/srv/irclogs.ubuntu.com/2012/02/28/#ubuntu-server.txt

wondermananyone around to help me with server crashing a few hours ago? related to apache i think00:11
hallynstgraber: one other thing: the default fssize with -B lvm is 500M, which isn't big enough for the GINOURMOUS ubuntu-cloud image00:12
hallynso i'll probably bump that to 1G :(00:12
wondermanalso i need to know why on restart ubuntu sets my apache modules based on some 'default'00:12
hallyni dunno, maybe i need to detect the templates and make it conditional.  but that's fugly00:13
hallynwonderman: if it crashed, it's probbably better to open a bug - so that there's a place to collect all the relevant info00:14
wondermanhallyn what if its not a bug? but my configs00:14
hallyn<shrug>  if it's not a bug, it may get marked invalid, but not until we've been able to help you00:15
stgraberhallyn: 1G has been my default size for containers (vserver, vz, lxc) for years, so I'm fine with that ;)00:15
hallynstgraber: oh, ok, i'll do that then00:15
hallynwonderman: what sort of crash then?00:15
wondermanlet me pastebin some relavant00:15
wondermanOOM00:16
wondermanhttp://pastebin.com/XnirkVnd00:17
wondermanFeb 27 15:21:59 server1 kernel: [ 1321.513390] Out of memory: kill process 1595 (apache2) score 1225215 or a child00:17
wondermanit swapped :O00:17
=== sixstringsg|afk is now known as sixstringsg
uvirtbot`New bug: #942338 in lxc (universe) "Make default lvm blockdev size 1G" [High,Confirmed] https://launchpad.net/bugs/94233800:31
hex20decCan someone help me? I've installed postfix, imap, etc. I can send and receive emails when doing it from terminal but I tried connecting through thunderbird and it won't let me.00:56
twbThat sounds like a problem for #thunderbird00:59
twbProbably on irc.mozilla.org00:59
hex20decNo it's actually my server.00:59
hex20decThat's for sure.01:00
hex20decIt's something with imap01:00
hex20decI installed roundcube and it spits out "problem connecting to imap server"01:00
twbimap is a protocol, not a daemon.01:01
hex20decI realize that, I'm just telling you what is going on.01:02
hex20decAny idea what could cause such problem?01:02
twbWhat imap daemon are you using?01:03
hex20decI'm new to this, I just did what the guide told me to do, so how would I find that out?01:04
twbWhat guide?01:04
hex20decubuntu guide01:04
hex20dechttps://help.ubuntu.com/community/PostfixBasicSetupHowto01:05
hex20dechttps://help.ubuntu.com/community/Roundcube01:05
twbDon't follow that.  Couriour is old crap.01:05
twb*Courier01:05
hex20dec=[01:05
hex20decSo what do I do?01:05
twbThere is an official guide which uses dovecot, it is linked from topic for 10.04 version01:05
twbtopic as in /topic01:06
hex20decHuh?01:06
hex20decWhere is that guide?01:06
hallynwonderman: apache is the one that got killed, but it looks just as likely that cron executing /usr/lib/cgi-bin/awstats.pl -config=awstats -update >/dev/nul was what made you run out of mem01:07
hex20dectwb, I'm sorry, I'm new to it all.01:09
hex20dechttps://help.ubuntu.com/10.10/serverguide/C/dovecot-server.html01:10
hex20decNevermind, found it.01:10
hex20dectwb, do I need to uninstall all the stuff I just installed?01:11
twbIf you are running 10.10 that is the right URL yes.01:14
twbYou should probably uninstall courier stuff, uninstalling postfix is probably not necessary01:14
wondermani disabled that cron, but that was the default one installed by ubuntu when i installed awstats, and there is no site in that location or anything, i have another that runs and that hasnt crashed in the past01:25
wondermano well, ill see how it goes01:26
twbIsn't there something better than awstats by now?01:26
twbLike maybe google analytics01:26
twbIIRC when $boss wanted shiny graphs I said "go use analytics and stop bothering me"01:27
=== lifeless_ is now known as lifeless
=== sixstringsg is now known as sixstringsg|away
twbIs there a channel for apcupsd?03:02
patdk-lapman, the internet here is so bad :(03:28
patdk-lap300ms pings. and >50% packet loss03:28
stgraberhallyn: I just saw Daniel's e-mail about LXC 0.8, do you know if there's any patch that we still need to forward upstream?03:40
hallynstgraber: there probably are.  well there are some which aren't in bc they rely on our lxcbr0.  i need to check what others i haven't pushed03:41
hallynbbl03:41
hallynstgraber: http://paste.ubuntu.com/860057/   i think that's the list of not yet applied patches03:59
hallynno, 0048-warn-if-container-started is there04:02
=== sixstringsg|away is now known as sixstringsg
dindemanhey guys, I have a RAID 1 question04:48
dindemanI would like to know if my setup would allow hot swapping04:48
twbSATA drives are always hot-swappable; bays are not.04:55
twbmdadm should allow hot-swapping of a RAID1 node just fine, but you will need to ssh in and tell mdadm to fail/remove/add the node04:55
dindemantwb: thanks, sorry I left the keyboard for a moment05:25
dindemansorry I asked the question here because I didn't really know where to ask but my question was more purely hardware related (not so much to Ubuntu server)05:25
dindemanI'm indeed using SATA drives, it's a hardware RAID 1 on Intel Server Board S3420GP05:26
twbDon't use ICH10R raid05:28
twbIt's a fakeraid05:28
twbIn fact, unless you paid at least $200 for your RAID card, and it has a BBU or equivalent, it is useless fakeraid and you should stick to mdadm05:29
linociscohi all05:56
hex20decIt's been about 40 something hours that I'm awake trying fix this mail server and it keeps giving me problems, I need help desperately, is anyone willing to help set it up the right way? Please don't ask what is the problem, because I messed with so much settings that nothing is working now.05:56
dindemantwb: thanks agaih, let me Google this ICH10R you mentioned06:10
twbdindeman: ICH10R is your southbridge controller06:10
twbdindeman: i.e. the thing on your motherboard that gives you raid06:10
dindemanoh k06:10
dindemanso sorry, what's the issue with this controller ?06:10
twbit's fakeraid06:12
dindemanso you're telling me I'd better off use software raid instead of that ?06:13
=== aarcane_ is now known as aarcane
dindemantwb: thanks for your tips anyway, since this will be a ESXi host I might pursue with VMware peepz to see what's their take on raid when it comes to ESXi06:20
humungulousoffhand i do not believe esxi support intel storage array06:21
humungulous"intel storage matrix" sorry06:21
humungulousyeah http://communities.vmware.com/message/1876032?tstart=006:22
dindemanthe Matrix RAID from Intel ? (checking your link)06:22
humungulousyeah the raid onboard the ICH chipset06:23
dindemanthere's actually two06:23
dindemanESRT2 and Matrix06:24
dindemanyour link sort of mentions Matrix, ESRT2 seems more recent06:24
dindemanI can just try any, install ESXi and see whether there is some support about RAID from the vSphere client06:25
humungulousyou'll know if it wont work if the installer exits with an screen that says no supported storage adapter found06:26
humungulous(to point out the obvious)06:26
dindemanah ok06:27
hex20decAny mail experts here?06:27
twbIf it's esxi you can't use mdadm06:28
dindemanhumungulous: what about hot swapping ?06:28
twbI assumed since you were here, you were running ubuntu on it06:28
twbesxi is its own crackhead linux distro06:29
hex20dec dindeman: He left the channel.06:29
dindemantwb: sure, my mistake, I wanted to ask general questions about raid in fact and wasn't too sure where to ask06:29
twbNo problem06:29
dindemanI will pop by #vmware and see what's peepz take on this over there06:30
twbEither buy an expensive RAID card, use mdadm (you can't on ESXi), or use shit raid and deal with when it inevitably falls over and dies06:30
dindemantwb: what about hot swapping tho ?06:30
linociscowhat is the meaning or purpose of "sudo pkill -9 squid"  ?06:30
twbdoubt ich10r supports it06:30
twblinocisco: it sends a KILL to all processes named squid06:31
twblinocisco: see man 7 signal06:31
linociscotwb, when should I run this command , after stoping squid or starting squid?06:31
dindemantwb: thanks tho for your input on this, bye06:31
twblinocisco: never06:32
linociscodont understand06:32
twblinocisco: who told you to run it06:32
=== sixstringsg is now known as sixstringsg|away
linociscohttp://www.youtube.com/watch?v=LnBG_LEvvVw told me07:14
linociscotwb, http://www.youtube.com/watch?v=LnBG_LEvvVw told me to run07:15
rickspencer3good morning all, what's the word on the street about beta 1?07:22
humungulousrickspencer3: https://launchpad.net/ubuntu/+milestone/ubuntu-12.04-beta-107:23
rickspencer3hi humungulous07:23
humungulousrickspencer3: possibly you can find people discussing it in #ubuntu+107:24
rickspencer3thanks for the suggestion humungulous07:25
koolhead17hi all07:51
=== smb` is now known as smb
linociscohi all, who have done setup for failover between two ISP links?07:59
SpamapSlinocisco: do you mean with BGP routing?08:05
SpamapSlinocisco: or something simpler?08:06
linociscoSpamapS, there is no specific protocols. ISP is not sharing anything with us. THey allow us only port 8080,443 and NSlookup for DNS 5308:06
