[00:58] <Wyleyrabbit> can anyone suggest a place to get help with an LTS upgrade problem?
[01:08] <pmatulis> Wyleyrabbit: this is a good place for you
[01:09] <Wyleyrabbit> pmatulis, really? I asked a question about an hour ago and got no response.
[01:09] <Wyleyrabbit> OK, I upgraded a server from 10.04 LTS to 12.04 LTS, and now my database-driven website doesn't work.
[01:10] <Wyleyrabbit> seems the CMS (Called MODx) cannot create a database connection.
[01:11] <Wyleyrabbit> not sure if related, but an apache module called "fcgi" continually segfaults.
[01:11] <Wyleyrabbit> everything was working before the upgrade.
[01:15] <pmatulis> Wyleyrabbit: when you try to start apache it doesn't work or only crashes when it is solicited/used?
[01:15] <Wyleyrabbit> only when solicited
[01:16] <pmatulis> Wyleyrabbit: and solicited using mysql/php?
[01:16] <Wyleyrabbit> right
[01:17] <pmatulis> Wyleyrabbit: i would begin by looking for error messages in mysql/php logs
[01:17] <pmatulis> Wyleyrabbit: does mysql start?
[01:17] <Wyleyrabbit> Apache logs show lots of these:
[01:17] <Wyleyrabbit> "[error] mod_fcgid: process /home/domain/fcgi-bin/php5.fcgi(30368) exit(communication error), get signal 11, possible coredump generated"
[01:18] <pmatulis> Wyleyrabbit: i would begin by assuring that mysql and php are in good shape, then look at apache side
[01:18] <Wyleyrabbit> yeah, Mysql starts fine.
[01:18] <Wyleyrabbit> and I can install and use an old version of the CMS.
[01:19] <Wyleyrabbit> seems the new version uses "xpdo" or something to talk to database, and that seems to be a problem now.
[01:19] <pmatulis> Wyleyrabbit: ah ok, when you upgrade your cms everything goes pear-shaped?
[01:19] <Wyleyrabbit> ha ha ha
[01:20] <pmatulis> Wyleyrabbit: is that the case?
[01:20] <Wyleyrabbit> just testing exactly that.
[01:20] <pmatulis> Wyleyrabbit: ok
[01:20] <Wyleyrabbit> set up a new subdomain for testing, installed version 1.x of cms, and it worked fine.
[01:20] <Wyleyrabbit> will know in about 3 minutes about upgrading to 2.x
[01:30] <Wyleyrabbit> pmatulis, argh. php having issues. Setup for Modx 2.x is saying date.timezone setting isn't set (it is!!!).
[01:35] <Wyleyrabbit> anybody know where php5 logs its errors (in Ubuntu 12.04)?
[01:37] <pmatulis> Wyleyrabbit: i believe that is set in the main/global PHP config file.  forget where that is.  you shouold also be able to increase verbosity in the same file (if needed)
[01:40] <pmatulis> Wyleyrabbit: but it's prolly a known issue.  might head over to the modx community (if it exists)
[05:10] <pentagon> 37.9.53.2
[05:10] <pentagon> leadpile.com
[05:10] <pentagon> How is it making a reverse vnc connection past ufw
[05:18] <pentagon> 37.9.53.2RURussian Federation66Saint Petersburg CitySaint Petersburg59.894430.2642Petersburg Internet Network ltd.Petersburg Internet Network ltd.
[05:50] <planet> what is the quickest way of stripping ubuntu server of everything not needed for the bare system to run
[08:47] <sanderj> When I boot ubuntu 12.04 it says I cant find "bnx2-mips-09-6.2.1a.fw"
[08:47] <sanderj> ,  but it's located inside /tmp/lib/firmware/3.2.0-24-generic/bnx2
[08:49] <sanderj> I'm wondring how I repack initrd
[08:51] <RoyK> sanderj: man mkinitramfs
[08:52] <sanderj> RoyK, when I extracted the initrd. The file it was complaining about is there. Do you know what could be wrong then?
[08:55] <RoyK> dunno...
[09:14] <jacobw> hi, i'm trying to make a preseed, how can i generate a hash to use at the root password value?
[09:20] <melmoth> mkpasswd ?
[09:22] <jacobw> thanks
[10:01] <RoyK> jacobw: using that alone, only generates the des-56-encrypted password - better use -m something-cool
[10:13] <sanderj> How do I create a mkinitramfs for a diffrent architecture? mkinitramfs only does it on the same.
[10:17] <sanderj> Is it just a matter of spesifying the correct kernel?
[10:51] <eagles0513875_> hey guys i have dovecot + postfix for email how can i set it up to use multiple domains ?
