[00:20] <markus27> Hi.  I'm running apache2 web server on ubuntu.  I'd like to ring a bell or flash a light or do some other action in the physical world every time a certain page is visited.  I'm sure this has been done before. Are there any existing devices that support this?
[00:29] <tsrk> markus27: you could install the "beep" package, and run beep every time the page is visisted
[00:50] <markus27> tsrk:  Thanks.  Might be a quick first step in that direction
[00:50] <rdw200169> markus27: yeah, you just gotta figure out the link between /var/log/apache2/access.log and the beep, and whatnot
[00:51] <rdw200169> markus27: there's also this: http://llg.cubic.org/led/
[02:20] <kieppie> hi guys. is anyone familiar with link aggregation or bonding?
[02:22] <twb> !anyone
[02:28] <kieppie> twb: k. I'm trying to make the best of a server w 2 NIC's & has a pretty hefty net load. basically a NAS. I'm looking for good, current guides on setting it up & what mode to use
[02:30] <Slyboots_> HEy; Perhaps someone can throw me a boe here; had to update Samaba to fix a bug (Compiled it outside the repo)
[02:30] <Slyboots_> but now I dodnt have a init.d script or that "service" thing that ubuntu likes to use
[02:30] <Slyboots_> So.. not sure how to start/stop at boot and whatnot
[02:31] <markus27> tsrk:  I'm not seeing a beep package in the list of mods - http://en.wikipedia.org/wiki/List_of_Apache_modules  Which one were you referring to?
[02:31] <tsrk> markus27: it's an ubuntu package
[02:31] <tsrk> markus27: you can call "beep" from the command line to make the computer beep, but you could also call it from PHP or somewhere else
[02:32] <markus27> Got it.
[02:34] <Slyboots_> Anyone any idea?
[02:34] <twb> kieppie: last time I looked, trunk like that was pretty trivial, as long as the other end of the trunk runs compatible gear (linux, cisco or sun).
[02:36] <kieppie> [
[02:37] <kieppie> twb: I'm not too sure re infrastructure @ this point, but it's a 1000 Mb/s connection (& 1 spare) serving about a dozen XP & W7 hosts doing some pretty intense I/O to SMB shared (CAD & Animation)
[02:38] <twb> OK, trunking only matters at the two endpoints of the trunked connection
[02:38] <kieppie> so I'm trying, at the moment, to figure out the best mode to use
[02:38] <twb> e.g. between your NAS host and the switch or router it's plugged into
[02:39] <twb> If your two gigE lines go from the NAS into a dumb switch, you can't have trunking
[02:40] <kieppie> I think it would be best to assume that scenario. are you referring only to the trunking, or to link aggregation too?
[02:40] <twb> (Unless you trunk further upstream, which is almost certainly useless, unless your inter-switch fabric is at least 2gbps)
[02:40] <twb> trunking = bonding = aggregation
[02:41] <twb> they all mean "combining multiple (link-layer) links into a single logical (link-layer) link"
[02:41] <kieppie> ok. thanks.
[02:42] <kieppie> I'm reading though some of the ubuntu docs, and they seem to indicate that there are modes that do not require anything special
[02:43] <kieppie> https://help.ubuntu.com/community/LinkAggregation#Bonding%20Modes
[02:43] <kieppie> " IEEE 802.3ad Standard Dynamic Link Aggregation (not required if using adaptive load balancing or active/backup). "
[02:43] <kieppie> so I'm trying to do the best with the kit I have on hand
[02:44] <twb> Ah, that rings a bell
[02:44] <twb> I admit I'm more familiar with the other way (802.1q) -- multiple logical links over a single physical link
[02:45] <kieppie> I'm not familiar with that. do you mean running multiple IP's over a single NIC?
[02:47] <kieppie> it looks like mode=6 (balance-alb) is the safest fall-back option if I'm unsure re kit
[02:52] <twb> Well, the specific example I have is a one-NIC server providing DHCP for several discrete networks
[02:52] <twb> It was a SFF host, so I couldn't put more NICs in it
[02:53] <Slyboots_> Hmm..
[02:57] <Slyboots_> This is driving me nuts
[02:57] <Slyboots_> Trying to find a deb file or a rep that I can install a newer version of Samba
[02:57] <twb> Slyboots_: why?
[02:58] <twb> And what version do you want?
[02:58] <Slyboots_> Because Windows 7 is crapping out on me and breaks Smbclient
[02:58] <Slyboots_> Need.. 3.5.6
[02:59] <Slyboots_> Something to do with Windows sign-in asssist or some crap breaks Samba; which is fixed in 3.5.6
[03:01] <Slyboots_> OIt exists in Natty Narwall.. but how to I install that on top of my M. Meercat
[03:01] <Slyboots_> .. I think Im on meercat..
[03:01] <Slyboots_> aye; 10.10
[03:01] <twb> http://paste.debian.net/102963/
[03:02] <Slyboots_> ... What?
[03:03] <Slyboots_> Im not sure what that link is.. supposed to mean
[03:04] <twb> It means the version you want is available in Ubuntu
[03:04] <twb> It just isn't available in maverick
[03:05] <Slyboots_> Yea.. But its in Narwall; but how do I install that into a Mavrick system wihtout you know.; breaking
[03:05] <twb> very very carefully
[03:06] <Slyboots_> :P
[03:06] <Slyboots_> This is why I hate linux sometimes :P
[03:07] <twb> Actually what you have to do is find or create a "bbackport"  For a simple package, creating a backport is mostly just recompiling the source package (apt-get --build source foo).  Likely samba is going to be too complicated for that.
[03:08] <twb> Slyboots_: if it wasn't linux, you'd be fucked, because all you'd be able to do is call up the vendor ask beg them to create backwards-compatible version.  At least here you have the option of employing someone to do so.
[03:09] <twb> e.g. if you wanted to run the latest IE on a really old version of Windows, the ONLY people that can make that happen is microsoft, so they have a monopoly on doing it
[03:09] <Slyboots_> Its not really the same thing
[03:09] <Slyboots_> i want the *next* revison :P
[03:09] <Slyboots_> I can build aind instlal samba from source but it falls to bits as I dont have any init.d or "service" scripts
[03:13] <Slyboots_> Headache :P
[03:17] <Slyboots_> no idea what to do baout this tbh
[03:17] <Slyboots_> sod it' will stick with the default version and think of a way around this..
[03:24] <twb> For something as hairy as samba, that would certainly be my recommendation
[03:24] <twb> (Well, I'm *assuming* Samba is hairy.)
[03:49] <carljm> Hi all. Is there any way I can the Ubuntu exim4-config package to regenerate all the split config files in /etc/exim4/conf.d/ as they would be on a fresh install?
[03:49] <carljm> dpkg-reconfigure doesn't do it, reinstalling doesn't do it...
[03:49] <twb> I guess you could purge and reinstall it
[03:50] <carljm> tried that
[03:50] <twb> You *purged* first?
[03:50] <carljm> just get a conf.d with several empty dirs
[03:50] <twb> Purged *all* the exim4 packages?
[03:50] <carljm> hmm, i may have just purged exim4-config
[03:50] <carljm> i'll try purging all of them.
[03:50] <carljm> thanks
[03:50] <twb> I don't know much about exim4, but that's what I'd try
[03:50] <twb> Or just create a new chroot (e.g. debootstrap) and install exim4 in there, then copy the config files out
[05:13] <Psi-Jack> Okay, so I have some sudden strange problems with ocfs2 under pacemaker. I have 2 nodes trying to use ocfs2, and it was working fine up until suddenly earlier today. Right now, 1 node can take the ocfs2 mount, and the second one fails. When I try to mount manually cause it does seem to load o2cb on the second node. Anyway, the error I get trying to mount:
[05:14] <Psi-Jack> mount.ocfs2: Configuration error discovered while trying to join the group
[05:14] <Psi-Jack> They're both using 10.04.1 with pacemaker and ocfs2 stuff from the clustering PPA.
[05:15] <Psi-Jack> Also the mount that pacemaker's trying to do, stays actively in the process list.
[05:15] <Psi-Jack>  1582 ?        D      0:00 /sbin/mount.ocfs2 /dev/vdc1 /home -o rw,acl
[05:16] <Psi-Jack>  2022 ?        D      0:00 blockdev --flushbufs /dev/vdc1
[05:16] <Psi-Jack> That too. Both D status.