SpamapSlinocisco: doesn't sound like much of a "service provider"08:07
linociscoSpamapS, I agree. They bought public IPs and they sells us private subnetted IPs08:08
linociscoHi08:40
linociscohow to download squid package and how to install ?08:40
sorenDaviey: I'm curious.. How long time does it take to install a box using netinst these days?09:24
Davieysoren: Honestly, i couldn't tell you from a wallclock perspective, but it's better than it was.09:27
Davieysoren: We have some ideas to make it faster for 12.10..09:27
sorenDaviey: Is it more like 2 minutes than 10 minutes?09:27
Davieyprobably ~409:28
sorenDaviey: Cool.09:28
Davieysoren: It can be done in the time it takes me to make a cuppa.09:28
Davieysoren: One idea we are thinking about for next cycle is using squashfs on the server :)09:29
Davieyie, like Desktop.09:29
sorenDaviey: I've come up with a little interesting hack that does it in ~30 seconds.09:29
sorenIt's all cheating, of course.09:30
Davieysoren: oh?09:30
sorenbut it works.09:30
sorenI'm using the cloud images.09:30
sorenand qemu-img from initramfs.09:30
sorenShoving the image onto the disk.09:30
sorenFetching it over http.09:30
Davieyheh, interesting concept... how big is your initramfs?09:30
sorenGood questino. Let me check.09:31
Davieysoren: Couldn't you loopback mount the .img, and use that as the rootfs?09:31
Davieymaybe pivot_root09:32
sorenDaviey: It costs ~3 MB in the initramfs.09:33
sorenYes, but then I'd have to use raw images.09:33
sorenRight now, I'm using a qcow2.09:33
soren...which is loads faster on account of not having to transfer all the unused bits of the original disk.09:34
Davieyright09:34
Davieysoren: is it persistent ?09:35
sorenAs in?09:35
Davieysoren: save data across reboots?09:36
sorenDaviey: WEll, sure. It's for installing, not for "live CD"-ish sort of stuff.09:37
sorenI don't use the image after it's done installing.09:38
sorenIt boots into the initramfs, fetches the image and writes it to disk, and reboots.09:38
Davieysoren: so do you dd?  Gah, where is the script? :)09:38
sorenDone.09:38
sorenNo, I use qemu-img.09:39
sorenqemu-img convert http:/blah -O host_device /dev/sda09:39
soren-ish.09:39
Davieysoren: I think i prefer the squashfs-in-di approach TBH.09:39
Davieysoren: then you touch users and hostname?09:40
sorenI don't do any of that yet.09:40
DavieyHmm, that might be inappropriate09:40
soren...but I expect to try to rely on cloud-init for stuff like that.09:40
Davieysoren: right09:40
Davieysoren: There will be a UDS session purely about this.09:41
sorenDaviey: I'm not sure I understand you squashfs-in-di idea.09:43
Davieysoren: Same as ubiquity on the desktop, but for alternate/di09:45
sorenDaviey: How would that work over the network?09:45
Davieyso a satic image that gets unpacked to disk, and customised based on instaler options.09:45
Daviey(ie users / hostna,e / network config)09:45
Davieysoren: I imagine the netinstaler would have to wget the squashfs for mini.iso.09:46
* Daviey adds some extra l's09:46
sorenWhere would it put it?09:47
soren"I don't know" is a valid answer :)09:47
DavieyI have an educatumacated guess, but i don't know09:47
Davieysoren: Colin sees it as viable fwiw09:48
sorenI see it working for ISOs.09:48
sorenI just wonder how you'd do it over the network.09:48
Davieysoren: I imagine the squashfs would be either on cdimage/releases or the same part of the archive which houses the mini.iso09:49
sorenDaviey: Perhaps I'm missing something.. You need to mount the squashfs somewhere and then copy it to a fresh filesystem... right?09:50
Davieysoren: Hmm, well i don't want to speak for Colin... but the way i see it.09:50
Davieymini.iso gets you into d-i09:50
Davieywget http://squashfs09:50
Davieydd squashfs to /target09:50
sorenStoring it where?09:50
uvirtbot`New bug: #942487 in openssh (main) "scp -f option not documented" [Undecided,New] https://launchpad.net/bugs/94248709:50
sorenSorry, go on.09:51
Davieysoren: You don't think we have enough ramfs these days?09:51
Davieycustomise /target based on preseed09:51
Davieysoren: I suppose the other option is that the squashfs is exposed via iscsi|nfs .. OR part of the server-mini.iso (but that is no longer mini :)09:52
sorenMy motivation is: I want to unify the way I provision cloud instances and metal. I.e. shove an image onto a disk and rely on cloud-init to do the customisation. I get annoyed having to deal with two completely separate systems for this stuff.09:54
soren...and I think we've got the whole shared-image-followed-by-first-boot-config solved pretty well with cloud-init.09:55
sorenSo why not use it?09:55
sorenI'm having some networking problems I need to solve, but it's mostly functional.09:55
sorenAnd then I need to find a few to feed data to cloud-init.09:56
koolhead17zul: essex-3 has some new python deps, http://paste.ubuntu.com/860317/  I am finally able to get it working from source :)09:56
Davieysoren: right!09:56
koolhead17i meant keystone essex-3 :P09:57
mandelhello, does anyone know why the squid package on ubuntu was not compiled with the --enable-ssl flag? is there a reason, or should I just compile own with it? I want to be able to use the https_port directive in the configuration10:04
mandelhttp://www1.it.squid-cache.org/Versions/v3/3.HEAD/cfgman/https_port.html10:04
lynxmanmorning o/10:05
lynxmanmandel: the squid package comes straight from debian, I'd suggest building your own package in your PPA, filling a bug and attaching a patch to it10:06
mandellynxman, you could have said that too me yesterday, puto!! ;)10:06
lynxmanmandel: it was just too fun replying "lol" instead ;)10:07
lynxmanmandel: if you need any help let me know, I'll be glad to build the package for you10:07
mandellynxman, don't say it to loud or I'll delegate hehe10:08
lynxmanmandel: that's why I'm offering, I like being your helper ;)10:08
mandellynxman, if is not too much work for you, I'd really appreciate it10:09
lynxmanmandel: it's really not, let me fix it up fo ryou10:09
mandellynxman, I'll get you a beer whenever I go to london or barcelona :)10:09
mandellynxman, thx!10:09
linociscoduring an installation with apt-get install, and if I press Ctrl+C to cancel. what will be the effect in system?10:10
lynxmanlinocisco: packages half installed, mayhem, dogs and cats living together10:12
lynxmanlinocisco: but just issue "apt-get -f install" and it should just catch up10:13
linociscook10:13
linociscothanks10:13
linociscocan I clear traces of that cancel by using apt-get -f install ?10:13
lynxmanlinocisco: most of the time it should yeah10:14
linociscolynxman, thanks10:15
maxagazhi10:24
maxagazI always get this error in my auth.log: Feb 28 11:23:49 catacombe sshd[18795]: Failed password for root from 207.183.255.142 port 38861 ssh210:26
maxagazwhat does it mean ?10:26
maxagazsome bot trying to log on my server ?10:27
greppymaxagaz: most likely, yes.10:29
maxagazgreppy, what can I do against that ?10:30
davepigottDaviey: zul or adam_g: jamespage gave me your contacts because I'm having problems with my openstack implementation and he thought you might be able to help. I have four servers, one running all the nova, glance and swift services (which I'll refer to as the controller), the others just running nova-compute and nova-network. I've got two nics per server, using a vlan config. If my instance runs up on the10:30
davepigottcontroller I can ssh to it and ping to it. If it ends up on any of the compute nodes, it's running but I can't connect to it. The console log shows it waiting for 120 seconds for a network connection, which it never gets. The logs are here: nova-compute.log: http://pastebin.ubuntu.com/860301/ nova-network.log: http://pastebin.ubuntu.com/860304/ Any clues?10:30
maxagazgreppy: it's polluting my logs10:30
maxagazI have many things like "Invalid user manikswe from 204.57.82.8"10:31
greppymaxagaz: you can do a couple of things, install fail2ban, block that IP using iptables, notify the owner of the IP of the attempted unauthorized access, move your ssh daemon to listen to a non-standard port, like 1422 instead of 22.10:31
maxagazwith different nicks, but same ip10:31
maxagazgreppy: great :-)10:33
maxagazgreppy: fail2ban works like a charm, I forgot it :)10:33
maxagazI don't know why it's not installed by default on ubuntu server10:34
linociscohi10:34
linociscoin nano, how could I see line no.?10:34
greppymaxagaz: because where do you draw the line then?  what else should be installed by default? :)10:35
greppylinocisco: have you read the man page for nano?10:35
maxagazgreppy: hmm... fair enough10:35
linociscogreppy, yes10:36
lynxmandavepigott: I'd say the nova-network configuration is borked somehow, could you please paste it somewhere?10:36
lynxmandavepigott: also it'd help pinging the #openstack channel10:36
davepigottlynxman: Already pinged there but got nothing. Posted on the openstack forum as well. So far no response.10:37
davepigottlynxman: Do you mean the nova.conf file?10:37
greppylinocisco: use 'nano -c' to invoke nano.10:37
lynxmandavepigott: I'd say from my experience that there's something not right either in the vlan creation or the nova.conf10:37