[10:56] <mld> eagles0513875: tried googling it? have a look at http://library.linode.com/email/postfix/dovecot-mysql-ubuntu-10.04-lucid
[10:57] <mld> eagles0513875: here's another take on it: http://www.exratione.com/2012/05/a-mailserver-on-ubuntu-1204-postfix-dovecot-mysql/
[10:57] <ikonia> eagles0513875: you've set this up befor e- we've been thorugh it before
[10:57] <ikonia> why are you not doing the same as your server that worked
[13:01] <hallyn> smb: awesome, good idea on the saddr=daddr=0 patch :)
[13:02] <smb> hallyn, It is a bit poking in the mud... At least it works with the testcase... :-P
[13:03] <hallyn> smb: do you have a kernel built in ppa?
[13:04] <hallyn> stgraber and SpamapS were able to hit it very quickly, so confirmation from them would go a long way (of course 'i didn't hit it' won't be "proof" :)
[13:04] <smb> hallyn, No I only got 64bit testkernel debs right now which will puke a lot of messages
[13:04] <smb> I could shove them over to people if they want to
[13:05] <hallyn> smb: i think it'd worthwhile
[13:05] <smb> ok
[13:05] <hallyn> (given what you'd posted in the bugzilla before)
[13:06] <smb> stgraber, SpamapS, hallyn, http://people.canonical.com/~smb/clonetst
[13:20] <stgraber> smb: thanks! I'll try that one once I'm done going through my e-mails.
[13:20] <randomDude> Just physically moved server location, one of my cloud servers had a firewall rule allowing ssh access only from a specific ip address (which i no longer have access to), I do however have the cloud server offlined and its harddrive attached to a recovery server and mounted at /mnt/recovery... can I edit a file to reset the firewall rules ?
[13:46] <hallyn> randomDude: depends how you were loading the rule before.  if using ufw, check ufw-framework manpage
[13:46] <randomDude> hallyn: it was with zentyal
[13:47] <randomDude> perhaps if i put a `ufw reset` somewhere in the init scripts after zentyal has loaded all its bits?
[13:56] <stgraber> smb: running your test kernel now, will do a bit of stresstest today (rebuilding all my test containers, playing with arkose/lxc-start-ephemeral to see if I can make something bad happen)
[13:57] <smb> stgraber, Ok cool. Lets see what happens. Though I somewhat still hope to get some upstream feedback... :)
[14:03] <smb> Daviey, Seems you where not adventurous last Friday... ;)
[14:17] <Daviey> smb: oh?
[14:18] <Daviey> the upload, ah :)
[14:18] <smb> :)
[14:37] <ziggyzero> Can anybody help me with installing mod_proxy_html v3.1 so I get rid of the Invalid command 'ProxyHTMLEnable', perhaps misspelled or defined by a module not included in the server configuration error message
[14:38] <mercsniper> Do you know how to compile from source?
[14:40] <mercsniper> Ziggy
[14:40] <ziggyzero> mercsniper: sorry I was trying to ask the same question in httpd too
[14:41] <ziggyzero> mercsniper: but it won't let me post :-(
[14:41] <mercsniper> I have no idea on httpd channel
[14:41] <mercsniper> but do you know how to compile?
[14:42] <ziggyzero> mercsniper: I've seen instructions on how to compile the c source using the apache module compiler but I have not done it before, no.
[14:42] <mercsniper> I have not compiled with apache, but I know how to compile using gcc
[14:42] <ziggyzero> mercsniper: I am paranoid about breaking apache2 package that I installed via apt-get :-|
[14:43] <ziggyzero> mercsniper: basically I just want to proxy to an internal web server, but atm it's redirecting to absolute paths
[14:44] <ziggyzero> mercsniper: I am led to believe that mod_proxy_html 3.1 will resolve the issue but apt-get repository only has 3.0.11
[14:45] <mercsniper> see this http://www.apachetutor.org/admin/reverseproxies
[14:47] <SpamapS> smb: installing your test kernel now
[14:48] <ziggyzero> mercsniper: He's using Apache 2.2 and mod_proxy_html 3.1
[14:49] <ziggyzero> mercsniper: I also want to use mod_proxy_html 3.1
[14:49] <mercsniper> that article details how to install the module with the apxs command
[14:50] <ziggyzero> mercsniper: ProxyHTMLEnable On   and ProxyHTMLURLMap  /      /app1/ are the lines are need
[14:51] <ziggyzero> mercsniper: Ah, I see the section 'Building Apache for Proxying'
[14:51] <mercsniper> then after that section it mentions how to compile the module
[14:51] <mercsniper> using apxs
[14:52] <ziggyzero> mercsniper: I am paranoid about breaking apache2 package :-|
[14:52] <mercsniper> how would you break it
[14:53] <mercsniper> the module is only activated if you have the ProxyHTMLEnable on
[14:53] <mercsniper> building the module will not inherantly break it
[14:53] <mercsniper> the apxs command just builds it
[14:53] <mercsniper> but you need to know if your version has the apxs command by
[14:53] <mercsniper> "which apxs"
[14:55] <ziggyzero> mersniper: I don't have apxs installed. Also not in the repository
[14:57] <ziggyzero> mercsniper: I also have another problem (unrelated to the proxy one)
[14:58] <ziggyzero> mercsniper: I have a SAMBA share setup on another server. I automount this in FSTAB, which works fine
[14:59] <ziggyzero> mercsniper: I map this using an account called Bill with 755 privilges.
