViewer...
[18:49] <TheLordOfTime> and that's not a good solution for what you need though
[18:51] <sda> TheLordOfTime, Truth is I have a server, I would like to start the server with CLI and then ask the users to connect to it to have a full desktop enviroment
[18:52] <TheLordOfTime> ahh...
[18:54] <sarnold> sda: I believe that's one of the features of e.g. http://www.edubuntu.org/documentation/ltsp-live
[19:35] <keithzg> Hmphh. Even new text files I upload seem to mostly get placed as unknown files if they don't have a file extension. That's quite annoying.
[19:39] <keithzg> Oops, wrong channel, although in fairness my owncloud instance IS running on ubuntu server ;)
[19:40] <keithzg> Here, though, I'm still wondering how to make the damned MOTD show up when logging in via Mosh instead of SSH.
[19:41] <sw_> hi. how can I set up on my Ubuntu server that whenever some user logs in via SSH the history is cleared when they exit SSH? I want it to be for all users
[19:41] <sarnold> keithzg: hrm, moshing to one of my systems shows the usual "N packages can be upgraded..." message
[19:42] <keithzg> sarnold: Hrmm. Odd. I've even tried it now with two 13.10 servers, both running Mosh 1.2.4a, but no luck, no upgrade/capacity/load messages, just straight to a prompt line.
[19:42] <sarnold> sw_: check out the HISTFILE environment variable in the bash(1) manpage
[19:43] <pmatulis> sw_: that's a shell-specific question.  if bash, study /etc/bash.bash.logout
[19:43] <sw_> sarnold: I seen that actually, so set it to some place it'll be lost? also where should this be set to be picked up for all users?
[19:45] <vlad_starkov> Question: I'm choosing SSD for Ubuntu+KVM host OS. The guests will be on HDD RAID 10. What capacity of SSD will be enough?
[19:45] <sarnold> sw_: yeah, I'd aim it at /dev/null -- it's funny, the manpage says to unset the variable to not save the history, but it's hard to place an "unset" in the right place to keep it unset... well, I guess if a user's ~/.bashrc or ~/.bash_profile over-writes the value, it'll over-write the value...
[19:45] <keithzg> sw_: If you set HISTFILESIZE=0 in /etc/bash.bashrc it should work for all users assuming you're okay with users being able to override that if they want.
[19:47] <sarnold> vlad_starkov: I've got a pandaboard "server" (though it has a full desktop install on it) that takes 26 gigabytes. If you go for 64 gigabytes, that'll give you some overhead space for e.g. a squid-deb-proxy, some swap, and some free space to prevent excessive writing
[19:48] <vlad_starkov> sarnold: the server has 16Gb RAM. So what size the swap partition should be?
[19:49] <sarnold> vlad_starkov: my laptop's got 16 gigs RAM, I sometimes have ten VMs running at once (768 M ram in the VMs) and I don't think I've seen more than one gigabyte of swap use
[19:50] <vlad_starkov> sarnold: I see
[19:50] <sarnold> vlad_starkov: (I did go with 16 gigs swap on my laptop, primarily so I could use hibernate if I wanted.)
[19:51] <sw_> keithzg: I have an /etc/bashrc, no /etc/bash.bashrc, is that the same?
[19:51] <vlad_starkov> sarnold: how many free space should I reserve on SSD? Should I partition it somehow or just leave unpartitioned?
[19:52] <sarnold> vlad_starkov: I like having a separate /home partition but if this machine won't have any "real" local users, you can probably skip it
[19:52] <keithzg> sw_: Yeah, that should be the same, that's the older convention IIRC (I see bash.bashrc on a pair of newly configured 13.10 machines, but I remember it used to be just /etc./bashrc)
[19:53] <sw_> keithzg: thanks! so is it better to 'unset HISTFILE', or 'HISTFILE=/dev/null', or 'HISTFILESIZE=0'?
[19:53] <sw_> I guess they all do the same thing, but would one be more logical? :p
[19:56] <keithzg> sw_: Yeah, the result is pretty much the same. I'd gravitate towards the last two;  I know with HISTFILESIZE=0 the history will exist for the session and then be lost upon logging out, I assume the same is true for the /dev/null method but I don't know for sure.
[20:02] <keithzg> sarnold: Hmm, I don't see /etc/motd on the servers I'm trying to get a message via my Mosh login, whereabouts is that pulled from nowadays then?
[20:02] <vlad_starkov> sarnold: thanks
[20:03] <sarnold> keithzg: check out /etc/update-motd.d/
[20:06] <keithzg> Hrmm. Okay, and then it runs and writes to /var/run/motd for every login. Interesting.
[20:06] <keithzg> Apparently if I create /etc/motd manually it spits it out upon login via Mosh.
[20:07] <sarnold> keithzg: do you have the pam_motd module configured in /etc/pam,d/sshd ?
[20:07] <sarnold> /etc/pam.d/sshd
[20:19] <MrBill> Im looking to restore data from an old dat tape but don't know what program was used to back the data up. Can anyone here help me with this?
[20:20] <sarnold> MrBill: first thoughts: tar pax cpio lonetar (commercial program, dunno if you can still get it..)
[20:20] <sarnold> MrBill: perhaps amanda? it might use other tools to do the actual tape work..