[05:22] <twb> D means waiting for I/O
[05:23] <twb> Or just generally stuck in kernel-space
[05:23] <Psi-Jack> Right, uninteruptable sleep, I know that. ;)
[05:23] <twb> Sorry, I'm used to #ubuntu-* querents being duuuuumb
[05:23] <Psi-Jack> But that happens right after it tries to mount the ocfs2 mount after loading o2cb. ;)
[05:24] <Psi-Jack> Yeaaaah.. Understood.
[05:24] <Psi-Jack> That's why I'm here, not there. This is a server as well, so. ;)
[05:24] <twb> This is an #ubuntu-* channel, dude
[05:24] <twb> But whatever; I don't know OCFS2 so I can only triage.
[05:26] <twb> So I notice the Ubuntu Server Guide only describes NFSv3 for networking /home.
[05:27] <twb> That is SUPER sucky, because NFSv3 basically trusts anyone on the network anyone NET_RAW capabilities (i.e. everyone).
[05:29] <Psi-Jack> heh
[05:29] <twb> Hmm, on further reading it appears to be exporting both NFSv3 and (unauthenticated) NFSv4.
[05:29] <twb> At the client's choice
[05:29] <Psi-Jack> Yeah, and NFS sucks.
[05:29] <Psi-Jack> v3 and v4, especially for failover support.
[05:29] <twb> Psi-Jack: what do you suggest, CIFS or AFS?
[05:29] <twb> I am.... unimpressed by pam_mount
[05:30] <Psi-Jack> My /dev/vdc device is a multipath'd iSCSI device provided as a kvm virtio drive to the server.
[05:30] <twb> (Recall that this is for mounting users' $HOME.)
[05:30] <Psi-Jack> twb: Out of those? Maybe AFS.
[05:30] <twb> I'm open to other suggestions, I just haven't seen anything even THAT production-ready
[05:30] <twb> *seen anything else
[05:31] <Psi-Jack> But my /home is an ocfs2 shared iSCSI device. :p
[05:31] <Psi-Jack> twb: Well, really depends on what you need, and the overal goals. You can't just "choose" randomly what you want and hope it works for the best.
[05:35] <twb> I have a heterogeneous mix of mostly linux workstations, laptops and servers in a single-site LAN running over 100baseT on a good day.  I need users to be able to login (solved with LDAP/SSL) and I need them to be able to access both their home directory and a shared data directory, both stored on a central server.  The access should be subject to at least POSIX DACs, MACs (e.g. apparmor) are optional.  Advisory lock
[05:35] <twb> ing is sufficient.  Circumventing the DAC should be difficult, preferably non-trivial.  Site policy prevents me from dictating what software the users run, e.g. I cannot prevent users having root on their own machines.
[05:36] <twb> Whatever solution I deploy MUST work on 2.6.32, SHOULD work as far back as 2.6.18 (pref. 2.6.8), and if it works on legacy FC3 and SunOS, that's a mild bonus.
[05:37] <Psi-Jack> Hmmmm
[05:37] <Psi-Jack> NFSv3 then.
[05:39] <twb> My chief objection to NFSv3 is that if you have root on any client box, you have full access to the NFS fileshare
[05:40] <Psi-Jack> use rootsquash.
[05:40] <twb> root squash does not fix it.
[05:40] <twb> viz: su - fred, now you can edit all fred's file
[05:40] <twb> *files
[05:41] <Psi-Jack> Hmmm.. I see your problem. And there's not much you /can/ actually do about it.
[05:41] <Psi-Jack> besides policy changes.
[05:41] <twb> I think nfsv4+kerberos would fix it...
[05:41] <Psi-Jack> Or isolated mounts.
[05:41] <Psi-Jack> In other words, only mount their home directory, period.
[05:41] <twb> Psi-Jack: that's how cifs+pam_mount works
[05:42] <Psi-Jack> Yeah, but CIFS != POSIX.
[05:42] <Psi-Jack> You can do the same approach with NFS, automounting their home directory by NFS.
[05:42] <twb> THat doesn't stop root on a client running mount whenever they want
[05:42] <Psi-Jack> Course, they could still su to the other user and it'd happen the same way, but you can restrict what systems can mount by NFS export rules.
[05:43] <Psi-Jack> In other words, isolate the mounts by workstation IPs.
[05:44] <twb> Hum, that's reasonable
[05:44] <Psi-Jack> Yeah, that would only become a problem IF they need to be able to login from any system, and if that were the case, then you /need/ a policy to prevent full use of root access.
[05:50] <twb> Talking to the other sysadmins here, I think I have a viable mix of policy change and locking down the NFSv3 server (per IP) and the root on the clients
[05:52] <Psi-Jack> Yeah. if it's company equipment, locking down root is essential, especially if you want to comply with any kind of security policies, like PCI, SOX, etc.
[05:59] <twb> It's probably a bit different for my organization, which is an ad-hoc bag of rent-a-sysadmins
[05:59] <twb> So they're even worse than the engineering staff, who want root access so they can compile random shit into /usr/local :-)
[06:02] <Doonz> engineers should not have access to anything
[06:02] <Psi-Jack> heh, exactly.
[06:02] <twb> I know
[06:02] <Doonz> pen and paper ONLY!
[06:02] <twb> And an HP-48 of course
[06:02] <Doonz> even then make sure the pen isn't sharp
[07:23] <xampart> to what permissions i set my /var/www -directory (og = www-data) if i want another user to be able to create/remove directories/files under it?
[07:24] <xampart> should i add the user to group www-data also
[07:29] <blaenk> anyone here running nginx? I'd like to install it cause it's lighter than apache and supports phusion passenger, but I'm wondering how straightforward it would be to use PHP and Rails with it?
[07:31] <joschi> blaenk: you'll need to invest a little more time into its initial configuration. other than that, it's just a web server
[07:32] <blaenk> joschi: cool thanks, appreciated
[09:19] <twb> Has anybody successfully gotten slapo-ppolicy working in a cn=config-style openldap slapd?
[09:19] <twb> I *had* it working in 8.04, but this new-fangled crap is... not obvious.
[09:27] <BlackDex> Hello there..
[09:27] <BlackDex> i just did an upgrade of the packages
[09:28] <BlackDex> and now i get the following message
[09:28] <BlackDex> dpkg: error processing grub-pc (--configure):
[09:28] <BlackDex> subprocess installed post-installation script returned error exit status 1
[09:28] <BlackDex> Errors were encountered while processing:
[09:28] <BlackDex> grub-pc
[09:28] <BlackDex> E: Sub-process /usr/bin/dpkg returned an error code (1)
[09:29] <BlackDex> How can i fix this?
[09:30] <twb> dpkg --configure -a
[09:31] <twb> But the problem is further up than what you pasted
[09:31] <twb> It's probably something like grub-pc running os-prober, which tries and fails to read /dev/cdrom.
[09:31] <BlackDex> that is all i get now when i type "apt-get upgrade"
[09:31] <BlackDex> except for the message that 1 is not fully installed
[09:32] <BlackDex> sudo dpkg --configure -a
[09:32] <BlackDex> Setting up grub-pc (1.98-1ubuntu9) ...
[09:32] <BlackDex> dpkg: error processing grub-pc (--configure):
[09:32] <BlackDex>  subprocess installed post-installation script returned error exit status 1
[09:32] <BlackDex> Errors were encountered while processing:
[09:32] <BlackDex>  grub-pc
[09:32] <BlackDex> the same again
[09:32] <twb> !pastebin > BlackDex
[09:36] <BlackDex> The output of dpkg --configure -a > http://paste.ubuntu.com/546539/
[09:36] <BlackDex> sorry btw
[09:42] <yann2> hello - for KVM, what brings the best io performance for disks - qcow, or raw? would my disks maybe be faster if I converted my qcow disks into raw?
[09:42] <twb> It depends on your write profile
[09:43] <yann2> write profile?
[09:43] <twb> copy-on-write is a more expensive operation, but I think it'd pay of for non-sequential writes.