[15:00] <mercsniper> ok
[15:00] <ziggyzero> mercsniper: But when Bill created files and folders on the share the privileges it creates them with is 700, but I want 755 so others have readonly access
[15:00] <mercsniper> to install apxs: http://knowledge-republic.com/CRM/2011/09/ubuntu-missing-apxs-fo-compile-apache-module/
[15:01] <ziggyzero> mercsniper: Thanks mercsniper, I'll install that now
[15:02] <ziggyzero> mercsniper: on my other issue the automount command I am using is: //192.168.1.2/docs /mnt/docs cifs auto,iocharset=utf8,uid=bill,gid=staff,credentials=/root/.cifscredentials,file_mode=0755,dir_mode=0755 0 0
[15:02] <ziggyzero> mercsniper: but when bill creates files they are created with 700. The umask is showing as 0022. Any advise also appreciated
[15:11] <SpamapS> smb: any help on how I can run that test kernel of yours with bcmwl ?
[15:11] <SpamapS> smb: I get a build failure from dkms complaining about GPL/non-GPL
[15:18] <smb> SpamapS, Did you install all of the debs? I am cheating there a bit because it would otherwise need an ABI bump
[15:23] <SpamapS> smb: yes installed them all
[15:23] <SpamapS> smb: http://paste.ubuntu.com/1254068/
[15:24] <SpamapS> smb: http://paste.ubuntu.com/1254069/
[15:24] <SpamapS> smb: second one is the dkms fail
[15:24] <smb> SpamapS, Hm, ok. Can you paste me the error that dkms complains about? Wonder whether it is because of the the lockdep enablement... ok thanks
[15:26] <SpamapS> smb: can I assume then that I can't just cheat and copy the one from -16?
[15:26] <smb> SpamapS, Yeah sounds like wl cannot be compiled when the kernel has lockdep enabled. *sigh*
[15:26] <SpamapS> OH
[15:26] <smb> SpamapS, Nothing guaranteed but its worth a try
[15:27] <smb> Though I fear that the module abi hashes don't match and so it won't work
[15:27] <SpamapS> smb: can I just lie to dkms that wl is gpl compatible somehow?
[15:27] <SpamapS> I promise I won't distribute wl.ko to anybody except me ;)
[15:29] <smb> SpamapS, Never tried it... not sure its enough to modify the MODULE_LICENSE line in the source code, but you could try
[15:30]  * smb did not say that and denies any recollection of the incident
[15:30]  * xnox that's what she said.... we have logs!
[15:31] <xnox> http://www.xkcdb.com/logo.png
[15:36] <CharlieSu> Hi all.  I'm running Ubuntu Server on one of my servers that acts as a SFTP server and I'm noticing that it is starting to take a long time to login.  The console-kit-daemon process is eating up a lot of CPU.  What does this process do?
[15:36] <holstein> login? or boot?
[15:37] <Jeeves_> CharlieSu: Good question. I've been wondering about that for some years now
[15:37] <Jeeves_> No manual entry for console-kit-daemon
[15:38] <SpamapS> smb: well I don't have any non-bcmwl machines here.. so, can't test your kernel. :-/ (changing MODULE_LICENSE caused breakage elsewhere)
[15:39] <chmac> How can I disable the auto update check? Can only find instructions that involve clicking in the forums, etc.
[15:41] <holstein> http://askubuntu.com/questions/172524/how-can-i-check-if-automatic-updates-are-enabled seems relevant chmac
[15:43] <SpamapS> smb: hang on, don't start a non-lockdep build or anything, I might have it..
[15:43] <smb> SpamapS, Oh I was not (even yet) ;)
[15:46] <SpamapS> ok booted and testing now
[15:47] <CharlieSu> Jeeves_: https://bugs.launchpad.net/consolekit/+bug/284229
[15:50] <chmac> holstein: Awesome, that looks like exactly the ticket, thanks a lot. :-)
[15:50] <stgraber> hallyn: FYI (not sure it's really worth spending a lot of time on), having commented fstab entries make lxc-start-ephemeral fail, at least on 12.04
[15:50] <stgraber> hallyn: found that out while helping one of the unity guys debug a broken lxc :)
[15:56] <SpamapS> smb: [  701.391431] unregister_netdevice: waiting for lo to become free. Usage count = 2
[15:57] <SpamapS> smb: not fixed :(
[15:57] <SpamapS> root     13315  0.0  0.0  27532  1112 ?        Ds   08:57   0:00 lxc-start --daemon -n clint-local-ci-u2-0 -l DEBUG -o /home/clint/.juju/data/clint-loc
[15:57] <smb> SpamapS, It would be nice if you could send me the full dmesg somewhere
[15:58] <smb> Could be just not the case of daddr=saddr=0
[15:59] <SpamapS> [  821.326828] rt-key: hash=fa11 daddr=7259bd5b saddr=600a8c0 oif=0
[16:03] <SpamapS> smb: recreate with base LXC tools..   lxc-create -n foo -t ubuntu-cloud -- -r precise  ; lxc-start -n foo ; lxc-stop -n foo ; lxc-destroy -n foo ; # Usage count = 2 starts printing
[16:05] <smb> SpamapS, Yeah, I guess that means that preventing any route is not the way forward... Rather needs the route cache purged at some point...
