ruben23 hi, how do i create lvm on my vg..in command line00:01
cemcruben23: you mean LV in an existing VG?00:05
ruben23 cemc: yes i have exixting VG00:06
cemcruben23: man lvcreate00:06
ruben23 cemc: i would like to increase my /var/  cemc:00:07
=== erichammond1 is now known as erichammond
cemcugh, how I hate the default VG and LV names :) should be something more like dev/mapper/<hostname>/<partition_name>, IMHO00:10
cemcruben23: resizing is not creating, you have to look that up on the net. I've done that a couple of times, but I'm not 100% sure how, was some time ago00:11
cemcruben23: but if you have the space and time, I would create another bigger LV for /var and copy it, then delete the old one. not a fan of resizing ext3 (or any fs for that matter), dunno why. should be safe tho00:12
mathiazkirkland: hey - did you run the new eucalyptus.conf files system via upstream?00:18
kirklandmathiaz: i did00:19
kirklandmathiaz: nurmi wants me to get rid of /usr/share/*00:19
mathiazkirkland: ok - so the issue I see is that the CC is not able to load its configuration file00:19
kirklandmathiaz: and just put that stuff into /etc/eucalyptus/eucalyptus.conf00:19
mathiazkirkland: right - the cluster.c file only knows about *two* configuration file00:19
kirklandmathiaz: yeah, i'm sorry i didn't update you00:19
kirklandmathiaz: nurmi implemented half of it00:19
kirklandmathiaz: while i implemented the other half00:20
kirklandmathiaz: but we miscommunicated on the file names00:20
mathiazkirkland: so the whole include stuff doesn't work since the cluster config parser doesn't understands includes00:20
kirklandmathiaz: and the design changed, after i talked to him, was my multi-hour conversation with you and slangasek00:20
kirklandmathiaz: right, it just looks for KEY=VALUE00:20
* mathiaz would just suggest to use yam;00:20
* mathiaz would just suggest to use *yaml*00:21
mathiazkirkland: so what's the current plan00:21
mathiazkirkland: ?00:21
kirklandmathiaz: well, i told nurmi that i need to collect my notes to recollect why we needed 3, and not just 200:22
kirklandmathiaz: i remember i lost the argument in favor of just 2 files00:22
kirklandmathiaz: but i don't recall which reason was the reason i finally gave in to00:22
mathiazkirkland: right - in order to be able to stick new default options without prompting for a conffile00:22
mathiazkirkland: if new options are available and need to be defined, we'd rather not update eucalyptus.conf which is a conffile00:23
mathiazkirkland: that being said we could update eucalyptus-local.conf programatically00:23
mathiazkirkland: ie from the upgrade script00:24
kirklandmathiaz: it certainly sounds like upstream would prefer it this way00:24
mathiazkirkland: since eucalyptus-local.conf should not be modified by the end user (even though we should take care of keeping local changes - just to follow the debian policy)00:24
kirklandeucalyptus.conf and eucalyptus-local.conf00:24
mathiazkirkland: ok - so to accomodate both upstream and the debian policy I'd suggest to move eucalyptus-local.conf to /var/lib/eucalyptus/00:25
mathiazkirkland: and have all the default options set in eucalyptus-local.conf00:25
mathiazkirkland: maintainer scripts and euca_conf write to eucalyptus-conf00:26
mathiazkirkland: maintainer scripts and euca_conf write to eucalyptus-local.conf00:26
mathiazkirkland: if new options are needed, we can mangled eucalyptus-local.conf00:26
kirklandmathiaz: okay00:26
mathiazkirkland: probably using a token to delimit the section updated by euca_conf00:27
kirklandmathiaz: how much effort do you suggest i put in supporting upgrades from today's version to the next version?00:27
mathiazkirkland: none00:27
kirklandmathiaz: i'm inclined to call today's bust00:27