[09:43] <twb> yann2: like are you writing to /var/log or to a postgres database
[09:43] <yann2> I get very bad perfs with qcow2 when the disk needs to grow, which is the only way vmbuilder can create them, which is a bit unfortunate
[09:44] <twb> Uh, qcow2 size is decided up-front.  Unless by "growing" you're referring to population of what is effectively a sparse file.
[09:45] <twb> If you're doing raw to a *file*, that literally is a sparse file.
[09:45] <yann2> by growing I mean when a 50GB qcow2 file takes only a couple of GB in the beginning (the space effectively used) on the host
[09:46] <yann2> and yeah I was talking about raw to a file, instead of a qcow2 file
[09:46] <twb> yann2: well, they're both growing in the same way
[09:46] <twb> compare ls -hl foo.raw with du -h foo.raw
[09:46] <yann2> am using qcow2 right now
[09:47] <twb> My personal preference would be to assign raw LVs to VMs, rather than files
[09:47] <yann2> from what I remember you can create qcow2 images that grow, or with space preallocated (man qemu-img)
[09:47] <twb> Or entire disks and leverage intel VT-d, but I can't justify that expense
[09:48] <yann2> you mean, use LVM more?
[09:48] <twb> yann2: having the virtual disk as a file on another filesystem means you're paying the filesystem layer cost TWICE
[09:48] <yann2> I wonder if converting my qcow2 to raw and then back to qcow2 would be beneficial, as it'd allow preallocating
[09:49] <twb> If you just say kvm -hda /dev/mapper/foo-bar, then you aren't
[09:49] <yann2> LVM has a cost too
[09:49] <twb> Granted
[09:51] <yann2> I think I'll try converting to raw and back to qcow2 so that the qcow files dont grow anymore... I believe its the growing thats killing my perfs
[09:52] <twb> I believe you're silly, but whatever
[09:54] <yann2> really?
[09:54] <yann2> once I noticed that the qcow2 file was growing by slices of 2-3MB
[09:55] <yann2> and that my speed was good while copying a few MB, then suddenly dropping to something ridiculously low
[09:55] <yann2> and that just copying a GB file over and over and over, and then removing it (thus increasing the qcow2 file) was significantly improving write speed
[09:55] <yann2> but that was with an older version of kvm
[09:57] <dewey_> could someone help me with a upstart & path problem?
[09:58] <dewey_> I have a upstat script that's trying to start a ruby daemon, but the ruby file can't require some dependencies
[09:59] <masACC> Hm. Anyone have a feeling for when we can expect postgres 9 turning up? :)
[10:00] <twb> dewey_: fix the $PATH in your script/end script stanza?
[10:03] <dewey_> twb: doing that, but it probably has to do with rvm
[10:03] <dewey_> + I could do it wrong of course
[10:03] <twb> I guess that's the ruby interpreter?
[10:04] <dewey_> yes, it's an easy way to swtich between ruby versions
[10:05] <dewey_> twb: http://pastie.org/1397451
[10:05] <twb> "export HOME=/home/sysadmin".... nnnngh
[10:05] <dewey_> just to make sure...
[10:06] <twb> [[ is a bashism.  It is not valid in sh (which is what upstart uses... I hope.)
[10:06] <dewey_> ok
[10:06] <twb> But congrats, that's the ugliest upstart job I've ever seen
[10:06] <dewey_> newb as you can see, but trying :)
[10:06] <dewey_> haha
[10:06] <dewey_> well, because it isn't working
[10:06] <twb> "start on startup" is also very likely wrong; it should at least wait until you have a root filesystem
[10:07] <yann2> twb, can't blame the author, upstart documentation non existent as far as I remember
[10:07] <twb> To be honest you're probably safer sticking this shit in /etc/rc.local, which runs at the very end of the boot process
[10:07] <twb> yann2: granted, but I'm busy ranting about slapd right now so I don't want to get started on upstart... :P
[10:08] <dewey_> well ok, will google again
[10:10] <yann2> sysadvent published a nice article on upstart a couple of days ago, I promised myself to take time to read it http://sysadvent.blogspot.com/2010/12/day-19-upstart.html
[10:12] <twb> I get the impression that upstart's driving forces have moved to Google and/or switched to developing systemd, so I don't intend to invest any more time in it than I have to
[10:13] <yann2> oh? :( too bad, seemed promising
[10:15] <twb> yann2: go look at the systemd salespitch
[10:24] <qman__> IMO systemd is a FAR better approach
[10:24] <qman__> I would be very happy to see it replace upstart in ubuntu
[10:24] <twb> IMO *insserv* is a better approach than upstart
[10:25] <qman__> I can't disagree there
[10:25] <twb> Hell, I even prefer minit/cinit.  At least they don't mix two discrete DSLs in a single file like a flipping PHP user
[10:25] <qman__> upstart is faster than sysv, but it's just a pile of headaches and work
[10:25] <twb> qman__: is it faster than startparized sysvinit?  Because I doubt it's that much faster.
[10:26] <qman__> either way, it's not faster enough to justify the mess
[10:26] <twb> I mean, my laptop is sid and it boots in under five seconds, compared to the one second or so for my (empty) lucid servers.
[10:26] <twb> But the *BIOS* on the enterprise servers I deal with (which probably includes linux in the ILOM) takes like two minutes, so I don't care about an extra four seconds...
[10:26] <qman__> I like the idea of systemd because it's basically eliminating startup scripts
[10:27] <qman__> not completely, but it's a lot less involved and a lot more automated
[10:27] <twb> qman__: that's like saying computers will eliminate paper in the office
[10:30] <yann2> twb, what brand? on many of them you can disable some checks, which is sometimes useful
[10:31] <twb> I don't have one in front of me now, but I've had dells, HPs and IBMs do it recently
[10:32] <qman__> yeah
[10:32] <qman__> especially anything with several RAID controllers
[10:32] <qman__> I don't think I HAVE a BIOS that loads in under 5 seconds
[10:33] <qman__> except for my laptop, startup time is just not important to me
[10:33] <qman__> I would much rather have it work
[10:34] <qman__> of course, compared to windows, even 8.04 was fast
[10:34] <yann2> hum. using virtio with raw devices under kvm... fail :(
[10:35] <twb> qman__: OpenFirmware!
[10:36] <qman__> before I finally nuked it for good, my windows install was taking in excess of five minutes to boot, it was pathetic
[10:36] <qman__> never had any linux take that long
[10:36] <qman__> I think an extra five seconds or so is pretty insignificant
[10:39] <twb> I wouldn't know about that.  Last Windows I ran was NT5.0
[10:39] <qman__> you didn't miss anything
[10:40] <twb> Well, I missed out in hard links
[10:41] <qman__> I've hard linked on win2k before
[10:41] <qman__> php.ini workaround
[10:43] <Slyboots_> Hmm..
[10:44] <Slyboots_> Anyone any idea on a pratical solution to running steam on a headerless ubuntu server?
[10:44] <qman__> steam as in Valve's Steam?
[10:44] <Slyboots_> Not intrested in ctually playing games; just using the download manager of the client
[10:44] <Slyboots_> mmm
[10:44] <qman__> probably not going to happen without X
[10:45] <qman__> if it's server data you could use hl2ds/srcds
[10:45] <Slyboots_> Nah; want to run a copy on the server to make it simpler to backup games and download the huge ones overnight
[10:45] <Slyboots_> Prefer NOT to use X though; since its a waste of disk-space
[10:45] <qman__> well, for steam to work, you need wine
[10:45] <qman__> and steam doesn't have a command line interface either
[10:46] <qman__> it's going to need an X display
[10:46] <Slyboots_> heh; so Im crap out of luck
[10:46] <twb> Xvfb
[10:46] <Slyboots_> Could install X with a reall light WM I suppose I just use VNC
[10:46] <twb> Or Xvncserver4, if you need to connect to it
[10:46] <twb> Or, of course, just run X over the network
[10:47] <twb> Nor do you need a WM to use X, just start clients with the appropriate -geometry option.  Default behaviour is focus-follows-mouse.
[10:47] <Slyboots_> Mmm
[10:47] <twb> qman__: does steam even have a native Linux client, or does one just use wine and hope for the best?
[10:49] <Slyboots_> Wine only
[10:49] <twb> I wish the gaming industry would get a clue...