* koolhead17 does whois lynxman to confirm if its really him!! :P10:37
davepigottlynxman: OK. nova.conf on its way. One moment10:37
lynxmandavepigott: cool10:38
linociscogreppy, no . not like that. like we can see line no. in gedit10:38
davepigottlynxman: http://pastebin.ubuntu.com/860354/10:39
davepigottlynxman: Same on all four servers10:39
greppylinocisco: not possible.  use gedit then.10:40
greppylinocisco: or use vi10:40
linociscogreppy, ok thanks. vi is more difficult than nano10:41
linociscogreppy, thanks anyway10:41
lynxmandavepigott: Have you looked at this? http://wiki.openstack.org/VlanNetworkSetup10:46
davepigottlynxman: No. First I've seen of it. I'll read through. You think my db is out of sync with the config?10:48
lynxmandavepigott: if you didn't do it then yeah :)10:48
davepigottlynxman: Fair enough. :)10:49
uvirtbot`New bug: #942541 in nova (main) "Short descriptions mistakes" [Low,New] https://launchpad.net/bugs/94254110:55
sorensmoser: How easily could you be pursuaded to extend cloud-init to accept a url on the kernel command line where it should go and find its user-data and meta-data?10:56
lynxmansoren: if you offer him candy or a bug report I'm sure he'll be happy to have a look :)11:03
lynxmansoren: how're you doing btw, all good?11:10
davepigottlynxman: OK. Database looks ok. I did do all that, based on the "Beginners guide" so it's fine. Or at least would seem to be.11:10
lynxmandavepigott: then I'd try to ping the openstack channel, I've never handled vlan connections on openstack myself :)11:11
wondermanhey, can anyone tell me why this cron is running everyminute, for 1 hour, then stops  > * */16 * * * curl --silent11:11
lynxmandavepigott: always used flat network11:11
davepigottlynxman: OK. Thanks. I probably should have used flat network but not sure how to switch it over.11:11
lynxmanwonderman: because it's programmed to do so at 4pm (16:00) for every minute of that hour11:11
lynxmandavepigott: hmm good luck :)11:11
wondermani see, how would i make it do it every 16hours?11:12
wondermanthis should do it every 24hours  >   0 0 * * * curl --silent11:13
wondermani also need one to run every 16hours11:13
lynxmanwonderman: 0 */16 * * * curl --silent11:13
jamespagedavepigott, how is the network switch configured?11:13
lynxmanwonderman: eitherway your mileage may vary, running something every 16 hours is a bit bizarre :)11:14
davepigottjamespage: Pretty much as it came out of the box. Cisco switch.11:14
wondermanhmm lol11:14
jamespagedavepigott, ah11:14
davepigottjamespage: ah?11:14
wondermanwell, what its doing atm is totally wrong!11:14
jamespagedavepigott, one second11:14
davepigottjamespage: sure11:14
jamespagedavepigott, I think that the switch needs some extra config11:14
davepigottjamespage: To allow vlan?11:15
wondermani thought this is what i was doing   0 */16 * * *   what is the difference, apart from the obvious, what does setting 0 mean over setting * ?11:15
lynxmanwonderman: 0 is minute 0 whereas * is every possible option in the minute, so every single min11:15
jamespagedavepigott, yes - the ports to your servers need to be trunk ports that support the VLAN's that you have configured nova with11:15
wondermanah ok11:15
jamespagedavepigott, I've not done that on cisco for a while11:16
* jamespage tries to remember11:16
davepigottjamespage: OK. Not over familiar with configuring cisco switches for that kind of thing.11:16
jamespagedavepigott, you might be better to not use VLAN mode and revert to one of the other networking types11:16
davepigottjamespage: Ugh. How big a deal is that to undo it all and then re-do it?11:17
davepigottjamespage: I have two switches. One on the main net, one on the private. Do they both need vlan config?11:18
lynxmandavepigott: I think this helps your config https://lists.launchpad.net/openstack/msg05531.html11:20
lynxmandavepigott: you'd need to configure every switch port that is configured in vlan mode this way11:20
* lynxman thankfully is good at Cisco so jamespage sparked his brain11:21
davepigottlynxman: How do I get onto the switch? Over serial it's all menu driven11:21
lynxmandavepigott: oh my :)11:21
lynxmandavepigott: you need to telnet into it, have a user or a term password then the enable password11:21
lynxmandavepigott: you don't have anyone around that did actually configure those switches or are you on your own?11:22
jamespagelynxman, that config looks good11:22
davepigottlynxman: I'm trying to telnet but it's not responding11:22
davepigottlynxman: On my own I'm afraid11:22
jamespagedavepigott, I think it only needs to be applied to the private network interfaces i.e. one of the switches11:22
davepigottjamespage: OK. That's good at least. :)11:22
lynxmandavepigott: what james page says11:22
lynxmandavepigott: then try to access through serial cable, it shouldn't ask you for any password and just give you a prompt11:23
lynxmandavepigott: then just type "enable" and the default password should be Cisco11:23
lynxmandavepigott: do "sh run" to show your running configuration, write down the ports that you'll be using for your internal network11:23
lynxmandavepigott: then add that config to them using "config term"11:23
lynxmandavepigott: once you're done type exit and "write mem" to save your changes11:23
* jamespage bows to the awesomeness of lynxman's cisco knowledge11:24
lynxmanjamespage: I used to manage spanning tree vlans on Cisco, but I'm rusty at best11:24
davepigottlynxman: It just gives me a menu. No shell11:25
jamespagelynxman, I had todo some integration work between IBM/HP BladeCenters and Cisco switches but that was in 2007 :-)11:25
jamespagedavepigott, if you don't need it I would suggest that you don't use VLAN networking11:26
jamespagedavepigott, this was what I was looking for - http://unchainyourbrain.com/openstack/13-networking-in-nova11:27
davepigottjamespage: If this solves it, I'll stick with vlan. If not I'll go back and reconfigure everything11:27
jamespagethis is what the juju charms for openstack do11:27
lynxmandavepigott: I do agree with jamespage, if you don't feel comfortable managing vlan switches it's better to go to plain basic flat network11:28
jamespagedavepigott, I don't think switching is that hard TBH11:29
jamespageyou already have the right bits installed - they just need reconfiguring!11:29
davepigottlynxman: Why am I menu driven and not command line on the switch?11:30
lynxmandavepigott: it depends on your cisco switch model, all Catalysts should run Cisco IOS but maybe you have a SoHo one that doens't11:31
davepigottlynxman: OK. There's a VLAN menu option. Has a VLAN ID of 1.11:32
lynxmandavepigott: that's the default vlan, the question is if the switch allows trunk config per port11:33
kahrnHi all11:33
davepigottlynxman: OK. If it doesn't vlan is out, right?11:33
lynxmandavepigott: afraid so11:33
kahrnI'm having some problems with my ubuntu server, would anyone be able to help?11:34
=== Leseb_ is now known as Leseb
kahrnI can't upgrade/use dpkg/apt.. specifically, I messed up Perl11:34
kahrnand it keeps complaining about strict.pm being missing, so I tried to reinstall, but it still complains11:35
kahrnit prevents me from using the package manager11:43
kahrnwhich is annoying as I need to upgrade the server11:43
kahrnhttp://ubuntuforums.org/showthread.php?p=1172494611:49
sorenlynxman: Good, thanks. Of course I'd be even better if cloud-init would do this for me :)12:03
davepigottlynxman: OK. Looks like I can set up trunk config per port from the menu interface12:21
sorensmoser: Oh, hang on. It already does this somehow, doesn't it?12:44
Davieyutlemming: hey, it looks like i won't be able to be at the weekly meeting.  Are you ok driving the development part?12:45
lynxmansoren: I'm sure it somehow does, cloud-init does the kitchen sink and everything12:52
ttxsmoser/utlemming: what's your take on including acpid in the cloud images in order to support KVM soft reboot ?13:13
ttxReferences:13:13
ttxhttp://wiki.libvirt.org/page/Tips#Debian.2FUbuntu_guests_under_KVM_don.27t_shut_down_properly13:13
ttxhttps://bugs.launchpad.net/nova/+bug/93955713:13
uvirtbot`Launchpad bug 939557 in nova "'nova reboot' under KVM always does a hard reboot" [High,Triaged]13:13
Davieyttx: I have never agreed that acpid should be absent.. iirc, soren had views on it.13:21
sorenDaviey: I certainly do.13:21
sorenDaviey: It should totally be included.13:21
sorenDaviey: That's my view.13:22
sorenDaviey: I also have views on how to get it included.13:22
Davieysoren: Hmm.. i'm *sure* you said that it shouldn't be there.13:23
DavieyInfact, i remember saying so.. and you saying just apt-get install it! :)13:23
sorenDaviey: When did you ever pay attention to what I say?13:25
sorenacpid is important. Include it. In -minimal or -standard or whatever. Don't backdoor it into the cloud images.13:26
DavieyAHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHH13:26
Picigezundheit13:26
ttxH.13:26
sorenHeck, or even in -server (which gets installed in cloud-init, iirc).13:26
Davieysoren: You change your mind like the weather man!13:27
sorenDaviey: Maybe.13:27