[16:06] <SpamapS> smb: is there a command that can be run to purge the route cache manually?
[16:07] <smb> SpamapS, Maybe /proc/sys/net/ipv4/route/flush...
[16:08] <smb> echoing 1 into
[16:08] <SpamapS> smb: that doesn't seem to do anything unfortunately :-/
[16:10] <smb> SpamapS, Darn, ok, back to thinking... But thanks for testing
[16:15] <SpecialEd> Hi, would anyone know of a good Ubuntu 12.04 VPS hosting company that is physically located in Amsterdam?  I am hoping to configure this VPS to receive LVM Snapshots of my production VPS for use as a failover. I currently have a VPS hosted in France $15USD/mo 512MB RAM, 100GB HDD, 100Mbps uplink...     If this is NOT the place to ask I apologize, any suggestions where I should ask would
[16:15] <SpecialEd> be greatly appreciated!  Thanks :)
[16:18] <RoyK> SpecialEd: I don't think it's the politically correct place to ask, but I don't know if too many bothers - why does it have to be in Amsterdam, when the internet contains so many other species?
[16:21] <SpecialEd> I like the netherlands :)
[16:22] <SpecialEd> for real, its a customer request, don't understand the full scope of it to be honest
[16:22] <SpecialEd> something to do with needing to have a hot site that is specifically so many miles away
[16:22] <RoyK> np
[16:22] <SpecialEd> and I know what you were thinking :P
[16:23] <RoyK> really, all I was thinking was that it doesn't matter much for me where a VPS is, physically, so long as the setup is stable and latency isn't too bad
[16:24] <SpecialEd> true
[16:24] <SpecialEd> may I ask what country your from?
[16:24] <SpecialEd> Im in the states, and I used to work for a data backup company that was all ubuntu driven creating Windows VMs and storing them in an 8 petabyte cloud
[16:25] <SpecialEd> so, given the country your from I can probably tell you reasons why physical location matters (mostly to idiot auditors and government regulations)
[16:26] <SpecialEd> but I personally think at your level with that :)  as a veteran of the disaster recovery field I can tell you that 80-90 % of disasters are the result of human stupidity :)
[16:36] <bananapie> I used the network interfaces on my ubuntu 10.04 to setup network bonding. I have eth1 and eth3 as slaves. I am using active backup ( mode 1 ), when the server boots it uses eth3 as the primary and eth1 as the backup. I want eth1 as the primary and eth3 as the backup. Anyone know how I can fix this ?
[16:38] <SpecialEd> not sure what the best suggestion would be, but maybe fooling the NIC order in /etc/udev/rules.d/70-persistent-net.rules  ?
[16:38] <sarnold> bananapie: I was thinking vaguely along SpecialEd's line of thought, though thought about fiddling with the order in /etc/network/interfaces
[16:39] <SpecialEd> yeah network interfaces would definitely be the place to start
[16:39] <bananapie> You think it's because eth3 is being brought up first ?
[16:40] <RoyK> what parses networking interfaces, btw?
[16:40] <SpecialEd> I dont mind building a 10.04 vm real fast to try for ya
[16:40] <SpecialEd> and I dont believe /etc/network/interfaces is parsable
[16:42] <smoser> kirkland, ping
[16:42] <bananapie> Physically changing the wires at this point is complicated because the server is 350 kilometers away. I had a look at persistent rules, I see that eth1 and eth0 are listed before eth2 and eth3
[16:42] <SpecialEd> sorry, can't do the VM, just realized i dont have my host machine's second nic available
[16:42] <bananapie> eth1 is using tg3 ( according to comment ) and eth3 is sundance ( according to the comment )
[16:43] <SpecialEd> u mind doing a pastebin of /etc/network/interfaces  ?
[16:43] <SpecialEd> I've had to deal with VERY remote network troubleshooting like this, its like walking on eggshells ...
[16:43] <bananapie> Yes
[16:44] <bananapie> fortunately, this machine has something called 'ilo' and is non-critical.