[10:50] <qman__> steam does not have a native linux client
[10:50] <qman__> but srcds/hl2ds do
[10:50] <qman__> there might be a way to get those to download regular game data
[10:50] <qman__> or there might be a separate tool you can use
[10:51] <qman__> I'd look into that before attempting to rig up an X just for steam
[10:58] <twb> srcds?
[10:59] <qman__> the dedicated server
[10:59] <qman__> it downloads the data the same way steam does, and runs on headless linux natively
[11:00] <twb> Is it FOSS?
[11:02] <qman__> not sure, I don't think so
[11:02] <qman__> it is free to use, but I think you can only get binaries
[11:05] <twb> I hear you can't even play single-player games without putting the windows box on the network
[11:06] <twb> (And allowing it egress connections to steam's server.)
[11:06] <qman__> yeah, it has to phone home at least once before the game will play
[11:06] <qman__> once it has, you can go in offline mode
[11:06] <twb> Once per game, or (say) once per game per day?
[11:07] <qman__> once per game, ever
[11:07] <twb> OK, that's not as bad as I was lead to believe
[11:07] <qman__> the only time it has ever really been a problem was at the launch of half-life 2, and their servers couldn't keep up
[11:08] <qman__> they learned quick after that one
[11:08] <Slyboots_> Aye; Steam is not to shoddy
[11:09] <yann2> qman__, starcraft 2's release wasnt too great either :)
[11:10] <qman__> YankDownUnder, yes, but that's blizzard, they go down after most of their patches, a game launch it's fully expected
[11:29] <Slyboots_> Hm..
[11:30] <Slyboots_> Dont they make a decent rsync client for windows?
[11:32] <twb> Slyboots_: cwrsync
[11:33] <twb> It's just the relevant bits of cygwin (libcygwhatever, rsync and ssh)
[11:43]  * RoyK just uses basic cygwin/rsync with windoze
[11:56] <Slyboots_> Mm..
[11:57] <Slyboots_> Right that doesnt actually seem to give me the rsync program :P
[12:04] <RoyK> what?
[12:04] <RoyK> cygwin comes with rsync
[12:05] <twb> cygwin provides a package picker thingy
[12:05] <twb> You might need to run that and say "gimme rsync"
[12:05] <RoyK> cygwin's setup.exe
[12:06] <Slyboots_> Using cysync
[12:06] <Slyboots_> Just keeps saying "rsync.exe not found"
[12:06] <twb> Slyboots_: maybe take it to ##windows
[12:07] <Slyboots_> aye
[12:07] <RoyK> Slyboots_: just install cygwin with rsync and ssh - that works
[12:22] <dewey_> does someone know why upstart gives the following error: [21595]: Error: Could not execute su
[12:22] <dewey_> it's running as root
[12:29] <twb> Presumably because it can't execute su
[12:33] <dewey_> I've changed it to /bin/su and now it's working
[12:33] <dewey_> sweet
[13:01] <dewey_> is there  a way to find a pid file?
[13:02] <joschi> dewey_: `find`
[13:02] <joschi> dewey_: PID files are really not standardized in any way
[13:02] <dewey_> yeah but the name is not clear
[13:02] <joschi> dewey_: and basically really only contain the PID of the process in question
[13:03] <dewey_> but i need it for monit to monitor
[13:03] <dewey_> can't find it anywhere :p
[13:03] <joschi> dewey_: does the process create one?
[13:03] <dewey_> yes
[13:03] <twb> I think Debian policy and start-stop-daemon(8) encourage people to put them in /var/run/
[13:03] <joschi> dewey_: so you know it's there
[13:07] <RoyK> dewey_: lsof -p pid-of-whatever-process-having-the-pid-file-open
[13:08] <joschi> RoyK: the PID file usually only written on startup of the daemon and then closed
[13:08] <dewey_> RoyK: thanks, that worked out
[13:09] <joschi> hm, is there any use in keeping the PID file open?
[13:09] <RoyK> not really :P
[13:29] <BlackDex> Hello..
[13:29] <BlackDex> i have a problem with upgrading
[13:29] <BlackDex> here is the error: http://paste.ubuntu.com/546539/
[13:29] <BlackDex> how can i fix this?
[13:53] <riktking> hi i need some help with apache, can only get one page to show, no sub pages from an external pc
[14:09] <rigved> BlackDex: https://bugs.launchpad.net/ubuntu/+source/grub2/+bug/692163 and https://bugs.launchpad.net/bugs/693059
[14:26] <BlackDex> rigved: Thx.. i almost got afraid i did something wrong :)
[14:33] <hallyn> zul: don't suppose you had any plans of doing a new samba-daily upload anytime soon?  :)
[14:37] <emre> hi can anybody help me with ubuntu server networking settings (two nics)
[14:37] <pmatulis> !ask | emre
[14:38] <emre> I have installed ubuntu server in a vmware machine
[14:38] <emre> I have two nics (hostonly and bridged)
[14:38] <emre> eth0:192.168.123.12 (static)
[14:38] <emre> eth1:192.168.234.40 (dhcp)
[14:38] <emre> I can ping everything in lan wihtout issue. I cannot ping outside lan.
[14:38] <emre> My resolv.conf file shows correct dns server
[14:38] <emre> If I disable eth0 than I can ping outside of lan
[14:38] <emre> how can I make this two eths work together
[14:38] <patdk-wk> did you define a gateway for eth0?
[14:39] <emre> yes
[14:39] <patdk-wk> well, there is your issue
[14:39] <emre> it is 192.168.123.1
[14:39] <RoyK> emre: pastebin /etc/network/interfaces
[14:39] <RoyK> !pastebin
[14:39] <pmatulis> emre: you made 2 gateways
[14:40] <emre> if remove gateway for eth0 than I cannot ping my host 192.168.123.1 which is the host that is running vmware
[14:41] <pmatulis> emre: you made 2 default gateways.  you should create a static route for contacting the host
[14:41] <RoyK> there is no such thing as a gateway for a nice
[14:41] <RoyK> there are routers for networks, default gateway is for network 0.0.0.0/0
[14:42] <emre> so should I disable eth0 gateway?
[14:42] <patdk-wk> hmm, if you have 192.168.123.12 on eth0
[14:42] <patdk-wk> and you can't access 192.168.123.1
[14:42] <patdk-wk> I wonder what your netmask is
[14:42] <RoyK> emre: pastebin /etc/network/interfaces
[14:42] <patdk-wk> assuming your probably using /24, it should just work, no gateways needed
[14:42] <emre> than how can I ping 192.168.123.1 is host's ip adres and 192.168.123.12 is guest. I have a netmask 255.255.255.0
[14:43] <hallyn> zul: do you know if anyone ever bothered to look at https://bugzilla.samba.org/show_bug.cgi?id=6724 ?
[14:44] <emre> I am on a different network to chat with. I cannot reach vmware now so I cannot get interfaces file contents to paste here.
[14:45] <emre> I will gather that or ask for irc access from within company
[14:45] <emre> thanks
[16:05] <axisys> ok i am still struggling with this box.. at one point i were able to get to previous linux kernel from grub and then I got fsck status 4 .. I need to run fsck on disk /dev/md1 .. so I booted from live CD and picked Rescue mode.. but I do not see fsck in there.. any idea how can I fsck now ?
[17:13] <bdmurray_> bug 581941 might be the same as bug 690401
[17:13] <axisys> ok i booted from live CD .. but I do not see the /dev/md1 .. can anyone help?
[17:14] <bdmurray> SpamapS: ^
[17:22] <axisys> i need to fsck /dev/md1 from live cd since it is in fsck status 4
[17:24] <SpamapS> axisys: cat /proc/mdstat .. nothing?
[17:24] <axisys> rebooting it.. will check again as soon as comes back up
[17:24] <SpamapS> bdmurray: looking.. I think you're right.
[17:25] <axisys> SpamapS: nothing
[17:26] <axisys> SpamapS: there is no mdadm installed on live cd
[17:26] <SpamapS> axisys: you don't actually need mdadm if the kernel sees your partitions as raid autodetect.
[17:27] <axisys> SpamapS: fdisk -l shows all those sd?1 are linux raid autodetect
[17:27] <axisys> but no md device
[17:31] <SpamapS> axisys: weird
[17:32] <SpamapS> axisys: well install mdadm and ask it to scan
[17:33] <axisys> mdadm --scan ?