sorenDaviey: I'd be happy to see quotes of me saying otherwise.13:27
sorenDaviey: I've talked many times about *relying* on its presence.13:28
Davieysoren: UDS-Jaunty real life, and prior on irc. :)13:28
ttxHe was young then13:28
sorenDaviey: Where I've told people to just go and apt-get install it, because we can't travel back intime and add it to hardy or lucid or whatever.13:28
* jpds flips another coin.13:28
wondermanman, can someone tell me what causes lots of 408s in apache?13:29
Davieysoren: I'm going to add it to just server for now.13:32
sorenDaviey: I remember, for instance, a bug about rebooting kvm's from libvirt, where I spent way too much time explaining why you couldn't just rely on acpid for it to work.13:32
sorenDaviey: ...because even if acpid got added to the development release at the time, it still wouldn't work for older version of ubuntu.13:32
sorenDaviey: I can't imagine I'd be opposed to adding acpid by default.13:33
sorenDaviey: But, if I did, I'm sure I'd have excellent reasons :)13:33
Davieynaturally13:33
kaimorning folks13:35
* soren hugs kexec13:38
Davieysoren: seen pxe-kexec ?13:42
sorenDaviey: I did see it in the archive, yes. Didn't quite get it, though.13:43
sorenDaviey: Perhaps its name confuses me. Does it actually do PXE at all?13:44
Davieysoren: well, depends what you define as PXE :).. it pulls via tftp the conifg and kernel, then kexec's it13:46
Davieyso not quite PXE 'booting', but the same end resut.13:46
sorenDaviey: ...but I don't need a bootp server?13:51
sorensmoser: If I pass ds=nocloud on the kernel command line, isn't cloud-init supposed to not try to contact the EC2 metadata service?13:56
Davieysoren: it tries to find that from dhcp (unreliable), normally you declare it on the command line13:57
sorenDaviey: Ah, I see.13:58
sorenDaviey: PXE invariably means something that involves DHCP in my head.13:58
Davieysoren: Think outside the box (head).13:58
larsemilquestion about iscsi and lvms...14:01
larsemili have a raid, and i divide it using lvm.14:01
Davieylarsemil: Are you asking for us to ask you a question?14:01
larsemilmy initiators mount the different luns14:01
larsemilcan the server mounting the lun, use lvm (again) on the iscsi lun that is originally a logical volume14:02
larsemil?14:02
larsemilDaviey: there it came, the question14:02
Davieylarsemil: I can't see why not?14:02
Davieylarsemil: All you are doing is nested lvm, right?14:03
larsemilyes14:03
wondermansomeone please tell me how i can disable modules that apache load by default in ubuntu, there are too many, and once i disable, if i reboot server they all become enabeld again14:04
sorenwonderman: How are you disabling them?14:06
OZ8AAZwonderman: a2dismod is your friend14:06
OZ8AAZwonderman: ...and a2enmod to enable...14:09
wondermanyes i use that14:12
wondermani disabled mod_reqtimeout14:12
wondermanwhen i rebooted server it was back enabled14:12
sorenwonderman: So on reboot, /etc/apache2/mods-enabled/reqtimeout.load magically reappears?14:15
wondermanyes14:20
zulivoks: ping14:20
wondermanas well as other useless modules ubuntu seems to have enabled by default14:20
ivokszul: pong14:20
zulivoks: the nova bug you raised is with juju right?14:22
ivokszul: yes14:22
zulivoks: k14:22
ivokszul: why?14:22
zulanywhere you can point me to get juju working with nova?14:22
OZ8AAZwonderman: can you see the module in $apache2root/modules-available/ ?14:22
zulivoks: starting to work on it14:22
ivokszul: sure14:22
wondermannot now that i disabled it again14:23
wonderman..14:23
ivokszul: where do you wanna start?14:23
zulivoks: at the begging14:23
ivokszul: i didn't get it working 100%, but it's almost there...14:23
zulk14:23
davepigottjamespage: lynxman: I've now got instances I can ping, but unless its on cloud01 I can't ssh. Weird.14:31
zulDaviey: im going to try to  get the console patch accepted again upstream14:36
=== Leseb_ is now known as Leseb
sorenwonderman: Well, Ubuntu certainly isn't resurrecting files on the filesystem on its own. Is there anytihng special about this system?14:36
wondermanwhat about init scripts?14:37
wondermanare you 100% sure about that, because it is lol14:37
sorenwonderman: I'm sure we'd have heard about it before if it did.14:37
Davieyzul: awesome14:39
sorensmoser: ping14:42
lynxmansoren: maybe it's a spin called Zombibuntu? Where your files come back to life14:44
sorenlynxman: Could be :)14:44
sorenlynxman: I don't see how ds=nocloud is ever supposed to work.14:45
sorenlynxman: Sorry, not you :)14:45
sorensmoser: I don't see how ds=nocloud is ever supposed to work.14:45
sorensmoser: cloudinit.DataSourceNoCloud.get_datasource_list calls list_from_depends with net and disk depends. list_from_depends returns only data sources whose dependencies match exactly the list of dependencies passed (rather than the data sources whose dependencies are fulfilled by the list of dependencies).14:48
sorenThis matches the "docstring" for list_from_depends, but goes against my sanity as well as what you put in the commit message for r323.14:48
sorenThe net effect is that the NoCloud ds isn't considered, because both net and disk is available already.14:49
jamespagedavepigott, thats a step forward at least!14:49
jamespagehave you configured security groups to allow SSH access?14:49
davepigottjamespage: On the server? Yep.14:54
davepigottjamespage: And just found that by default the switch blocks ssh. Changed that but still can't ssh. Hmm. Unless I need to restart networking across the cloud14:55
uvirtbot`New bug: #942646 in nova (main) "No logrotate files for nova-manage and nova-dhcbridge" [Undecided,New] https://launchpad.net/bugs/94264614:58
jamespagedavepigott, I mean have you configured a security group and associated it with your instances?  If thats what "On the server" means then +114:58
davepigottjamespage: Hmmm. Well, I did an euca-authorize on tcp port 22, if that's what you mean?14:59
jamespagedavepigott, thats what I mean :-)14:59
davepigottjamespage: Then yes. :)14:59
davepigottjamespage: Going to restart the services on the cloud and see if they recognise the change on the switch15:00
smosersoren, nocloud is run at 'cloud-init start-local'15:00
smosersmb, ping15:00
sorensmoser: Err.. ok?15:01
smosersoren, i'm fairlys ure it does work. i can test to make sure though.15:01
sorensmoser: Not sure what to make of that. All I can tell is that if I pass ds=nocloud on the kernel command line, the vm still spends a lot of time trying to contact the ec2 md service.15:01
smosercloud-init has 2 main "init/datasource" jobs15:01
smbPotentially I currently have shared troubles with soren15:01
smosersoren, ah.15:01
smosersoren you're hoping to completely use the command line ? ie, no /var/lib/data/seed?15:02
smoseryeah... that is potential, smb.15:02
smoseri will check to make sure i didn't break that. hold on.15:02
sorensmoser: I haven't gotten that far yet. Right now I just want it to stop bothering with the ec2 md service.15:03
* smb is using /var/lib/cloud/seed though (without ds=nocloud but have checked with it)15:03
smosersoren, right.15:03
smosereasiest thing in precise is to attach an iso or vfat formated disk15:04
smoserwith 'user-data' and 'meta-data' files.15:04
wondermananyone help with diagnosing 408 errors?15:04
sorenFor a VM, perhaps.15:04
smosersoren, right.15:04
sorenI need to deal with metal, too.15:04
smoserbut if you're doing an install and just want to ditch it, easiest thing to do would be to seed and dpkg-reconfigure.15:04
smoseroh..15:04
smoserbut you dont want to ditch.15:04
smoserk.15:05
smoserjust a minute15:05
sorenPassing ds=nocloud-net works, too, but ds=nocloud should work.15:05
smosersoren, yeah, clearly it should.15:06
smosersmb, on bug 93735215:06
uvirtbot`Launchpad bug 937352 in cloud-initramfs-tools "root partition may not be grown" [Undecided,In progress] https://launchpad.net/bugs/93735215:06
smoseri woke up loast night night with an idea.15:06
smoserwhich may not be valid, but wanted to run it past you15:06
sorensmoser: It'll pass.15:06
smbsmoser, Hm no wake up...15:06
smoserslangasek pointed out that when you call sfdisk to resize, it calls the ioctl once at the beginning (to see if its busy) and then once again.15:07
smoserhis suspicion is that the first call (even though nothing has changed) is generating kernel events , spawning udev events...15:07
smbOops, that might be slightly fatal15:07
smoserbut if there is no change, there *should* be no events generated15:07
smoserbut my thought last night was this:15:08
smoser * kernel comes up, sees /dev/vda (and vda1), and gets CHS from disk/driver somehow15:08
smoser * udev runs15:08
smoser * partition scanning code runs, and notices that the partition table seems to have a CHS that does not match that of the disk/driver15:08
smoser * sfdisk runs with ioctl to say "re-read"15:09
smoser * kernel now says "oh my, its different than it was"15:09
smoser * udev gets event15:09
smoser * blkid ... BUSY15:09
smoserdoes that seem possible ?15:09
smoserit seemed strange tome that the kernel could potentially change its view of CHS for a disk, but maybe that udev somehow tried to "fix" it.15:09