[16:44] <bananapie> http://pastebin.com/3Agq0jCW
[16:44] <SpecialEd> actually i can bridge two virtual nics to a physical and replicate it
[16:44] <SpecialEd> yeah, but if it breaks it still sucks to have to troubleshoot it with an end user (if your lucky to have that )
[16:44] <bananapie> the network interfaces file is the same as another machine, I think my problem is related to the difference in drivers on the cards.
[16:45] <SpecialEd> i got a 10.10 server 64bit ISO locally on this machine, that should be close enough to 10.04 i think
[16:45] <bananapie> Yea, networking stuff doesn't change a lot
[16:47] <RoyK> SpecialEd: 10.10 isn't supported anymore, though
[16:47] <SpecialEd> yeah b/c 10.04 was lts
[16:47]  * RoyK only uses LTS for servers
[16:47] <SpecialEd> but since im gonna try and replicate his issue real fast in a vm it should be close enough
[16:48] <SpecialEd> for this issue
[16:48] <SpecialEd> yeah, I'm pretty much all 12.04 myself right now
[16:48] <bananapie> I am hoping to migrate to 12.04 on my servers in the new year.
[16:48] <RoyK> I still have a few machines/VMs on 10.04 and even an old hardy installation
[16:49] <SpecialEd> hardy was what 8.04 ?
[16:49] <RoyK> mhm
[16:49] <SpecialEd> i think that was my first intro to ubuntu
[16:49] <bananapie> I had a hardy installation, we wiped it out this morning :D
[16:49] <bananapie> my first intro was version 5 something, 5.10 maybe ?
[16:50] <RoyK> problem is - that hardy installation runs zimbra on 32bit and migrating that to 64bit is a PITA, so I haven't gotten around to it yet...
[16:50] <SpecialEd> still installing the OS btw, almost done
[16:51] <bananapie> I think I found the problem, I think the network cards I installed don't support bonding properly and can not read the status of the card.
[16:51] <hallyn> stgraber: is that with the bash or python version of lxc-start-ephemeral?
[16:51] <bananapie> and that's mixing up bonding
[16:52]  * RoyK thought bonding was done in software...
[16:52] <bananapie> yea, but the monitoring ( miimon ) has hardware stuff, no ?
[16:52] <RoyK> bananapie: what sort of NICs are these?
[16:52] <bananapie> 10/100 mbps pci cards ( it's an older server )
[16:52] <RoyK> what chipset?
[16:53] <bananapie> I think it's an HP DL380 g4, eth 3 is a cheapo card from an electronics store.
[16:53] <RoyK> lshw will tell
[16:53] <bananapie> hang on
[16:53] <bananapie> rebooting
[16:53] <bananapie> I think I found a solution.
[16:54] <stgraber> hallyn: bash
[16:54] <SpecialEd> are eth0 and eth2 in another bond?
[16:54] <bananapie> yes
[16:54] <stgraber> hallyn: python should be unafected (didn't try though)
[16:54] <bananapie> ok
[16:56] <bananapie> I am uploading my solution to pastebin
[16:57] <SpecialEd> vm is up
[16:57] <bananapie> http://pastebin.com/xrmQryyf
[16:57] <SpecialEd> im in bash shell
[16:57] <SpecialEd> and playing with it now
[16:57] <bananapie> I used bond-arp instead of bond-miimon, and it boots with the proper card. I think that whatever miimon is using is not working properly on the card.
[16:58] <bananapie> Thanks for your guys' help :D
[16:59] <SpecialEd> not sure bout this...
[16:59] <SpecialEd> network 192.168.2.0
[17:00] <SpecialEd> shouldn't that line be:
[17:00] <SpecialEd> gateway 192.168.2.0
[17:00] <SpecialEd> ?
[17:00] <RoyK> no, .0 is the network address with 24bit mask
[17:01] <RoyK> gateway is the address of the router on the network
[17:01] <SpecialEd> kk
[17:05] <SpecialEd> i can't test ur static route in there
[17:05] <SpecialEd> and im replicating ur config on a 10.
[17:06] <SpecialEd> failed to bring up bond
[17:06] <SpecialEd> hmm
[17:06] <SpecialEd> 1 sec
[17:13] <bananapie> The file '/sys/devices/virtual/net/bond1/slave_eth3' contains 'unknown', it should say 'up'. Anyone know where I need to go next ?
[17:26] <Daviey> zul: wtf is openstack-resource-agents ?
[17:26] <zul> Daviey: its the openstack HA stuff that roaksoax was working on
[17:27] <zul> roaksoax: ^^^
[17:28] <roaksoax> Daviey escoex talk al
[17:28] <roaksoax> err
[17:29] <Daviey> zul / roaksoax: bug 1054022 needs more data before i can even consider reviewing it.
[17:29] <Daviey> roaksoax: Has it been uploaded to Debian yet?
[17:29] <roaksoax> they are scripts that allow monitoring of openstack components in pacemaker based clusters
[17:29] <roaksoax> Daviey not yet
[17:29] <Daviey> roaksoax: For universe ?