[17:33] <axisys> mdadm --assemble --scan gave me something
[17:34] <axisys> now I see the md devices..
[17:34] <baggar11> axisys: just hopped in, did you already have a raid array on your drives?
[17:34] <axisys> md1 (root) and md10 (data)
[17:34] <axisys> baggar11: yes
[17:34] <baggar11> axisys: will probably need to tell mdadm to assemble your array with --assemble and list the devices
[17:35] <axisys> baggar11: yep
[17:35] <axisys> baggar11: thanks
[17:35] <baggar11> axisys: then mdadm --detail --scan >> /etc/mdadm/mdadm.conf
[17:35] <axisys> baggar11: i am on live cd
[17:35] <axisys> i guess i need to mount the md1 and then put it there
[17:36] <baggar11> axisys: are you just trying to mount then?
[17:37] <axisys> baggar11: fsck went fine.. now I can mount md1 fine.. so I will just update the mdadm.conf and reboot
[17:40] <axisys> i also modified the /a/etc/default/grub .. so I can see the grub menu .. hidden timeout was `0' .. changed it to `5' .. so I can pick a kernel of my choice.. how do I upgrade the grub now that root is mounted on /a ?
[17:40] <axisys> do I chroot and then run upgrade-grub ?
[17:41] <baggar11> axisys: you could also just hold down the shift key when you want to pick another kernel
[17:41] <baggar11> on boot...
[17:41] <axisys> baggar11: that did not work.. tried so many times.. I guess just for this time i modify the grub.cfg instead
[17:47] <Velmont> Question; I just can't understand why my new, totally idle server has a load of 1.21  1.16  1.09. It's not doing anything. Nothing is using CPU in htop! I have software mirroring RAID though, but that shouldn't use that much power.
[17:54] <SpamapS> bdmurray: agreed, that bug does seem to be the same statd issue
[17:54] <baggar11> axisys: did you use the left shift? and hold it down right after the bios leaves the screen?
[17:54] <bdmurray> SpamapS: okay, shall I mark one as a duplicate of the other?
[17:55] <SpamapS> bdmurray: seems like there is a lot more information in bug 581941 ... I assume the merge proposal from bug 690401 will be moved to it upon marking it as a duplicate?
[17:56] <bdmurray> SpamapS: I would not put money on it
[17:57] <SpamapS> bdmurray: well I can reassociate it manually.. I'll mark the duplicate tho
[17:57] <bdmurray> SpamapS: okay thanks
[17:59] <axisys> any idea why my system is stuck here now?
[17:59] <axisys> http://pastebin.com/iLjzm5J5
[17:59] <axisys> it rebooted fine with md1 until there
[18:01] <baggar11> axisys: can you pastebin your /proc/mdstat ?
[18:01] <axisys> baggar11: i have not got a prompt yet
[18:02] <axisys> see my last pastebin.. i got stuck
[18:04] <baggar11> axisys: is md1 your root partition?
[18:05] <axisys> baggar11: yes
[18:05] <Velmont> OK. "top" is clearly better than "htop". It showed me the sinners:  md0_resync and md0_raid1.   However, in a fresh install of Ubuntu Server, why do they use 10% CPU?  I installed RAID right in the installer, so there should be no use resyncing. Anyone have an idea`
[18:06] <baggar11> Velmont: which level of raid?
[18:07] <axisys> i am guessing it is doing the fsck of /dev/md10 (raid10) .. 200GB .. might be why it is stuck there..
[18:07] <axisys> baggar11: ^
[18:07] <axisys> baggar11: i will go grab a bit real quick and that should probably enough time for that fsck to complete
[18:08] <Velmont> baggar11: RAID1, mirror.
[18:08] <axisys> s/bit/bite/
[18:09] <baggar11> Velmont: cat /proc/mdstat and it will tell you if it's rebuilding or not
[18:10] <baggar11> axisys: ok
[18:10] <baggar11> Velmont: or you can mdadm --detail /dev/md0
[18:10] <gobbe> it takes while to build raid, even that you jsut installed it
[18:11] <Velmont> baggar11: Thank you. It seems to be, yes... But why? Do the Ubuntu install first only use the first disk? And then the real machine has to sync afterwards?   [[18:11] <gobbe> yes
[18:12] <gobbe> that's how software-raid works
[18:12] <gobbe> in fact, even empty disk takes while to sync
[18:12] <gobbe> so you can either do the raid and wait until sync is ready, or just install and wait afterwards
[18:13] <RoyK> linux software raid is block-based, so it doesn't matter how much data is stored there, it'll take the same amount of type to sync up the mirror even if the filesystems on it have data or not
[18:13] <baggar11> Velmont: the install process starts the syncing process, when you reboot, it finishes it
[18:14] <gobbe> RoyK: yes, that's what i tried to say also :)
[18:14] <RoyK> baggar11: the install process makes the md devs, the kernel starts the sync, then, after reboot, the kernel continues to sync. the installer knows nothing about the syncing
[18:14] <Velmont> baggar11: Hmmm. But the drives are both totally clean. I guess it should take only the fraction of a second to "sync" it.
[18:15] <baggar11> RoyK: I think you know what I meant ;)
[18:15] <RoyK> Velmont: read what I just wrote - it's block based - it doesn't see data
[18:15] <gobbe> Velmont: no it doesn't
[18:15] <Velmont> RoyK: OK. :-)
[18:15] <gobbe> it needs to copy every block
[18:15] <RoyK> Velmont: as opposed to zfs, where an empty zpool takes about zero time to 'rebuild' whereas a fuller one takes longer
[18:16] <Velmont> gobbe: RoyK: Sorry I didn't read the parts without hilight. :-) Did now. Very nice.
[18:16] <gobbe> Velmont: that's ok :) just hold on, it will take while
[18:18] <Velmont> Good to know my new, nice server isn't having over 1 in load by just idling as I first thought. :P
[18:18] <axisys> i am back.. still at this line
[18:18] <axisys> fsck from util-linux-ng 2.17.2
[18:19] <axisys> /dev/md1: clean, 59312/2523136 files, 479698/10078960 blocks
[18:19] <axisys> i am guessing it will take a while for the raid10 to finish ?
[18:19] <axisys> the data device /dev/md10 that is
[18:19] <baggar11> axisys: do you have activity on your drive lights?
[18:20] <gobbe> axisys: fsck might take forever with bigger disks
[18:20] <axisys> does not look like it..
[18:20] <axisys> baggar11: ^
[18:21] <baggar11> axisys: what's the scenerio here? is this is fresh raid install, did you add raid to an existing system?
[18:21] <axisys> gobbe: raid10 of 6 disk 73GB disks
[18:21] <RoyK> I'm quite glad I don't need to fsck this one
[18:21] <RoyK> pbpool/tmp             97T   60K   97T   1% /pbpool/tmp
[18:22] <gobbe> axisys: so you are trying to fsck that?
[18:22] <axisys> RoyK: zpool scrub is slow too.. you may have to run that sometime ;_0
[18:22] <RoyK> axisys: it'll probably take a while to scrub that, yes...
[18:22] <axisys> gobbe: i think the system is trying to fsck that before booting
[18:22] <axisys> before giving me the OS prompt
[18:22] <gobbe> axisys: are the numbers growing or does it look like it's stalled?
[18:22] <RoyK> axisys: but currently we only have a couple of terabytes on it :P
[18:24] <axisys> baggar11: i have data in /dev/md10
[18:24] <axisys> but latest kernel broke md..
[18:24] <axisys> finally i got back to it using the old kernel
[18:25] <baggar11> axisys: hmm, is this 10.10 or 10.04?
[18:25] <axisys> baggar11: 10.04
[18:25] <baggar11> axisys: I'm running 10.04 with raid1 and raid5 volumes, updated to new kernel with no issues
[18:25] <axisys> i have few servers went through this.. no issue.. except this x4150
[18:28] <baggar11> axisys: does the system run on the old kernel fine still?