smosersmb, ^ is that remotely possible ?15:10
smoserwe've generally ruled out the mount/unmount yesterday as I saw there were still issues right before the first mount15:11
smbsmoser, I actually think there is not so much complicated code. You / sfdisk calls reread partition table, the kernel does it and generates events. No checking for whether this is the same as before15:11
smosersmb, well, that is not what i see.15:12
smoservery clearly, just calling blockdev --rereadpt does not generate udev events.15:12
smoserthat is easily demonstratable15:12
smbsmoser, Hm, it was what I saw when just using that blockdev --rereadpt command instead of sfdisk for testing...15:13
smoseri thought we'd seen that too15:13
smoserbut i just tested it yesterday15:14
smbsmoser, So probably I go and look at the code. Then we should be sure. :)15:14
gsuesshey everyone. any way to get apache kerberos auth working with a require group auth?15:16
gsuessdirective^15:17
smoserie, 'stop udevd' 'udevd --debug' 'blockdev --rereadpt /dev/vdb', nothing new on udev screen15:17
smosersmb, ^15:17
smoserijust tested that now15:17
gsuessi keep getting access to / failed, reason: require directives present and no Authoritative handler.15:18
gsuessit works with require user ...15:18
gsuessrequire valid-user doesn't work either.15:18
gsuesssame error.15:18
smbsmoser, I had been adding function trace points to blkdev_get and blkdev_put and saw them used by blkid and cdrom-id, at least when running rounds in a loop15:19
smosersmb, so something is strange then15:20
gsuessany ideas?15:21
smbsmoser, yes agree. Well give me a few minutes to check the code in the kernel. And try some modified tracing15:22
smbsmoser, In the mean time for clarification: to make cloud-init use the files in /var/lib/cloud/seed/nocloud, I need to put ds=nocloud on the grub command  line, too?15:23
smosersmb, you do not.15:24
smbsmoser, Hm, ok. Still it does not seem to want to use the files I got there. Is there a way to make it more verbose?15:27
smosersmb, soren, it seems lke precise might have regressed that :-(15:27
smbOh15:27
smosersmb, soren precise is regressed for both of you :-(15:34
gsuesswhat are the ./configure args used for ubuntus apache?15:35
smbsmoser, Good to know. At least I can now stop wondering what I did wrong. So for now I add precise to my decloudify list :)15:36
smoserhttp://paste.ubuntu.com/860653/15:37
smosersmoser, there is a fix15:37
smoserand i just verified with that change that both populating /var/lib/cloud/seed/nocloud15:37
smoserand15:37
smoser sudo DEBUG_PROC_CMDLINE="BOOT_IMAGE=/boot/vmlinuz-3.2.0-17-virtual root=LABEL=cloudimg-rootfs ro console=ttyS0 ds=nocloud;h=foo;i=i-abc" cloud-init start-local15:38
smoserfind the local data source15:38
smosersoren, ^15:38
smbsmoser, Yes, it does seem to work now15:40
smoserbug 94269515:44
uvirtbot`Launchpad bug 942695 in cloud-init "cloud-init local data source broken" [Undecided,New] https://launchpad.net/bugs/94269515:44
uvirtbot`New bug: #942695 in cloud-init (main) "cloud-init local data source broken" [Undecided,New] https://launchpad.net/bugs/94269515:56
Davieybroken, smoken15:56
smbsmoser, I checked the code and did some experiments. Both does verify that there will be events for every rereadpt done. Basically this removes all partitions and creates fresh ones after scanning15:56
smbsmoser, So if sfdisk really does that ioctl twice when resizing there is really a good chance of failing because some events still may be running15:59
uvirtbot`New bug: #928383 in glance (main) "python-glance package contains stuff about API and registry" [High,Invalid] https://launchpad.net/bugs/92838316:42
hallynstgraber: you're not touching ubuntu:lxc this week at all?  (wondering how frequently i should check for updates :)16:56
stgraberhallyn: nope, busy fixing the installer this week :)16:56
hallyncool.  then lemme put on my irc blinders and get to work17:00
stgraber:)17:01
=== sixstringsg|away is now known as sixstringsg
smosersmb, i dont understand how you could have seen that.17:13
smoserwhat was wrong with my test ?17:13
smoservery explicitly i saw no output of udev17:13
smbsmoser, You tried to start the udevd in the foreground with debug, right. Wonder whether that is really the same as logging the events in debug mode...17:14
smbsmoser, Have you tried with "udevadm control --log-priority=debug" and then looking in /var/log/syslog?17:15
smoseri can try that really quick17:17
smosersmb, nothing17:18
smoser(i only see stuff in /var/log/udev, not /var/log/syslog though)17:19
smoserthis is precise17:19
smbsmoser, I am also looking at precise. A cloudimage running on my local xen17:19
smosersmb, here, try my instance17:20
smoserubuntu@10.55.60.14817:20
smosersmoser, tell me what i'm doing wrong17:20
smoseri'm just running 'sudo blockdev --rereadpt /dev/vdb'17:21
smoserand expecting to see output in /var/log/udev17:21
smoserbut there is none17:21
smbsmoser, There is something else broken. I do not even see any output from the rereadpt... Checking for a partition...17:24
smosersmb, hm...17:24
smoserwell, what is wrong then?17:25
smbsmoser, not sure right now17:26
smbbut clearly /proc/partitions has not seen your vdb117:26
smosersmb, if you are correct, then sfdisk without --no-reread is generally broken17:26
smoserhm..17:26
smoserinteresting17:26
smoseri fdisk'ed it17:27
smoserthats very interesting.17:27
smbNormally there should be messages in dmesg like vdb: vdb117:27
smoserthats just a new instance, and then i ran 'fdisk' to put a partition table on it.17:27
smbnothing there17:27
smoserwell, the partition table wasn't ther eon boot17:27
smbjust like the ioctl somewhere is ignored17:27
smoseri added it17:27
smosersmb, you can have that instance.17:27
smoseri think that is something you should debug17:28
smoseras it may be related here.17:28
smoserand we should open a bug on sfdisk, as if you're correct, it is broken in its design.17:28
smosersmb, in your xen guests, you have a different driver than in canonistack instance17:29
smbsmoser, hm, ok the ioctl returns 1 not 0 , just need to find out what that means again17:29
smosersmb, k. i'll leave you at that for a minute and chase these other 2 cloud-init bugs17:29
smbsmoser, aye :)17:29
smoserutlemming, jamespage Daviey i'm hoping that i'll be able to fix bug 941955 and bug 942695 before beta17:30
uvirtbot`Launchpad bug 941955 in cloud-init "util.islxc() method is broken without is-lxc-container" [Medium,Triaged] https://launchpad.net/bugs/94195517:30
uvirtbot`Launchpad bug 942695 in cloud-init "cloud-init local data source broken" [Undecided,New] https://launchpad.net/bugs/94269517:30
smoserboth should be straight forward enough.17:30
smbsmoser, Harrr! You had the device mounted. So rc 1 mean busy17:34
smosershoot.17:35
smosersorry, smb.17:35
smbsmoser, Now there are all the missing things :)17:35
smoserso sfdisk is simply broken and almost guaranteed to cause races17:35
smoserwithout --no-reread17:35
smoseri'll open a bug, and will address in cloud-initramfs-tools17:36
smbsmoser, It seems so. At least since rereading the partition table actually causes events and lots of things to happen17:36
smbTrying to use rereadpt to check for a busy disk is nearly guaranteed to have it busy in some way for some time after that17:36
* smb thinking whether that could be replaced by an open with O_EXCL17:37
=== Leseb_ is now known as Leseb
smbthough that may required an open to all partitions, not only the main block device. But I am not sure17:39
kklimondasmb: udevadm settle won't block until after rereadpt is done (and the disk isn't busy anymore)?17:40
smbkklimonda, It just ensures that the outstanding requests from a previous rereadpt are finished. Either sfdisk does something like that internally or one would need to use --no-reread and do any initial check before17:41
smbudevadm is not particularly related to the disk17:42
tyskahi guys, someone already used ldap to auth users on ubuntu with the automount of home directories?17:42
tyskai need some advice to do that17:43
kklimondasmb: yeah, but it tends to block while it's processing events, and it did help me when sfdisk was complaining that the disk is in use ;)17:45
kklimonda(but if sfdisk has a bug then it won't help obviously)17:45
smbkklimonda, Yes it does. Just that we found out that sfdisk does the rereadpt twice internally17:45
smbOnce before doing anything to make sure the disk is not in use and once after changing the partition17:46
kklimondahuh, and the second one bails with an error?17:51
smosersmb, still around ?17:54
smbsmoser, yes17:54
smoserso:17:54
smoserprintf "1,,L,*\n" | sudo sfdisk /dev/vdb17:55
smoserthat should basically guarantee failure at some point, right?17:55
smbsmoser, not sure I know sfdisk well enough to understand the instructions...17:56
smoserwell, that just says "add a linux partition the whole disk"17:57
adam_gsmoser: whats the bug no for that glace/nova AKI issue?17:57
smoserand i verify that this does result in val17:57
smoseradam_g, none at the moment.17:57
smoserit is upgrade related17:57