[17:30] <roaksoax> yes
[17:30] <vikram> hello... I am using Tomcat 7.0.29 ... my question is why we need to restart tomcat after making configuration changes ( changes to .properties file ) ......???
[17:30] <sarnold> vikram: does tomcat not provide a SIGHUP-style reload-configuration-please trigger?
[17:32] <vikram> sorry .... i am new to tomcat .... let me search it ..
[17:35] <Daviey> zul: ./run_tests .. -P .. what does the P do now?
[17:36] <zul> ignores the pep8 tests
[17:36] <adam_g> jamespage: around?
[17:37] <Daviey> zul: oh, cool
[17:37] <Daviey> zul: is it not pep8 clean now?
[17:38] <zul> Daviey: it is but not for the version we ship in quantal/precise
[17:38] <Daviey> zul: suck
[17:38] <Daviey> thanks
[17:38] <vikram> sarnold: r u talking about "reloadable" attribute of Tomcat's Context ??
[17:40] <sarnold> vikram: most servers allow you to do something as simple as 'kill -SIGHUP `pidof servername`' and they'll reload their configuration file.
[17:41] <jacobw> i want to learn about juju, where's are the resources collated?
[17:42] <vikram> sarnold:  'kill -SIGHUP `pidof servername` means they will stop the server ... i want to apply configuration changes without restarting the server
[17:43] <sarnold> vikram: that's the thing; since the "modem hangup signal" doesn't mean anything to servers the clear majority of servers use SIGHUP to ask for configuration reload.
[17:47] <sarnold> vikram: see here, the signal handler installed will catch SIGHUP and schedule a configuration file reload: http://www.mail-archive.com/tomcat-dev@jakarta.apache.org/msg11338.html
[18:26] <b0ot> What's the easiest way to keep email servers fully synced?
[18:27] <Jeeves_> Ehm, need more info
[18:27] <Jeeves_> define email servers
[18:30] <b0ot> So, when users send email it stores all the email until they are able to get online
[18:30] <b0ot> essentiall store and forward messenging
[18:37] <RoyK> b0ot: all servers do that
[18:39] <b0ot> RoyK, I need all the servers to fully duplicate all email though
[18:39] <b0ot> so any user could connect to any server to get their mail
[18:40] <RoyK> then you need some sort of cluster
[18:41] <RoyK> DRBD could do it, perhaps glusterfs, or if you're not worried about redundancy, NFS will work
[18:45] <smoser> Daviey, https://launchpad.net/ubuntu/+source/cloud-initramfs-tools/0.18-ubuntu1/+build/3869162
[18:46] <smoser> does that need some "ack" ?
[18:46] <smoser> it has been built but not copied to archive for ~ 1 hour
[19:03] <Daviey> smoser: yes, you can see because https://launchpad.net/ubuntu/+source/cloud-initramfs-tools/0.18-ubuntu1 shows (New)
[19:03] <Daviey> smoser:  cloud-initramfs-dyn-netconf_0.18-ubuntu1_all.deb
[19:03] <smoser> right. that is a new binary package.
[19:04] <Daviey> smoser: accepted
[19:34] <hallyn> zul: adam_g: hey, have either of you ever run into bug 985489?  if so, can you also repr it with lxc, or only kvm?
[19:34] <zul> hallyn: i havent but i dont run a canonistack in my basement
[19:37] <stgraber> smb, SpamapS: flushing the route cache in /proc/sys/net/ipv4/route/flush would likely only work if done from the right network namespace
[19:37] <stgraber> so the idea might still work, it's just that you'd need to run that against the "stuck" network namespace, which is relatively tricky to do (especially as it's in a pretty weird state, so not sure we can actually attach to it)
[19:38] <hallyn> zul:  i thought you did
[19:38] <TLoT> anyone here want to help confirm/refute a php5-fpm bug?
[19:39] <TLoT> just asking because it being confirmed determines if i upstream it to debian :p
[19:39] <zul> hallyn: nope
[19:40] <TLoT> https://bugs.launchpad.net/bugs/1059272  <-- for those who want to help see if this actually happens
[19:40] <lamont> anyone want to claim significant dnsmasq knowledge?
[19:40] <hallyn> smb: have you tried (a) bisecting or (b) reproducing the bug with and without the commit removing the routing cache?  (i.e. at commits 89aef8921b and 89aef8921b^) ?
[19:41] <hallyn> zul: anyway i'm trying to decide if i need to try and do a full kvm based cluster to test (hoefully i can do that with 2 laptops?) or if i can do it with lxc with just one
[19:41] <hallyn> i guess i'll try with lxc and see what i get
[19:42] <zul> hallyn: hmmm...i can try to reproduce it as well
[19:46] <hallyn> zul: that'd be great.  i suspect it's present in precise and fixe din quantal.  (bc several nastybugs were solved there)
[19:48] <zul> yeah
[19:57] <ziggyzero> Hi does anybody know anything about libxml2?