[18:29] <axisys> baggar11: yes.. md1 is recognized fine.. except md10 probably holding it up
[18:29] <axisys> will comment it in fstab through live cd and reboot again
[18:30] <baggar11> axisys: yeah, I would try re-adding the mirroring portion
[18:31] <pedahzur> Any rumors on when the bug-fix packages will be out for Postgresql?  New versions were released recently. Not seeing them yet on 10.04
[18:34] <axisys> on livecd .. mdadm --assemble --scan brings up md1 (root) and md10(data) but not md0 (swap) .. any idea why ?
[18:34] <axisys> md0 is raid1 of sda1 and sdb1 .. i see those
[18:35] <axisys> cat /proc/mdstat shows only md1 and md10 and both are active
[18:36] <baggar11> axisys: assemble your md0(swap) and mkswap on it again
[18:37] <baggar11> axisys: what does your mdadm.conf look like?
[18:37] <wizardslovak> hello people
[18:37] <axisys> baggar11: so mdadm --assemble /dev/md0, like that ?
[18:38] <wizardslovak> i got corsair memory 2g in pc , will anything happen if i add 2gb of kingston, same specs
[18:39] <baggar11> axisys: you may need to use --force if that doesn't do it
 try and mem test with a livecd or something, you will see
[18:40] <wizardslovak> nah think is i didnt order kingston yet
[18:40] <wizardslovak> i am just curious
[18:43] <axisys> ok.. after commenting the /dev/md10 .. i got my OS prompt back .. also my md0 is there now
[18:44] <axisys> baggar11: here is the how the /proc/mdstat looks like now
[18:44] <axisys> http://pastebin.com/ZheX6UwU
[18:44] <axisys> md10 was commented out in fstab
[18:46] <axisys> so how do I make md10 active ?
[18:47] <axisys> mdadm --assemble --scan says
[18:47] <axisys> mdadm: /dev/md10 has been started with 5 drives (out of 6).
[18:48] <baggar11> axisys: I think /dev/sdc is doing something funky, you may need to remove that from md10 and re-add it
[18:48] <baggar11> axisys: looks like it's trying to be apart of md_d10(?)
[18:50] <RoyK> axisys: one dead drive, then?
[18:51] <baggar11> RoyK: it's probably not a dead drive since it's show up on md_d10
[18:51] <RoyK> axisys: I don't understand that output - looks to me md10 is an inactive stripe
[18:52] <RoyK> s/$/ /set/
[18:52] <baggar11> axisys: try mdadm --stop /dev/md_d10 and then mdadm --manage /dev/md10 --add /dev/sdc1
[18:54] <axisys> baggar11: did exactly as you suggested
[18:54] <axisys> baggar11: http://pastebin.com/49bH6TjR
[18:54] <axisys> baggar11: i guess that looks better now
[18:54] <axisys> baggar11: can I mount it while it is rebuilding ?
[18:55] <axisys> baggar11: thank you so much for your help
[18:55] <baggar11> axisys: cool, I've had that happen before, for some reason the superblock probably got toast on that drive
[18:55] <baggar11> axisys: yes, you can mount while it's rebuilding
[18:55] <baggar11> axisys: read/writes will just be slower
[18:56] <axisys> baggar11: ok.. i will wait .. estimated time 10 mins..
[18:56] <axisys> baggar11: so sdc1 is fixed for now ?
[18:57] <baggar11> axisys: hopefully
[18:58] <baggar11> axisys: you might just check your mdamd.conf with what mdadm --detail --scan produces after the rebuild
[18:59] <axisys> baggar11: does this look kosher?
[18:59] <axisys> baggar11: http://pastebin.com/7gv5xKqR
[18:59] <baggar11> axisys: the /dev/md10 will probably remove the "spares=1" once it's done rebuilding, make sure those lines are in your mdadm.conf for reboot
[19:00] <axisys> just the ARRAY lines .. correct?
[19:02] <axisys> which timeout makes sure the grub menu stays there for 5 secs .. if i had that instead of `0' i could have switch my kernel and call it a day.. instead of trying to fix it for 2 days
[19:03] <axisys> here is my current default grub
[19:03] <axisys> http://pastebin.com/ABUHtW4V
[19:03] <baggar11> axisys: yeah, array lines
[19:04] <axisys> i also like the boot process show up on both tty0 and ttyS0
[19:04] <axisys> with grub it was easy.. i just put it in kernel line.. how do I do it with grub2?
[19:07] <axisys> in other words.. in grub my menu.lst looks like something like this
[19:07] <axisys> http://pastebin.com/W2bUctfn
[19:07] <axisys> how do I achieve similar setup with grub2 ?
[19:08] <baggar11> axisys: don't know
[19:08] <axisys> baggar11: no prob.. asking in the mailing list
[19:10] <axisys> baggar11: http://pastebin.com/dHkBLvQi
[19:10] <axisys> baggar11: mdstat looks complete
[19:10] <axisys> wonder what those unknown formats are all about
[19:13] <baggar11> axisys: remove the first "0" in 00.90
[19:13] <baggar11> axisys: you won't get those unknown messages anymore
[19:13] <axisys> heh
[19:14] <baggar11> axisys: time to reboot and find out if that worked... :)
[19:20] <hggdh> hallyn: want to discuss bug 691590?
[19:24] <ssureshot> anyone know if they have built 389-ds for ubuntu 10+ yet?
[19:25] <hallyn> hggdh: i'm just trying to tell whether it is blocking you, or whether you are objectint go libvirt's security scheme.  (IF the latter, then we probably need to take it up with them, not with jdstrand)
[19:27] <hggdh> hallyn: I can bypass (and will) by running a hack to chown the files back to the correct user under cron, or something like
[19:27] <hggdh> hallyn: but I think this will end up upstream...
[19:27] <RoyK> I wonder if btrfs will stabilise one day...
[19:31] <hallyn> hggdh: why doesn't setting libvirt user to root work?
[19:31] <hallyn> hggdh: i thought you commented that it left the files as is?
[19:33] <hallyn> hggdh: <shrug> maybe we can just add an extra cmdline switch to stop that happening.  Send it to libvirt list and get their comment.
[19:33] <hggdh> hallyn: they get chown-ed to root, my correction was re. the permissions
[19:33] <hggdh> hallyn: will do
[19:33] <axisys> baggar11: looks like all data in md10 is fine
[19:33] <axisys> baggar11: yep.. time to reboot it
[19:34] <hallyn> hggdh: you're going to write the patch?
[19:34] <hallyn> hggdh: (else I'll do it, just may not get to it this week)
[19:34] <hggdh> hallyn: no, sorry, never worked with the libvirt source, and I am not sure I want to start with a changethis area
[19:35] <hggdh> hallyn: what I will do is search the libvirt ML for any comments re. that -- this is a major change, and most certainly somebody has commented on it
[19:35] <hallyn> hggdh: ok
[19:36] <baggar11> axisys: cool, let me know how it goes
[19:37] <ssureshot> question on ppa.launchapd.net ... are these fine to use for production use?
[19:38] <axisys> baggar11: this is how my fstab looks like now
[19:38] <axisys> baggar11: http://pastebin.com/vmBkvTSC
[19:38] <axisys> may be I should put the UUID
[19:39] <baggar11> axisys: shouldn't matter, but it's probably neater to follow the same method for representing the devices
[19:46] <axisys> baggar11: it got stuck again at here
[19:46] <axisys> /dev/md1: clean, 59318/2523136 files, 479816/10078960 blocks
[19:46] <axisys> i guess i need to comment the md10 again fron live cd
[19:47] <axisys> may be i need add a switch in fstab ?
[19:48] <baggar11> axisys: maybe change 0 0 to 0 1, that's what I have on my server
[19:49] <axisys> ok
[19:51] <baggar11> axisys: ah, remove the quotes around the UUID
[19:51] <baggar11> axisys: I also don't have those
[19:51] <baggar11> axisys: not sure if that matters though
[19:51] <axisys> ok
[19:53] <baggar11> axisys: both of my raid volumes 1&5 are using errors=remount-ro
[19:54] <baggar11> axisys: I see your md10 is using defaults,
[19:54] <axisys> yep
[19:55] <hggdh> hallyn: https://bugzilla.redhat.com/show_bug.cgi?id=568935
[19:58] <hallyn> hggdh: heh, so I can post a patch there
[20:16] <axisys> i added the errors and 0 1 .. still stuck
[20:18] <axisys> baggar11: ^
[20:19] <baggar11> axisys: do you get to a login?