smoserbut i am confused on its extent otherwise, as I don't know why my last uploads of lucid failed. but i'm re-trying.17:58
smoser(it could have been luser error)17:58
smosersmb, , so i see failure like:17:58
smoser for((i=0;i<100;i++)); do printf "1,,L,*\n" | sudo sfdisk /dev/vdb > out 2>&1 || { echo "FAILED: $i"; cat out; break; } ; echo -n .; udevadm settle; done17:58
smoseryou think that is basically a valid test case, right?17:58
smbsmoser, Assuming we do not start while something is going on yes17:59
smbAt least as long internally do_fdisk is called (which is doing the reread check)17:59
smoserright.17:59
smoseri was looking at code too, its clear it calls it twice17:59
smbSo yes, that should be a valid case. Now, there could be side-requirement in a way of udev getting reasonably delayed in processing18:01
smbAlso the blkid and cdrom_id lock the mutex only while the open and close system calls are processed18:02
smbsmoser, seems to work in failing in testing here18:08
smoserwork in failing18:09
smoser:)18:09
smosernice18:09
smoserbug 942788, smb18:09
uvirtbot`Launchpad bug 942788 in util-linux "sfdisk without --no-reread is likely to cause race conditions" [Undecided,New] https://launchpad.net/bugs/94278818:09
smbsmoser, Oh well. :) It can sometimes take a while though. Just had a quite long run18:09
smoseryeah, but its quite obviously wrong.18:10
smoserthank you for your help, smb.18:10
smoseri'm really quite amazed that we never saw this issue until now.18:10
smoserthat code has been in basically as it is since 10.10.18:10
smbsmoser, Yes, at least since there is som much dynamic and user-space involved in rereading the partitions. Maybe we just did not notice or shrugged it away as temporary oddity. Or there is slightly more things triggered by udev or the commands run a bit slower or...18:12
smosersmb, when we upgraded to precise on canonistack, we upgraded kvm18:13
smoserand that changed the timings of disk access in vms18:13
smoserthat is what made us start seeing it now.18:13
smbTrouble with races is that they are racy... :)18:14
=== sixstringsg is now known as sixstringsg|away
hallyncan i have faster races in my kernel?18:18
smbhallyn, Upgrade your CPU :)18:23
eagles0513875hey guys where can one find the url's to the server so i can do a network installation18:25
SpamapSeagles0513875: http://cdimage.ubuntu.com/netboot/18:34
arosalesSpamapS: could you give me the pointer to the release notes bug once more :-/18:34
eagles0513875thanks SpamapS18:34
SpamapSarosales: https://launchpad.net/ubuntu-release-notes/precise18:34
arosalesSpamapS: thanks18:34
SpamapSone is glad to be of service18:34
smosersmb, hallyn has looked for an upgrade, but it seems faster transmeta chips are hard to find.18:35
hallynzing18:35
smb:)18:35
hallynkeep this up i'll have to take that thing to uds18:36
DavieySpamapS / arosales: Traditionally, we don't start populating ubuntu-release-notes project until further into the cycle.18:36
=== bladernr_ is now known as bladernr_afk
tyskai'm trying to automount home directories of users using this tutorial - https://help.ubuntu.com/community/AutofsLDAP, but i'm having some problems if someone could help me i will appreciate that18:37
arosalesDaviey: Following up on skaet looking to capture items for release notes as they land (ie per the release notes section in the template @ https://wiki.ubuntu.com/ReleaseTeam/Meeting/Agenda/TeamTemplate)18:40
Davieyarosales: ok18:42
arosalesDaviey: noob question. What is the process to _collect_ release note information? Do we queue requests in the LP branch as we know of items, or is there more of a concerted effort towards release?18:43
SpamapSDaviey: I had thought that beta1 was where we started filing those bugs.18:43
DavieySpamapS: Are there items you are sure will not be fixed for release at this stage?18:49
benjihallyn: I have an LXC issue that is biting me hard.  Will you take a look at http://paste.ubuntu.com/860883/ for me?18:53
arosalesDaviey: SpamapS: I guess we should not of significant package updates / new packages that would be good candidates for the release notes at this time, or is that not correct material for release notes?18:53
hallynbenji: that's probably bug 942144 ?18:55
uvirtbot`Launchpad bug 942144 in lxc "Using bindhome option of ubuntu template conflicts with ubuntu user" [High,Fix committed] https://launchpad.net/bugs/94214418:55
* benji looks with anticipation.18:55
hallynbenji: is buildbot uid 1000?18:55
hallynhm, no18:55
Davieyarosales: Well, traditionally it's been in the ramp up to release, we can document stuff which wasn't fixed, that users should know about.18:55
benjihallyn: it should be18:55
hallynprolly wouldn't cause a problem there yet18:55
benjihallyn: hmm, nope 10818:55
hallynoh, ok.  (i never tested it myself, not 100% sure how it manifeswts)18:55
DavieyIf there are issues which meet this criteria already, we should be concerned IMO.18:56
hallynbenji: that sucks, i dont' see any cause for the failure18:56
hallynbenji: you have enough disk space?18:56
benjihallyn: we do have an ubuntu user on the host which has uid 100018:56
hallynyeah but you're not trying to bind that in so that's fine18:57
benjiFilesystem      Size  Used Avail Use% Mounted on18:57
benji/dev/xvda1      8.0G  2.7G  4.9G  36% /18:57
hallynthis is on uptodate precise?  wahts' full cmdline?18:57
arosalesDaviey: agreed, we still have time to fix bugs.  Your thoughts on collecting new package and updated package notes at this time, or better towards release?18:58
benjihallyn: it's an EC2 AMI from the 25th or 26th, the full command line is on the first line of the paste18:58
Davieyarosales: TechnicalOverview is a rolling document for this, no?18:58
benjihallyn: I could try updating the host to see if it would help18:58
Davieyarosales: https://wiki.ubuntu.com/PrecisePangolin/TechnicalOverview18:59
benjiThe following packages will be upgraded:18:59
benji  libxml2 manpages ubuntu-minimal ubuntu-standard whoopsie18:59
arosalesDaviey: ah ok, thanks.19:00
* arosales sees how it fits together now :-)19:00
benjihallyn: same result after the upgrade: no discernable error in the output and then "failed to execute template 'ubuntu'"19:01
roaksoaxno g/me lunch19:01
* roaksoax lunch19:01
arosalesI think skaet's template update in the release notes also funnels into there (there = TechnicalOverview)19:02
adam_gDaviey: do i subscribe MIR bugs as normal to get them to list on the MIR section of http://status.qa.ubuntu.com/reports/ubuntu-server/release-bugs.html ?19:02
hallynbenji: and what is /etc/lxc/local.conf?19:03
Davieyadam_g: no, that is based on tags :/19:03
hallyn(trying with stock one)19:03
benjihallyn: http://paste.ubuntu.com/860910/19:03
arosalesDaviey: I just need to get updates from folks to update the TechnicalOverview now ;-)19:03
benji(yes, for some reason the first line is blank)19:03
Davieyarosales: lol, the hard part.19:03
arosalesDaviey: thanks for the info, and piecing that together for me.19:04
adam_gDaviey: ah, okay. well, we should probably add bug #941913 and bug #941916  however thats done19:05
uvirtbot`Launchpad bug 941913 in python-babel "[MIR] python-babel" [Undecided,Incomplete] https://launchpad.net/bugs/94191319:05
uvirtbot`Launchpad bug 941916 in python-tz "[MIR] python-tz" [Undecided,Incomplete] https://launchpad.net/bugs/94191619:05
hallynbenji: try adding '-F' at end of that cmdline19:06
benjihallyn: running...19:06
tyskaI'm trying to configure ldap + autofs to automount home directories of users in ubuntu (auth and mount home) - im trying to follow this tutorial https://help.ubuntu.com/community/AutofsLDAP - but i'm having problems - see this http://pastebin.com/raLeXikz19:06
tyskacould anyone help me?19:07
kirklandhallyn: hey, I'm having some odd behavior within a kvm guest, since I upgraded my host to precise on Friday19:13
kirklandhallyn: the load spirals out of control (up to ~40 or so) on a 4xCPU kvm guest19:14
kirklandhallyn: i've been running same guest as my dev environment for months on 11.10, no problem19:14
kirklandhallyn: problem showed up after i rebooted into precise 3.2 on the host19:14
kirklandhallyn: then guest is 11.1019:14
kirklandhallyn: ideas?19:14
hallynnone19:15
hallyndoes a fresh oneiric iso run fine?19:15
hallynbenji: it occurs to me that 'failed to execute ubuntu template' is a really really bad error msg19:16
ogra_arosales, can you please release the lock on the manifest wikipage ?19:16
arosalesogra_: just saved my updates19:16
benjihallyn: you might be onto something there ;)19:16
arosalesogra_:  its all yours19:16
ogra_thx19:16
jacobwkirkland: what's the current status of ecryptfs support for nss-pam-ldapd/getent?19:17
adam_gzul: https://code.launchpad.net/~gandelman-a/nova/libvirt_patch_refresh/+merge/95024  please note comments19:17
kirklandjacobw: hmm, none that I know of;  what's the bug number?19:17
hallynkirkland: can you open a bug with full cmdline and details on guest setup and host fs?  i'll try to reproduce.19:18
kirklandhallyn: sure;  you havent seen anything like it?19:18
benjihallyn: bad news, the -F didn't help; shall I pastebin the output?19:18