kirklandmathiaz: i can handle this, i think00:27
kirklandmathiaz: so what was going to /usr/share/eucalyptus/eucalyptus-defaults.conf now get's put into /etc/eucalyptus/eucalyptus.conf (as a conffile)00:28
mathiazkirkland: the package has been in the archive for just one day00:28
kirklandmathiaz: right00:28
mathiazkirkland: and it's a developement version - we  do our best effort00:28
mathiazkirkland: hm nope - /usr/share/eucalyptus/eucalyptus-defaults.conf goes into /var/lib/eucalyptus/eucalyptus-local.conf00:28
mathiazkirkland: /etc/eucalyptus/eucalyptus.conf stays at it is00:29
mathiazkirkland: /etc/eucalyptus/eucalyptus.conf stays *as* it is00:29
mathiazkirkland: maintainer scripts and euca_conf should *not* touch /etc/eucalyptus/eucalyptus.conf00:29
mathiazkirkland: /etc/eucalyptus/eucalyptus.conf is the override, while /var/lib/eucalyptus/eucalyptus-local.conf is the default configuration file00:29
kirklandmathiaz: WTF goes in eucalytpus.conf then?00:29
mathiazkirkland: what's in it now00:30
kirklandmathiaz: as of yesterday, everything (monolithic config file)00:30
kirklandmathiaz: as of today, 2 lines, sourcing -defaults and then -local00:30
kirklandmathiaz: as of tomorrow .... ????00:30
mathiazkirkland: tomorrow: 1 line sourcing /var/lib/eucalyptus/eucalyptus-local.conf00:31
kirklandmathiaz: installed by dpkg as a conf file?00:31
mathiazkirkland: and /var/lib/eucalyptus/eucalyptus-local.conf is the merge of eucalyptus-local.conf and eucalyptus-default.conf00:31
mathiazkirkland: yes00:31
mathiazkirkland: today's /etc/eucalyptus/eucalyptus.conf is all good IMO00:31
kirklandmathiaz: and when we add a new option00:32
mathiazkirkland: (modulo the source change)00:32
kirklandmathiaz: we add it to the manpage00:32
kirklandmathiaz: and mangle -local in the maintainer scripts?00:32
mathiazkirkland: yes! :)00:32
kirklandmathiaz: okay00:32
kirklandmathiaz: i'll make it happen00:32
kirklandmathiaz: you're not particular about eucalyptus-local.conf vs. eucalyptus.local.conf ?00:33
kirklandmathiaz: nurmi's code does eucalyptus.local.conf00:34
mathiazkirkland: not all all00:34
mathiazkirkland: right - noticed that00:34
mathiazkirkland: I'm testing the new setup discussed above00:34
kirklandmathiaz: his code also sources /etc/eucalyptus/eucalypts.local.conf, which i'll put a symlink there to /var00:34
mathiazkirkland: (ie eucalyptus.conf and a merged eucalyptus.local.conf)00:34
kirklandmathiaz: btw, does your -cc preseed file work?00:34
kirklandmathiaz: i'm using your eucalyptus values there and i'm not getting a cloud installation00:35
mathiazkirkland: :) - haven't tested it for a while00:35
mathiazkirkland: right - that's very well possible00:35
mathiazkirkland: the debconf questions have probably changed00:35
kirklandmathiaz: okay00:35
kirklandmathiaz: does order matter at all in preseed files?00:35
mathiazkirkland: I don't think so00:36
kirklandmathiaz: hrm, okay00:36
mathiazkirkland: the preseed file is read at once and all the values are dumped in the debconf database00:36
kirklandmathiaz: right, that's what i figured00:36
kirklandmathiaz: my node preseed is working well, but not the clc/cc/sc/walrus00:36
mathiazkirkland: the relevant scripts access the debconf database, not the preseed file00:36
* kirkland kinda wishes eucalyptus worked more that way :-)00:37
mathiazkirkland: hm - thinking about the filenames00:40
mathiazkirkland: .local.conf seems counter-intuitive00:40
mathiazkirkland: as .local.conf is the file edited by euca_conf and the maintainer scripts00:40
mathiazkirkland: how about switching the two?00:41
mathiazkirkland: eucalyptus.conf not being a conffile anymore00:41
mathiazkirkland: and eucalytpus.local.conf is a conffile00:41