[20:21] <MrBill> Alright ill take a look. Tar fails. that was my first guess. ill try the others you mentioned. I also tried to do jsut a raw dd dump of the tape so i wouldnt ahve to keep waiting around for the tape, but that kept throwing IO errors.
[20:22] <sarnold> ugh, yeah, I can imagine. It's been ages since I've done tape things, I don't know if I'd expect a 'dd' to work or not. I'd hope it would...
[20:23] <mdeslaur> MrBill: was the tape even written from a linux server?
[20:24] <MrBill> there was supposed to be a copy of some really interesting software on this particular tape, so im really hoping it is not bad.
[20:24] <vlad_starkov> About SSD: https://sites.google.com/site/easylinuxtipsproject/ssd#TOC-Limit-swap-wear-and-tame-the-inode-cache
[20:25] <MrBill> well, the main machine that ran the show at the business that this server was at was running some form of unix/linux, and there are linux style directories listed on it, so i would assume so.
[20:25] <mdeslaur> ok
[20:25] <mdeslaur> there's arkeia too
[20:25] <MrBill> listed on the label of the tape that is
[20:29] <sarnold> vlad_starkov: yikes, the advice on that page ranges all over the place -- using ahci is very nearly required, setting up fstrim in a cronjob is a good idea, noatime breaks some features where the default "relatime" drastically reduces atime updates but doesn't break features ..changing the swappiness will help some users and hurt other users, and disabling the browser caches is just insane.
[20:33] <vlad_starkov> sarnold: what about item "Ubuntu 12.04: change the scheduler to deadline" ?
[20:34] <sarnold> vlad_starkov: sounds decent enough.
[20:37] <vlad_starkov> sarnold: what Ubuntu-Server version do you recommend for KVM?
[20:38] <sarnold> vlad_starkov: 12.04 LTS -- it's the newest released LTS. :)
[20:38] <vlad_starkov> sarnold: OK)
[20:38] <sarnold> vlad_starkov: if you're looking to build something on 14.04 LTS when it comes out, then perhaps it'd be worth testing with trusty along the development path
[20:38] <vlad_starkov> ok
[20:38] <vlad_starkov> Thnks
[20:42] <vlad_starkov> A good explanation about TRIM feature for SSD http://www.youtube.com/watch?v=NnBk2IGYerU
[20:57] <keithzg> sarnold: Looks like it's set up in /etc/pam.d/sshd (http://pastebin.ubuntu.com/6590964/); and yeah, it works fine if I login via SSH, just not mosh.
[20:57] <sarnold> keithzg: odd, I thought sshd was responsible for setting up the mosh channel..
[20:58] <keithzg> sarnold: Well, it sets up the initial connection/handshake and starts the client-side mosh instance, but then it's mosh->mosh, at least as far as I understand.
[20:58] <keithzg> sarnold: I'm tempted to just symlink /var/run/motd to /etc/motd, heh. That should do the job, at least.
[20:59] <sarnold> keithzg: if you go that way it might be easiest to uninstall the dynamic motd package..
[21:00] <keithzg> sarnold: Oh, but in theory though, the dynamic motd will be running every time mosh does the ssh handshakey portion, no? And then mosh reads the new motd through the symlink, et voila.
[21:04] <keithzg> Or maybe I'm missing something and being overly optimistic, since it doesn't seem to be updating . . . even when I login via ssh. Hmm. Still info circa Dec 12.
[21:16] <vlad_starkov> Question: Is it fine choice for os SSD: http://ark.intel.com/products/74949/Intel-SSD-DC-S3500-Series-80GB-2_5in-SATA-6Gbs-20nm-MLC ?
[22:00] <sda> Hi all! I would like to create a remote session on my desktop from another device. I would also like to be able to choose the resolution reconnect. And be able to connect multiple different users.
[22:09] <TheLordOfTime> sda, that question on its own is vague.
[22:09] <TheLordOfTime> sda, [2013/12/17 13:51:01] <sda> TheLordOfTime, Truth is I have a server, I would like to start the server with CLI and then ask the users to connect to it to have a full desktop enviroment
[22:09] <TheLordOfTime> [2013/12/17 13:54:33] <sarnold> sda: I believe that's one of the features of e.g. http://www.edubuntu.org/documentation/ltsp-live
[22:10] <TheLordOfTime> sda, from about 3 hours ago
[22:12] <sda> TheLordOfTime,  yes, but it's not what i searched, i worked 3hours with it :) Truth is I would like something like RDP for windows. But xrdp on linux doesn't work as well as on Windows
[22:22] <soahccc> Can anybody give me some tips of how to pinpoint the cause of occasional lag on my server. I can't see anything with htop (delay 0.25) nor with iostat (iotop doesn't work unfortunately). Syslog doesn't show anything interesting and the raid is also operating normally. I don't know what to do know :(
[22:23] <soahccc> With lag I mean a webpage takes > 20 seconds to load, connection via SSH hangs, etc.
[22:24] <soahccc> The load is around 5 while the only thing I notice is an iowait of 2%
[22:54] <sarnold> soahccc: are you doing anything like logging hostnames in apache request logs?
[23:06] <wam> soahccc: vserver or bare metal?