[19:59] <ziggyzero> I have run apt-get install libxml2 but it has not given me the file I need.
[19:59] <ziggyzero> I was hoping to have /usr/lib/libxml2.so
[20:14] <hallyn> ziggyzero: maybe /usr/lib/x86_64-linux-gnu/libxml2.so.2.8.0 ?  check 'dpkg -L libxml2' to see what files it ships
[20:17] <ziggyzero> hallyn: Thanks, found it at /usr/lib/x86_64-linux-gnu/libxml2.so
[20:30] <SpamapS> stgraber: is that something we could work into lxc-stop as a workaround?
[20:31] <smw_> Hi all, how can I regenerate the default config for a package?
[20:31] <smw_> I deleted my /etc/apache2 dir and I want to make it the default again
[20:32] <stgraber> SpamapS: not in lxc-stop but in lxc-start, probably. (lxc-stop essentially just pokes lxc-start asking it to kill the container)
[20:32] <stgraber> hallyn: ^
[20:32] <SpamapS> smw_: its not generated, it is stored in the package.. you can remove the files and dpkg -i --force-confmiss file.deb ...
[20:32] <SpamapS> stgraber: *ah*
[20:32] <stgraber> hallyn: having lxc-start flush the routing tables of the netns (if not sharing the host's) when destroying the container
[20:32] <smw_> SpamapS, I don't have the deb file though...
[20:33] <SpamapS> stgraber: so perhaps as something to do right after tearing down the network, the route cache could be flushed? It seems like a reasonable workaround given the impact and nature of the kernel bug (and how close we are to kernel freeze.. ;)
[20:33] <smw_> SpamapS, and yes, I know it is not generated, I just want it to pretend it is installing apache2 for the first time
[20:33] <SpamapS> smw_: you should be able to get it. It may even still be in /var/cache/apt/archives
[20:33] <smw_> ah, right
[20:33] <smw_> I guess I can just unzip the deb and install the etc files manually
[20:35] <SpamapS> smw_: actually for that you can do  dpkg -i --force-confask --force-confnew apache-common_x.y.z-a_arch.deb
[20:36] <smw_> SpamapS, it works, thanks
[20:36] <smw_> sudo dpkg --force-confmiss -i apache2.2-common_2.2.22-1ubuntu1_amd64.deb
[20:36] <hallyn> stgraber: worth a shot
[20:37] <hallyn> stgraber: note that won't help in.vsftpd (which *should* have the problem too)
[20:37] <hallyn> perhaps we can do it in the kernel, at exit_net_ns
[20:38] <SpamapS> that seems more comprehensive
[20:38] <SpamapS> essentially if we know where the route cache causes problems, just flush it right after that point
[20:40] <hallyn> smb: ^ do you know offhand how you would do that?
[21:17] <mkeys> I have a ntfs formatted usb hard disk, /dev/sdc1 at the moment, and want to automount it at boot. It seems I must do a manual replug of the device for it to "see" the disk, it automounts it as /media/Seagate when I do. I snagged the uuid of the disk and put it in /etc/fstab, but like I said it doesn't see it without a manual replug and I'm forced to "press S to skip" at grub boot. Any suggestions?
[21:18] <sarnold> mkeys: if you set the 'auto' column to '0', you'll at least avoid the "Press S to Skip" prompt. Perhaps it'll even magically mount once the USB modules have loaded? (perhaps they load after fstab has been parsed?)
[21:20] <mkeys> fstab line is : UUID=D6D4E78DD4E76E65 /usbhd ntfs defaults 0 0
[21:20] <mkeys> uuid looks strange compared to the others, is that normal?
[21:20] <sarnold> that _is_ a funny looking UUID
[21:21] <mkeys> from blkid : /dev/sdc1: LABEL="Seagate Backup Plus Drive" UUID="D6D4E78DD4E76E65" TYPE="ntfs"
[21:26] <mkeys> mkeys@server:~$ sudo grub-probe -d /dev/sdc1 -t fs_uuid
[21:26] <mkeys> D6D4E78DD4E76E65
[21:27] <sarnold> mkeys: is sdc1 mentioned in your grub configuration?
[21:27] <hallyn> stgraber: SpamapS: we can't exactly flush the caches from end of lxc-start, of course, bc 'exec' doesn't return.
[21:29] <stgraber> hallyn: hmm, right, so kernel would be our best bet (though I'm not fond of the idea of getting a workaround like that in the kernel)...