mathiazkirkland: to be edited by the local sysadmin00:42
mathiazkirkland: eucalyptus.conf would then source .local.conf at the *end*00:42
mathiazkirkland: and point the sysadmin to edit .local.conf instead00:42
kirklandmathiaz: why have .local.conf at all?00:43
kirklandmathiaz: why not just make eucalyptus.conf not a conf file00:43
kirklandmathiaz: when we need to add something to it, we just jam it in00:43
mathiazkirkland: well - if we do so we need to make sure that local modifications are kept00:44
mathiazkirkland: when new options are added00:44
* kirkland has grown exhausted of this "challenge"00:45
mathiazkirkland: it's possible - the logic in the maintainer scripts can get ugly though00:45
mathiazkirkland: I totally understand you00:45
mathiazkirkland: it's complicated00:46
kirklandmathiaz: at the end of the day, eucalyptus does not "source" these files00:46
kirklandmathiaz: it does a crappy job parsing them instead00:46
kirklandmathiaz: for that reason, cascading any set of sourcing/including just doesn't fit eucalyptus' model00:46
* mathiaz nods00:46
mathiazkirkland: aren't the upstart job sourcing it?00:46
Sam-I-Ammathiaz: did you see my note earlier about being out of whack for a few months?00:47
kirklandmathiaz: ubuntu's code does, eucalyptus' code does not00:47
mathiazSam-I-Am: yes - that's life00:47
Sam-I-Amk, just wanted to make sure00:47
mathiazSam-I-Am: thanks for letting me know though00:47
Sam-I-Amhopefully its going to be fixed in a bit00:47
Sam-I-Ami'm sorry i kinda fell off00:47
mathiazSam-I-Am: that's ok - as long as you attend to the more important things ;)00:48
mathiazSam-I-Am: I hope everything will turn out well00:48
Sam-I-Amthings are starting to look good00:48
mathiazkirkland: ah ok.00:49
Sam-I-Amshould have a job back in the sys/netadmin world soon00:49
mathiazkirkland: I thought that having eucalyptus.conf was a hard requirement00:49
mathiazkirkland: so what we could do is to introduce /var/lib/eucalytpus/eucalyptus-defaults.conf00:49
mathiazkirkland: keep /etc/eucalyptus/eucalyptus.conf as it is now00:50
mathiazkirkland: and have euca_conf and maintainer scripts modify /var/lib/eucalytpus/eucalyptus-defaults.conf00:50
mathiazkirkland: and update the ubuntu upstart jobs to source both /var/lib/eucalytpus/eucalyptus-defaults.conf and /etc/eucalyptus/eucalyptus.conf00:50
mathiazkirkland: that also means updating the upstream source code to use /var/lib/eucalytpus/eucalyptus-defaults.conf as the default configuration file and have /etc/eucalyptus/eucalyptus.conf be the .local.conf file00:52
kirklandmathiaz: hrm00:52
mathiazkirkland: ie: EUCALYPTUS_CONF_LOCATION=/var/lib/eucalytpus/eucalyptus-defaults.conf00:52
mathiazkirkland: EUCALYPTUS_CONF_OVERRIDE_LOCATION=/etc/eucalyptus/eucalyptus.conf00:53
kirklandmathiaz: i'm starting to feel more like throwing in the towel ... back to a monolithic /etc/eucalyptus/eucalyptus.conf00:53
kirklandmathiaz: just not a conffile00:53
mathiazkirkland: that way eucalyptus.conf stays as a conffile00:53
kirklandmathiaz: users, euca_conf, dpkg edits that file00:53
kirklandmathiaz: upstream code stays the same00:53
kirklandmathiaz: we just seed it if it's not there00:54
kirklandmathiaz: and handle new options ad hoc00:54
kirklandmathiaz: we have a manpage now00:54
mathiazkirkland: the last part may be trickier though00:54
kirklandmathiaz: then we don't handle them at all in dpkg ...  we handle them in euca_conf00:54
=== erichammond1 is now known as erichammond
kirklandmathiaz: document them in the manpage00:55
kirklandmathiaz: and force eucalyptus to choose a sane (documented) default if unspecified00:55
mathiazkirkland: right - does the code currently have sane default if unspecified?00:56
kirklandmathiaz: don't know00:56