[20:19] <sparc> Hmm, I see amd64 arch for Lucid LTS
[20:19] <sparc> but not an x86_64
[20:20] <sparc> will it have intel extensions compiled in there?
[20:20] <axisys> baggar11: no.. it hangs at that line
[20:20] <axisys> /dev/md1: clean, 59318/2523136 files, 479816/10078960 blocks
[20:22] <baggar11> axisys: the only other thing I can think of is zeroing the superblock on that pesky sdc1 drive
[20:23] <baggar11> sparc: amd64 is the 64bit version and will work on amd and intel systems
[20:25] <axisys> mdadm --zero-superblock /dev/sdc1 ?
[20:26] <baggar11> axisys: might need to preface with --misc
[20:26] <baggar11> axisys: mdadm --misc --zero-superblock /dev/sdc1
[20:26] <baggar11> axisys: make sure it's not apart of a raid volume though
[20:33] <axisys> this time it is sdd[1](S) md_d10
[20:34] <axisys> hehe .. russian roullete
[20:34] <baggar11> axisys: nice
[20:37] <axisys> i am guessing it is not fstab.. for some reason md10 is not stable and because it is inactive .. it does not mount
[20:38] <axisys> one final reboot.. with md10 commented..
[20:38] <axisys> crossing my finger
[20:39] <baggar11> axisys: pretty weird, I'm sure once you get through all the drives, it'll be fine
[20:40] <axisys> baggar11: you mean run the zero block for each drive ?
[20:40] <baggar11> axisys: if that's what it takes, but not all at once...
[20:41] <baggar11> axisys: if it keeps coming up with a different drive on reboots
[20:41] <axisys> baggar11: yep.. as long as the drive is outside the array first..
[20:41] <axisys> rebooted fine w/ md10 detached
[20:42] <baggar11> axisys: check your /proc/mdstat, that will probably tell you if a drive is out of line
[20:42] <baggar11> check for those underscored md_** devices
[20:42] <axisys> this time it is for sdh
[20:45] <baggar11> axisys: pretty weird, but keep at it
[20:48] <axisys> is there a better raid software than md ? just curious .. cant change for this server
[20:48] <axisys> what is recommended or popular one this days for ubuntu ?
[20:49] <pedahzur> axisys: That's about it. No other software RAID for Linux that I know of.
[20:50] <baggar11> axisys: I've been lovin mdadm, much better than the raidtools approach
[20:50] <baggar11> no sure that there is anything else though
[20:50] <gobbe> axisys: you mean better in performance?
[20:51] <gobbe> software-raid is in fact in same cases better performancing than hardware-versions
[20:52] <axisys> gobbe: well yes that first.. and may be more stable
[20:52] <gobbe> why md is not stable?
[20:53] <axisys> gobbe: well i am having issue mounting a raid10 .. baggar11 is helping me with that
[20:53] <RoyK> gobbe: md is stable
[20:53] <RoyK> it's been stable for years
[20:53] <axisys> i can mount manually but if i have it in fstab it does not mount
[20:53] <gobbe> RoyK: i know
[20:53] <axisys> is there something like `dm' that can be used instead of `md' ?
[20:54] <RoyK> axisys: are the md modules in /etc/modules? that might help...
[20:54] <gobbe> RoyK: i'v been using it several years
[20:54] <RoyK> axisys: lvm, yes
[20:54] <RoyK> axisys: but not for raid[456]
[20:54] <axisys> RoyK: i have loop, lp and rtc in /etc/modules
[20:55] <RoyK> axisys: mount the md volumes and pastebin lsmod
[20:55] <RoyK> !pastebinit
[20:55] <axisys> RoyK: http://pastebin.com/wkB7Gczw
[20:58] <RoyK> axisys: probably overkill, but try to add these to /etc/modules http://pastebin.com/r7QDEAXL
[20:58] <RoyK> axisys: pastebin /etc/fstab too
[20:59] <axisys> RoyK: added.. let me uncommend the /dev/md10 is see if it gets stuck again
[21:00] <axisys> RoyK: http://paste.ubuntu.com/546712/
[21:00] <RoyK> k
[21:01] <RoyK> axisys: also, passno should be 2 for the last fs
[21:01] <RoyK> not 1
[21:02] <axisys> RoyK: oops .. too late.. may be on next reboot
[21:03] <axisys> RoyK: whats the diff between pass 1 and pass 2 (learning)
[21:03] <RoyK> man fstab :P
[21:03] <axisys> or man mount i guess
[21:04] <axisys> fstab did not say much
[21:04] <RoyK> axisys: on lucid, it does...
[21:05] <axisys> i got stuck at same place ..
[21:05] <axisys> /dev/md1: clean, 59319/2523136 files, 479989/10078960 blocks
[21:05] <axisys> nothing after that..
[21:05] <axisys> is there a way to fix the fstab without the live cd ?
[21:05] <gobbe> boot it with single user
[21:05] <RoyK> gobbe: not sure if that helps on ubuntu
[21:06] <gobbe> axisys: you could also run fsck manually
[21:06] <gobbe> axisys: after you get system up and running
[21:06] <RoyK> the problem is probably fsck order
[21:06] <RoyK> set that to 2
[21:06] <RoyK> not 1
[21:06] <gobbe> or you could even take it up with 0
[21:07] <axisys> RoyK: ok.. but w/o live cd ?
[21:07] <RoyK> try single
[21:07] <axisys> gobbe: 0 did not help
[21:07] <axisys> it was 0 originally.. then changed it to 1 .. next would be 2
[21:08] <RoyK> gobbe: 0?? that's shutdown
[21:08] <gobbe> axisys: but anyway you need to run fsck on your filesystem, that has nothing to do with actual raid (or atleast i dont see any reason)
[21:08] <jdstrand> hallyn: hey, you around today?
[21:08] <RoyK> run level 1 == single
[21:08] <gobbe> RoyK: not there :D
[21:08] <gobbe> RoyK: fstab
[21:08] <RoyK> ah
[21:08] <gobbe> single or just 1
[21:10] <axisys> it gets stuck at same line with rescue mode
[21:10] <axisys> /dev/md1: clean, 59319/2523136 files, 479989/10078960 blocks
[21:10] <gobbe> single user, not rescue
[21:10] <axisys> gobbe: i thought rescue runs in single user mode.. hmm
[21:10] <RoyK> gobbe: same thing
[21:11] <RoyK> axisys: boot on a cd
[21:11] <axisys> RoyK: ok
[21:11] <gobbe> uh
[21:11] <gobbe> yea, same thing. my bad
[21:12] <gobbe> axisys: you can also trigger ext4 to not run fsck on every now and then with tune2fs
[21:13] <axisys> gobbe: ok
[21:21] <hallyn> jdstrand: yeah, i'm around, but i have no kvm-capable laptops handy.
[21:21] <hallyn> jdstrand: actually, heading out now for some xmas shopping.  bbl
[21:23] <axisys> RoyK: changed pass to 2 .. rebooting ..
[21:24]  * axisys crossing my finger
[21:27] <axisys> RoyK: i am glad the server is under my cube and a monitor/kbd plugged into it.. i wonder how would i boot from live cd when the server is at remote location.. i only have remote console access thru cisco console server
[21:29] <axisys> btw .. back to same stuck screen ..
[21:29] <axisys> /dev/md1: clean, 59319/2523136 files, 479989/10078960 blocks
[21:29] <axisys> i guess boot from cd and comment the md10 again
[21:30] <axisys> wish i went with lvm .. hmm
[21:30] <axisys> wonder if i can still do it and save the data
[21:35] <RoyK> axisys: mount it, connect an usb drive, copy
[21:36] <axisys> RoyK: alright
[21:37] <RoyK> axisys: what sort of data do you have on this one?
[21:37] <axisys> RoyK: splunk
[21:37] <RoyK> ?
[21:37] <RoyK> raid10 for log data?
[21:38] <axisys> RoyK: mainly compressed files that is used by splunk (splunk.com)
[21:38] <axisys> for search real time indexed data
[21:38] <axisys> kind a like database
[21:39] <axisys> once i turn off splunk app .. this md10 is quite
[21:39]  * RoyK thinks of zfs
[21:39] <axisys> no transactions
[21:39] <axisys> zfs fs is slow if i put splunk on local zone
[21:39] <RoyK> hm.. it is?