hallyni haven't19:18
hallynbenji: sure.19:19
hallynbenji: and then i'll probably have to ask you to edit lxc-create and lxc-ubuntu template to get more output aobut where it dies19:19
lifelesshallyn: you could make that an option19:20
lifeless--debug19:20
zuladam_g: yeah merged19:20
jacobwkirkland: there's bug #29343319:20
uvirtbot`Launchpad bug 293433 in ecryptfs-utils "ecryptfs-utils does not work with LDAP/Kerberos users" [Medium,Fix released] https://launchpad.net/bugs/29343319:20
benjihallyn: http://paste.ubuntu.com/860931/19:20
hallynlifeless: if it fails i'd like the output to e there regardless.  but yeah19:21
kirklandjacobw: that one is fix-released;  is that the same issue you're having?19:22
jacobwkirkland: i can see its fixed now :)19:23
kirklandjacobw: alrighty :-)  does that help you out?19:23
jacobwkirkland: yeah, all i'm trying to do is make ecryptfs volumes for users auth'd by pam_ldap.so19:24
kirklandjacobw: hmm, i think that's more of a pam question, than an ecryptfs question19:25
kirklandjacobw: pam_ecryptfs.so just has to be in your stack19:25
hallynbenji: oh, wait.  this might be due to something i inadvertently fixed in the bzr tree19:25
kirklandjacobw: and has to receive your login password to unwrap your wrapped-passphrase19:25
hallynbenji: on line 636 of /usr/lib/lxc/templates/lxc-ubuntu,19:25
hallynthere is19:25
hallynif [ ! -z $bindhome ]; then19:25
hallynbenji: can you change that to "if [ -n "$bindhome" ]; then19:26
hallyn(the quotes being hte important part)19:26
hallynof course, that shouldn't be a problem since you *did* specify it...19:26
benjiah19:26
jacobwkirkland: yeah, since that bug was fixed this is possible19:26
hallynbenji: uh. what is the shell for that user?19:27
hallynand, does it have a shadow entry19:27
benjihallyn: /bin/sh19:27
hallyni'm pretty sure do_bindhome is dying somehwere (bc it's set -e) where something funk ys going on with that user19:27
hallyndoes 'getent shadow (user)' return success?19:28
benjihallyn: buildbot:*:15398:0:99999:7:::19:28
hallynhm19:28
hallynyo'ure just shooting down all my ideas19:28
benjiI try.19:30
benjihallyn: I put set -x in the ubuntu template, we'll see if that illuminates the failure19:30
benjihallyn: aha: http://paste.ubuntu.com/860953/19:31
hallynbenji: awesome, thanks19:32
hallynguess 'set -x' would make a good debug mode19:32
benjihallyn: I have no idea what would cause that though.  Thoughts?19:32
benjiI assume it's some thing along the lines of "the user's group is screwed up"19:33
hallynbenji: i'm guessing the default grou pfor that user doesn't have an entry in /etc/group?19:33
benjihallyn: unfortunately that's not it:19:35
benjiroot@ip-10-72-61-240:/var/lib/juju/units/buildbot-slave-0/charm# grep buildbot /etc/passwd19:35
benjibuildbot:x:108:65534:BuildBot system user,,,:/var/lib/buildbot:/bin/sh19:35
benjiroot@ip-10-72-61-240:/var/lib/juju/units/buildbot-slave-0/charm# grep 65534 /etc/group19:35
benjinogroup:x:65534:19:35
hallynright -1 is 'nobody' for groups, so getent group probably is *supposed* to return error19:36
hallynnot sure19:36
hallynanyway, from lxc's point of view, i'll just ignore and proceed in that case, so the container creation shoudl go fine19:37
benjihallyn: well, we at least know it's the user that is to blame because I just reran it with "-b root" and it appears to have worked (no error at least)19:38
hallynbenji: two lxc bugs should be scrolling by here in a few minutes.  if you want to work around this for now, you can either edit the template to ignore that failure (i can give you a patch), or you can add a buildbot group and set buildbot user's group to that19:43
benjihallyn: I'll give hacking the script a shot19:43
benjihallyn: your help is much appreciated!19:44
benjihallyn: wait, I may be confused: are you saying that there's no need for the "Make sure the group exists in container" block at all?  I should just remove it?19:47
uvirtbot`New bug: #942847 in lxc (universe) "add a debug option to lxc-ubuntu template" [Medium,Confirmed] https://launchpad.net/bugs/94284719:51
uvirtbot`New bug: #942850 in lxc (universe) "lxc-ubuntu: don't fail if getent group returns error" [Medium,Confirmed] https://launchpad.net/bugs/94285019:51
hallynbenji: yeah in your case you should be fine with that code removed19:53
hallynI should think nogroup is always defined in ubuntu systems19:53
benjihallyn: thanks, I'll try that19:53
hallyn(but, again, since i used 'getent group' and it apparently doesn't like nogroup...)19:54
benjihallyn: it worked!19:55
benjihallyn: thanks again19:56
hallynbenji: np, thank you.  man the # of fixes going up after freeze is getting high :)19:56
benji:)19:56
smoseradam_g, i just opened bug 942865 with info about my image upload strangeness.20:08
uvirtbot`Launchpad bug 942865 in nova "upgrade from diablo leaves existing images with kernel unbootable" [Undecided,New] https://launchpad.net/bugs/94286520:08
smoseri think i'm going to try just uploading a new kernel/ramdisk and using the newly generated aki/ari20:08
sorenDaviey: Random crack of the day: https://code.launchpad.net/~soren/reincarnate/trunk/20:12
smoseradam_g, that is some awesome bug.20:13
smosersoren, you might enjoy reading that one also20:13
Davieysoren: nice.. i'll try that at somepoint20:16
uvirtbot`New bug: #942862 in lxc (universe) "lxc-create failure creating a lucid container in a precise host" [Undecided,New] https://launchpad.net/bugs/94286220:17
philpemHi all. I've just had a disc failure on a server, and am taking this opportunity to do a clean Ubuntu 11.10 Server install. Thing is, I need to decide between RAID-1 only, or RAID-1 plus LVM. Why would I want to use LVM and md RAID together? what are the benefits/disadvantages?20:17
=== sixstringsg|away is now known as sixstringsg
greppyphilpem: depending on how you configure it, you can add disks of different sizes into the raid.  there are advantages to both RAID on LVM and LVM on RAID, it just depends on what you want to accomplish.20:26
philpemgreppy, well at the moment, the machine is rigged with a single '/' partition on software RAID1, and a swap partition.20:29
philpem(also RAID1'd)20:29
philpemI'm just wondering if there's any benefit to going LVM, or if I should just stick with the setup I have.20:30
=== sixstringsg is now known as sixstringsg|away
greppyphilpem: I use LVM to be able to tweak partitions.20:49
greppyI build the partitions based on what I think I will need, leaving a large amount of space unallocated.20:49
greppythat way if needs change, I can add more space to /home, or to /var or to /usr without having to tear it all down and reinstall.20:49
RoyKanyone here ever setup KVM with HA? I have central storage on NFS and want to use three nodes for KVM, moving the VMs when needed20:50
hallynbenji: fwiw, it turns out my group handling was just wrong altogether21:04
benjihallyn: I'm not good in group settings either.21:04
hallynbenji: i was assuming 'getent group $user' would get the user's group entry, but it needs to be given the group's name or #21:05
RoyKhallyn: groups user21:05
RoyKhallyn: groups $user21:05
RoyKdunno if that was what you wanted, though21:06
benjihallyn: that's shure what it looks like it does (as long as the group really exists)21:07
benji% getent group benji21:07
benjibenji:x:1000:21:07
Davieyjdstrand: Have i missed you?21:11
jdstrandDaviey: I am here21:11
=== matti_ is now known as matti
Davieyjdstrand: so... lynxman looked at ruby1.9 for puppet.21:23
Davieyjdstrand: he was less than confident21:24
jdstrandthat is unfortunate21:25
Davieyjdstrand: you got mail.21:26
roaksoaxzul: so i'm working on the cobbler-web separation and stuff... and I'm wondering if you have any idea of why postinst would not return when it restarts apache, and why would postrm fail because of it?21:26
jcorneliHello: I am having trouble after upgrading my Linode from 11.10 to 12.0421:27
jcorneli"Your Linode appears to have stalled mid-boot at around the time that the start up process is handed off to init. It looks like the kernel handed off control to init, as it didn't panic, however it doesn't appear that any further services were started after that."21:27
jcorneli"You'll want to contact the Ubuntu community for more information on known issues with 12.04 and virtualized environments."21:27
jcorneliThis is what the Linode support person told me.21:27
jcorneliWhen I log in via LISH, I just see the log, I don't get a proper console.21:27
jcorneliI copied the log here: http://pastebin.com/raw.php?i=4cMh5Gcr21:28
zulroaksoax: no idea21:28
zulroaksoax: cobbler was a former life :)21:28
roaksoaxzul: lol, you've never experienced similar issues?21:28
zulroaksoax: no i havent21:29
SpamapSroaksoax: perhaps apache is blocking on something weird like a fifo/socket ?21:29
roaksoaxzul: do you think you could take a quick look at it? ppa:andreserl/ppa --> just install cobbler21:29
SpamapSroaksoax: since the apache init script doesn't daemonize until after logs are open, its at least feasible.21:29
zulroaksoax: i can later probably..21:30
roaksoaxSpamapS: ah maybe... the weird thing is that it doesn't "fail" on the cobbler-web package21:30