mathiazkirkland: AFAICT no00:56
mathiazkirkland: as the CC doesn't start with the current empty eucalyptus.conf file00:56
mathiazkirkland: as the CC doesn't start with the current *almost* empty eucalyptus.conf file00:57
kirklandmathiaz: okay -- in that case, you're instructed to run "euca_conf"00:58
kirklandmathiaz: until you get a eucalyptus.conf that is not empty00:58
kirklandmathiaz: with sufficient configuration to start your cluster00:58
mathiazkirkland: well - that's how things were in karmic, with euca_conf being run in the postinst script00:59
mathiazkirkland: the problem is how to handle upgrades00:59
mathiazkirkland: where you need to add new (default) options00:59
mathiazkirkland: either we can rely on upstream to do that00:59
mathiazkirkland: or we can ship a default set of options that can be easily upgraded01:00
kirklandmathiaz: nurmi is here01:02
* nurmi waves01:02
mathiaznurmi: hi! :)01:02
nurmimathiaz: hey mathiaz!01:02
nurmimathiaz: what's up?01:02
mathiaznurmi: is it snowing down there?01:02
mathiaznurmi: kirkland and I have been discussing how to handle configuration files in eucalyptus01:03
nurmimathiaz: heh, hardly01:03
nurmimathiaz: nod01:03
mathiaznurmi: IIUC there was one monolitic configuration file (eucalyptus.conf)01:03
mathiaznurmi: and you've added an override (eucalyptus.local.conf)01:04
nurmimathiaz: that is correct01:04
nurmimathiaz: i also think it is important to clearly state how the files are parsed01:04
mathiaznurmi: IIUC it's key=value01:04
nurmimathiaz: there are some shell scripts that source the files (init scripts, euca_conf, etc), and, there are some C components that parse the files directly (key=val)01:05
mathiaznurmi: and isn't actually a shell file01:05
nurmimathiaz: right, that is the key point01:05
mathiaznurmi: right - so it's fair to assume that the configuration file are not shell fragment01:05
nurmimathiaz: yep, in fact, that will potentially break the C parser01:05
mathiaznurmi: but just key=val pair to keep the lowest commmon denominator01:05
nurmimathiaz: correct01:06
kirklandnurmi: is it "incorrect" that our upstart scripts (etc) actually source them as if they're shell?01:06
nurmikirkland: no, the files can be sourced by bash/sh, in that they're just KEY=VAL and comment lines start with '#'01:06
mathiaznurmi: do the different components of eucalyptus work correclty if there isn't any configuration file?01:07
mathiaznurmi: or rather if configuration files don't have all the options defined01:07
mathiaznurmi: for example, a new version introduces a new option - what should be done to upgrade existing configuration files?01:07
nurmimathiaz: oh, sure.  There are some options that are used by more than one component, but some are only used by an individual component01:07
nurmimathiaz: we have been using 'euca_conf --upgrade' to handle config file upgrades01:08
nurmimathiaz: sorry '--upgrade-conf'01:08
mathiaznurmi: ok01:08
mathiaznurmi: --upgrade-conf will add new keys with reasonable default values01:09
mathiaznurmi: and (may be) remove deprecated keys?01:09
nurmimathiaz: our common mechanism is to move the old file aside, put in a new default, then run '--upgrade-conf'01:10
nurmimathiaz: right01:10
mathiaznurmi: how do you handle local modified options?01:10
mathiaznurmi: for example, the list of nodes?01:10
nurmimathiaz: they persist when one uses 'upgrade-conf'01:10
mathiaznurmi: is this the case for any options, or only of the nodes option?01:10
nurmimathiaz: so any value that you had before, with a key that is still valid, persists01:11
mathiaznurmi: ok01:12
mathiaznurmi: where is the new default file that upgrade-conf uses?01:12
mathiaznurmi: in /usr/share/eucalyptus/... ?01:12