[21:29] <stgraber> SpamapS: I guess you could at least check that putting the flush line that smb gave you in a very late upstart job (in the shutdown sequence) indeed "fixes" it for you
[21:30] <stgraber> SpamapS: having a job that brings down all interfaces (ifdown -a) and then flushes the routing table should avoid the issue (if it's indeed where we think it's)
[21:30] <mkeys> sarnold : not that I can find
[21:30] <hallyn> stgraber: well, we'd be saying "flush the routes from this netns", since the netns is going away presumably that should be done anyway
[21:31] <sarnold> mkeys: makes me wonder if the UUID is being used at all by the tools. (Since it doesn't match the usual UUID format...) Perhaps try a /dev/disks/by-id/... link in the fstab instead?
[21:32] <SpamapS> stgraber: hm. so if that doesn't run and we forcibly stop a container.. there's no recovery except reboot. :(
[21:32] <stgraber> hallyn: indeed
[21:33] <stgraber> SpamapS: yeah, you'd have to live with lxc-shutdown for the time being
[21:33] <hallyn> smb` is probably eod, lemme go see if i can figure out whether what i'm saying even amkes any sense
[21:33] <mkeys> sarnold : so happens the uuid is the same in /dev/disk/by-uuid for sdc1. in by-id there's a wwn id for it
[21:33] <mkeys> wwn-0x5000cca34dd7edd6-part1 -> ../../sdb1
[21:33] <mkeys> whoops wrong one
[21:33] <mkeys> wwn-0x5000c50049a5812c-part1 -> ../../sdc1
[21:37] <sarnold> mkeys: aha, seagate :) http://en.wikipedia.org/wiki/World_Wide_Name
[21:37] <sarnold> (not that that means anything. It's just neat to find.)
[21:39] <mkeys> sarnold : works ok once it boots to desktop i can partprobe, mount -a, and then /usbhd/* exists
[21:40] <SpamapS> stgraber: thats pretty ugly. :-/
[21:40] <sarnold> mkeys: that still sounds like a lot of work. :/
[21:40] <mkeys> don't know how to force grub/kernel to look for it at boot
[21:40] <SpamapS> hallyn: "netns is going away" ?
[21:41] <hallyn> SpamapS: when the contaienr goes away, the netns goes away
[21:43] <mkeys> sarnold : hmm http://www.novell.com/coolsolutions/tip/17785.html
[21:44] <sarnold> heh, I don't think I've ever added 'hotplug' to an fstab line before..
[21:45] <hallyn> SpamapS: something like http://paste.ubuntu.com/1254876/
[21:46] <mkeys> must be a suse thing
[21:46] <mkeys> Oct  1 17:45:51 server udevd[3783]: invalid rule '/etc/udev/rules.d/99-mount.rules:2'
[21:49] <SpamapS> hallyn: makes sense to me
[21:51] <hallyn> SpamapS: doing a test build;  i suspect the main question is: do we get to the point of freeing the struct net, and do that leaving route entries orphaned, or do we not get to that point (in which case my patch won't help) bc of the left-over entries?
[21:51] <hallyn> but it's worth a shot.  will get back to you
[21:52] <SpamapS> hallyn: let me know if you want me to test
[21:52] <SpamapS> hallyn: you can reproduce w/ lxc-stop right?
[21:54] <hallyn> SpamapS: only once in a long while.  i'll shout if it works at all so you or stgraber can test :)
[21:55] <mkeys> testing hotplug, brb
[21:55] <SpamapS> hallyn: I get it nearly every time
[21:57] <hallyn> SpamapS: well while my build test runs, i went ahead and pushed to ppa:serge-hallyn/lxc-natty (figuring that wont' step on anyone's toes :)  so if it works it should be available there "in awhile"
[22:16] <mkeys> ok, i take that back. partprobe does not see it. manual unplug/replug and then it shows up in dmesg
[22:18] <mkeys> is there a way to force a usb bus scan?
[22:19] <mkeys> I'll try modprobe usb-storage manually, brb
[22:40] <lamont> smoser: around?
[22:42] <mkeys> still have no idea how to force a bus rescan. http://pastebin.com/27MNbwFd
[22:56] <sarnold> mkeys: hrm, try udisks --poll-for-media /path/to/dev ? Perhaps that can poke the disk in the eye and get its attention?
[22:58] <protoCall7> Hi All, I am working with a Cobbler setup which has created a local repository mirror (via debmirror) for me, and I would like to add custom packages to it, however all of the documentation I'm finding is leading me to believe that this isn't really possible
[22:58] <protoCall7> could someone either confirm that, or point me in the right direction to research the proper way to go about doing this?
[23:02] <sarnold> protoCall7: probably easier is to run a separate repository for your own packages
[23:02] <sarnold> (since Releases.gz is signed and all that, trying to inject packages into a mirror sounds like more work than it is worth -- but setting up a new repository is not horrible)
[23:03] <protoCall7> sarnold:  That is accomplished using reprepro, correct?
[23:04] <sarnold> protoCall7: no idea about "reprepro" :) sorry
[23:04] <protoCall7> no worries, i
[23:04] <protoCall7> ve at least got a starting point for research
[23:04] <protoCall7> thanks!
[23:04] <sarnold> have fun :)