roaksoaxuhmm interesting it now returned after a while waiting21:31
SpamapSroaksoax: perhaps DNS issues21:31
SpamapSroaksoax: I think if you configure apache in a certain way it will do a DNS lookup before daemonization too to lookup the servername21:32
roaksoaxSpamapS: it could mauybe be that21:33
=== bladernr_afk is now known as bladernr_
AzelphurIs there any way to get rebootless upgrades now ksplice is gone?21:39
roaksoaxSpamapS: when I manually restart apache2 (while it has hanged the psotinst), then the postinst resumes normally, so do you think it's either something being openned or a resolve?21:54
SpamapSroaksoax: hard to say. *hrm*21:55
roaksoaxSpamapS: does this looks sane to you? http://paste.ubuntu.com/861133/21:56
smoserstgraber, around ?22:03
stgrabersmoser: yep22:06
smoserwas looking at /etc/init/container-detect.conf22:06
smoserand had a question that i thin i resolved22:06
smoseryou search for a line that tsarts with VxID22:06
smoserbut you could fail on22:06
smoserVxIDABCDEFG22:07
smoserand also, just as a fun poke22:07
smoserhttp://uselessuseofcat.com/22:07
smosercat /proc/self/status | grep ^VxID | cut -f2)22:07
smoseri think faster:22:07
smoserif awk -F: '$1 == "VxID" { if ($2 > 1) x=0; }; END { exit(x); }' x=1 /proc/self/status;22:07
smoser  container="vserver"22:07
smoserfi22:07
stgrabersmoser: awk is in /usr, so can't use it22:08
smoserwell that sucks. :)22:09
smoserso is cut, though22:09
smoserso, touche!22:09
stgraberoh, I didn't know that ... guess I'll have to fix a few scripts then22:09
smosersh -c 'IFS=": "; while read key val; do [ "$key" = "VxID" ] || continue; [ "$val" = "0" ] ; exit; done; exit 1' < /proc/self/status22:11
Daviey$ /bin/busybox cut22:11
Davieycut: expected a list of bytes, characters, or fields22:11
Davieyi suck.22:12
smoseroh yes, you do.22:12
stgrabersmoser: line=$(grep ^VxID /proc/self/status) ; echo ${line##VxID: }22:12
Davieyoh, no - it is a builtin of cut.22:12
Davieyerr, busybox22:12
smoserstgraber, still, you need a :22:13
stgrabersmoser: indeed22:13
smoserut yeah.22:13
smoserbut yeah.22:13
smoserthat would be fine.22:13
kklimondajcorneli: hmm, it seems that initram script finishes, have you tried booting with init=/sbin/init --verbose? Do you use stock ubuntu kernel, or the one provided by linode?22:20
jcorneli@kklimonda: I use ubuntu kernel (just upgraded from 11.10 via sudo do-release-upgrade -d )22:21
jcornelihow would I set the init=/sbin/init --verbose option you indicated?22:21
jcorneliI have access to the filesystem from Linode's "Rescue mode"22:22
jcornelijust let me know what file to change and I can give it a whirl22:22
kklimondajcorneli: when I boot my linode from lish I get a full access to grub, I can configure it there22:24
kklimondabut you can also edit /etc/default/grub22:25
jcorneli@kklimonda: OK I will try that and post results22:26
jcorneli@kklimonda: since I can't access the console, I can't run sudo update-grub22:34
jcorneliI hope that won't be a problem...22:34
stgrabersmoser: http://paste.ubuntu.com/861186/ looks good?22:34
kklimondajcorneli: if you can access grub from lish you can change it there22:36
kklimondajcorneli: without update-grub changes you made to /etc/default/grub won't have any effect - you could try editing /boot/grub/grub.cfg directly22:36
jcorneliOK22:37
jcornelithere is no such file on my system tho22:37
kklimondajcorneli: hmm, right22:39
kklimondait's /boot/grub/menu.lst22:39
kklimonda(so I guess my comment about /etc/default/grub also didn't make much sense - I've completely forgotten you end up with grub1 when enabling stock ubuntu kernels on linode vps)22:40
jcorneliwell, I don't have a /boot/grub/ directory22:40
SmoziusHey guys, I'm setting up LVMs, where in the text install menu do I set the PE size for LVMs?22:40
SmoziusBy default its only 4MB I would like it to be 16MB22:40
jcorneliso maybe we need to look in a different place22:40
jcorneliI can also ask the Linode support guys22:41
SmoziusIt's not the reserve block thats 5% is it?22:41
kklimondajcorneli: huh, are you sure you are not running kernel provided by linode? Do you use pv-grub?22:43
jcorneliNot sure - I would have assumed that after several updates, I would now have an Ubuntu kernel22:44
kklimondajcorneli: by default all linodes are deployed with a custom linode kernel that isn't even installed on the system22:44
kklimondajcorneli: when you edit your linode what kernel is set?22:45
Zermannohi, i'm trying to install ubuntu server 10.04.4 on a two-disk raid 1. The install fails to set up grub, any advice?22:46
SmoziusWhat do you mean fails to install grubs?22:46
SmoziusDoes it attempt to boot?22:46
Zermannono, during installation procedure it fails to install grub22:47
Zermannoafter installing all packages22:47
Smoziusdoes it ask you where to install grubs?22:48
Zermannoyes, it asks me if i want to install to the mbr, i say yes22:48
simonjjI've got a problem with vm-builder, is anyone in here that knows this puppy fairly well22:48
simonjjhope this isn't OT22:48
SmoziusZermanno: After you say yes and it tries to boot, what does it do? Does it sit there with a flashing cursor?22:50
kklimondajcastro: if you don't have pv-grub set there, I'd at least try to change kernel to 3.2.1-linode40, which seems to be the most recent one.22:50
kklimondaerm22:50
kklimondajcorneli_: ^22:50
jcorneli_hi22:50
jcorneli_sorry I was disconnected22:50
Zermannoit does not try to boot, after i say yes the blue background become red and said Unable to install grub or something like that, i can't see the screen now22:50
uvirtbot`New bug: #942934 in lxc (universe) "update apparmor profile to restrict mounts" [High,Confirmed] https://launchpad.net/bugs/94293422:50
SmoziusHow are your drives hooked up?22:51
SmoziusAre you using 10.04?22:51
kklimondajcorneli_: check in your VM settings what kernel is set - if it's not pv-grub (I assume it's not) then you are running a kernel provided by linode, I'd try changing it to 3.2.0-linode40 to see if it helps - maybe the one your are booting currently is too old?22:52
kklimondait's a complete guess though22:52
Zermannotwo 1.5 TB hd, same partitioning scheme each. 1mb free starting, 1 34gb partition with /, 1 34gb swap, 1.4tb for /home22:52
Zermannoboth first parts are in md022:52
Zermannoboth seconds in md122:52
Zermannothirds in md222:52
SmoziusCreate an MD for /boot22:53
Zermanno10.04.422:53
Smozius200-500MB22:53
jcorneli_looks like 3.2.1-linode40 (to be precise!)22:53
smoserstgraber, i'd tihnk so.22:53
jcorneli_@kklimonda: changing to  that22:53
SmoziusI need to  head to class Zermanno, good luck.22:53
ZermannoSmozius, tnx bye22:54
smoserstgraber, im in an lxc container23:10
smoseri run 'running-in-container'23:10
smoserit exits 123:10
smoserhallyn, ^23:13
smoseram i doing something wrong ?23:13
smoseri just created container with 'lxc-create -c ubuntu'23:13
smoserok.23:15
smosernever mind.23:15
smoseruser-error, except for one thing23:15
smoserudo running-in-container23:15
smoserlxc23:15
smoser$ sudo running-in-container23:15
smoserlxc23:15
smoser$ running-in-container; echo $?23:15
smoser123:15
smoserso if i'm not root, it says i'm not in a container23:16
smoserbecause i can't run 'status container-detect'23:16
smoserbug 94296123:19
uvirtbot`Launchpad bug 942961 in upstart "running-in-container exits 1 as non-root even inside a container" [Undecided,New] https://launchpad.net/bugs/94296123:19
hallynsmoser: stgraber: ^ note, it works fine if you have dbus installed23:22
hallynbesides, it's a security feature :)23:23
hallynmaybe we should jsut go based on /run/container_type, no matter what?  run is never persistent...23:24
hallynthat, or install dbus in all containers23:25
smoseri dont know why not.23:25
hallynwhy not which?23:25
smoserwhy not /run/container_type23:25
hallyni'll wait for stgraber to chime in though23:26
=== sixstringsg|away is now known as sixstringsg
SpamapSyou're installing upstart in all containers, but not dbus?23:33
hallynbc i have to ask for it by hand23:33
hallynupstart gets installed automatically, why not dbus?23:34
jMCghallyn: upstart is init, pretty hard to go without init on Unix.23:36
hallynjMCg: agreed.  My point was if SpamapS' comment made sense then mine would as well :)23:37
hallyndbus is the preferred way to talk to upstart23:37
hallyn(according to upstart)23:37
hallynand yeah, upstart says if you want non-root to talk to it, you need dbus...23:38
hallynso i just dont' know if anything needs to change23:38
hallyn(re bug 942961)23:38
uvirtbot`Launchpad bug 942961 in upstart "running-in-container exits 1 as non-root even inside a container" [Undecided,New] https://launchpad.net/bugs/94296123:38
hallynsmoser: if you you want higher prio on that bug, pls feel free to raise it.  low was my view of it, but if it blocks something then that changes things23:44
=== sixstringsg is now known as sixstringsg|away
hallynoh, but it should print out an error msg in any case23:49
smoserhallyn, and exit somthing other than 123:52
smoserits not blocking anything for me23:52

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!