[21:39] <axisys> that is why moving to ubuntu.. splunk guys can help with linux more
[21:40] <RoyK> why don't you just run splunk on some system with native zfs?
[21:40] <RoyK> I mean, not in a zone
[21:40] <axisys> RoyK: yes, if you need to index about 50GB data per day .. constant data coming in from 100s of servers
[21:41] <RoyK> then why run it in a zone?
[21:41] <axisys> RoyK: i am.. one of the splunk server.. and it is working lot better.. not sure why local zone behaving like this.. it is the same release of sol 10
[21:41] <RoyK> you don't want to run that sort of machine in a zone
[21:41] <axisys> RoyK: i guess not..
[21:42] <RoyK> axisys: give openindiana or s11ex a try
[21:42] <axisys> RoyK: so system is back up with md10 commented .. oh well
[21:42] <RoyK> install on a single drive or mirror
[21:42] <RoyK> use zfs for the rest of the stuff, striped mirrors
[21:42] <RoyK> how many drives? 4? 6?
[21:43] <axisys> RoyK: i am running sol11ex .. i have been on opensolaris snv_141 for a while.. now snv_151 i think w/ sol11
[21:43] <axisys> RoyK: 6 on raid10
[21:43] <RoyK> k
[21:43]  * RoyK uses openindiana
[21:43] <axisys> RoyK: i was.. but sol11ex is higher snv .. so i swithced
[21:44] <axisys> RoyK: waiting for the 7410 to power up soon
[21:44] <axisys> RoyK: then NFS that as my splunk storage to test .. hehe
[21:45] <RoyK> http://pastebin.com/2XDHTNnX
[21:45] <axisys> http://paste.ubuntu.com/546726/ <-- keep doing the same thing.. this time sdh
[21:46] <axisys> RoyK: yep.. pretty familiar with that.. I have been using sol10 since day 1
[21:46] <axisys> i even have the opensolaris T shirt
[21:46] <axisys> free :-)
[21:47] <axisys> anyways .. any other suggestion with md10 ?
[21:52] <RoyK> axisys: I'd go back to zfs and not try to run splunk in a zone
[21:52] <RoyK> axisys: for such a setup, I'd use something dedicated, meaning _not_ a zone
[21:55] <adac> how to install sun (oracle) java?
[21:55] <adac> in which repository is it contained?
[21:58] <adac> oh it seeems to be in the partner repo
[21:58] <RoyK> http://gizmodo.com/5715650/googles-cr+48-laptop-totally-revolutionizes-safety-instructions
[21:58] <adac> can enable this without casing problems on ubuntu server?
[21:59] <guntbert> !java
[22:01] <adac> !partner
[22:02] <RoyK> nite
[22:02] <guntbert> g'night
[22:02] <adac> guntbert, night
[22:02] <guntbert> :)
[22:02] <adac> ahh
[22:03] <adac> hehe
[22:57] <blaenk> hey does anyone know of a guide or advice on how to manage web server users
[22:57] <blaenk> in other words, I currently have nginx and it's running as user nginx
[22:57] <blaenk> I believe php-fpm is running as user www-data
[22:58] <blaenk> I'm just wondering how to organize all of this, like, where to place my websites (/var/www/site.com ?), the permissions on that folder, etc.
[23:23] <Tm0> Hey i am new to this sorta, i have a Ubuntu server running, and i need to reenter a proccess i disowned. How do i do it. Or is there a better way i can disown a proccess so i can reenter it?
[23:35] <Tm0> Hey i am new to this sorta, i have a Ubuntu server running, and i need to reenter a proccess i disowned. How do i do it. Or is there a better way i can disown a proccess so i can reenter it?
[23:36] <examancer> maybe its because *I'm* new too, but I have no idea what you mean by disowning a "process"
[23:36] <examancer> i know how to change the owner of a file and change it back
[23:36] <examancer> heh
[23:36] <Tm0> proccess.sh &disown
[23:37] <examancer> oh. i just use screen. haven't tried that
[23:38] <Tm0> Well its for a server, i can't have putty open forever
[23:38] <JanC> Tm0: that's why most people use screen (on Ubuntu often through byobu)
[23:39] <examancer> yes, if you use 'screen <command>' or just type 'screen' and then execute whatever commands you want you can "detach" a screen using CTRL+A+D (default) and then reattach it using screen -r <screen number>
[23:40] <JanC> Tm0: screen is sort of like a terminal server for command shells
[23:40] <examancer> so you can detach, close putty, then open another session (on maybe another computer) and reattach the screen to see the output of that screen session
[23:41] <JanC> so you can ssh in and connect to screen when you need, disconnect when you want, things keep running on the server
[23:41] <examancer> the commands running in the screen session will of course continue running even when its not attached to anything
[23:41] <Tm0> I'm running Minecraft, what i want to do is enter the proccess it gives me so i can do stuff on that program's CLI, but i need to disconnect Putty from it.
[23:42] <examancer> screen "should" handle that just fine Tm0
[23:42] <Tm0> Great
[23:42] <examancer> disown may handle it fine as well... just not as familiar with that. looks similar to using nohup. not a bad solution, but i still prefer screen
[23:43] <JanC> Tm0: read up on screen & byobu and experiment with it, once you're used to it, you'll start to use it on all your systems  ;)
[23:43] <examancer> sweet! i suck at server administration, but i may have actually helped someone (even if they didn't get the answer they wanted)
[23:43] <examancer> :-P
[23:43] <Patrickdk> I dunno what &disown does, but if you use normal &, generally you can use fg to bring it back
[23:44] <Tm0> Well i have FG and screen now. I'll see what works. Just a matter of time now
[23:44] <examancer> disown appears to work similar to nohup in that it attaches the processes being run ot the parent process (init)
[23:44] <examancer> i don't know if you can even bring it back to the fg
[23:45] <Tm0> Nor do i. Its not my server, i'm just the server guy. Apparently running a home server means that
[23:46] <JanC> ☺
[23:46] <examancer> i have a theory that most administrators really have no idea what they are doing and just fumble around until things work
[23:46] <examancer> ... but i'm a developer, so maybe i'm supposed to think that
[23:46] <examancer> :-P
[23:50] <blaenk> hey, hwere in a daemon/server does one configure what to run it as. for example, I installed nginx and it's running as user nginx, this is fine, but I'd like to change another daemon's (php-fpm)'s user
[23:50] <Tm0> The only think i develop is HTML/CSS, (Markup at least xD) So i usually know what i'm doing, though procress wise its who knows.
[23:51] <blaenk> I figured it would be in the init.d script but doesn't seem like it
[23:51] <JanC> blaenk: it depends
[23:51] <blaenk> typically?
[23:52] <blaenk> cause I checked the conf/ini files and there's no mention of it, is it seriously hard-coded so I would have to recompile with the configure parameters?
[23:52] <JanC> some daemons do it themselves, and then there is often a setting in the daemon's configuration file somewhere
[23:52] <blaenk> cause I installed php-fpm through apt, and it automatically used www-data, I know that if you compile it yourself you can set the configure opts
[23:52] <blaenk> ok, what do you mean by do it themselves though?
[23:54] <JanC> blaenk: daemons that change user themselves, there are also some that you can run as another user yourself with su/sudo of course...
[23:55] <blaenk> ohhh I see what you mean, so it /would/ be hard-coded at compile time in the case of those former daemons
[23:55] <blaenk> I guess I'll just compile php-fpm myself, might be easier, is there a way you recommend of doing this?
[23:56] <JanC> it could be hard-coded at compile time or it could be in a configuration file...
[23:56] <blaenk> I can do dpkg -l  |grep php to see what packages I installed (i have a rough idea) and then just remove with sudo apt-get remove --purge package right?
[23:56] <blaenk> JanC: I've checked the configuration files but I didn't fidn anything, I'll keep looking
[23:56] <JanC> blaenk: also read the docs  ;)
[23:57] <blaenk> oh, I did
[23:57] <blaenk> man php-fpm
[23:57] <JanC> there might be a separate doc for the config files
[23:57] <blaenk> JanC: can you comment on my package uninstall method above though?