nurmimathiaz: no it would be /etc/eucalyptus/eucalyptus.conf01:12
mathiaznurmi: I meant in the package01:13
nurmimathiaz: first, we move the existing eucalyptus.conf aside (eucalyptus.conf.orig), then install the new default (eucalyptus.conf), then run '--upgrade-conf'01:13
nurmimathiaz: that is how our packages do conf upgrades01:13
mathiaznurmi: and you move the existing eucalyptus.conf file in the preinstall script?01:14
nurmimathiaz: nod01:14
mathiaznurmi: and /etc/eucalyptus/eucalyptus.conf is a conffile then?01:14
mathiaznurmi: ie it's shipped in the package as /etc/eucalyptus/eucalyptus.conf?01:15
nurmimathiaz: yes01:15
mathiazkirkland: what happens if a conffile has been moved?01:16
mathiazkirkland: is it reintalled? or is there an prompt made by dpkg?01:16
kirklandmathiaz: in which case?01:16
kirklandmathiaz: yesterday, or today's upload?01:16
nurmimathiaz: the new situation (with eucalyptus.local.conf), is an attempt to allow the system to keep eucalyptus.conf unmodified, so that pacakge upgrades can happen smoother01:16
mathiaznurmi: I'm trying to figure out whether dpkg would prompt the end user would be prompted01:16
nurmi.oO(more smoothly)01:16
mathiazkirkland: yesterday01:17
kirklandmathiaz: it would have been reinstalled01:17
mathiaznurmi: right - one of the goal here is to avoid useless prompts01:17
nurmimathiaz: nod01:17
=== erichammond1 is now known as erichammond
mathiaznurmi: ok - so on an upgrade by moving the existing eucalyptus.conf in preinstall, dpkg would install the *new* default eucalyptus.conf without prompting the user01:18
nurmimathiaz: so, if eucalyptus.conf was a 'sane default' file, and any run-time modifications were only made to eucalyptus.local.conf, then package upgrade should not ask any questions, if I understand correctly01:18
mathiaznurmi: kirkland: is that correct?01:18
mathiaznurmi: yes - that's one possibility01:19
kirklandmathiaz: i'm digesting that still ...01:20
nurmimathiaz: following this idea just for a moment; on upgrade, we could make '--upgrade-conf' do its thing with 'eucalyptus.local.conf' instead of 'eucalyptus.conf'01:20
nurmimathiaz: avoiding the need to mess with eucalyptus.conf at all01:20
nurmimathiaz: and, since eucalyptus.local.conf is not part of the package, the whole upgrade process could happen in post-install01:21
nurmimathiaz: okay, i'm done following that idea, thanks :)01:21
* mathiaz thinks01:21
* kirkland thinks eucalyptus does its own thing with its configuration files; it doesn't want dpkg or distro management thereof01:22
mathiaznurmi: right01:22
* kirkland thinks we make eucalyptus.conf a non-conffile01:22
kirklandand it's up to euca_conf to keep it sane01:23
kirkland^ at least, that's one option01:23
uvirtbot`kirkland: Error: "at" is not a valid command.01:23
kirklandthanks uvirtbot`01:23
mathiaznurmi: so having eucalyptus.conf be a conffile and eucalyptus.local.conf a configuration file edited by euca_conf and the local sysadmin01:23
mathiaznurmi: you wouldn't need to move eucalyptus.conf in the presinstall01:24
nurmimathiaz: nod01:24
mathiaznurmi: since eucalyptus.conf shouldn't be changed01:24
nurmimathiaz: we wouldn't need to mess with eucalyptus.conf at all01:24
mathiaznurmi: and if it is, then dpkg should prompt for a diff01:24
nurmimathiaz: well it *could* be, but that is up to the admin01:25
nurmimathiaz: nod01:25
mathiaznurmi: right - it could be - but then it's up to the admin and we don't wanna loose anything01:25
nurmimathiaz: nod01:25
mathiaznurmi: hence relying on dpkg conffile mechanism01:25
mathiaznurmi: and then euca_conf would modify .local.conf01:25
mathiaznurmi: and upgrade .local.conf as well in the postinst01:26
nurmimathiaz: nod01:26
mathiazkirkland: ^^ sounds like a good plan to me01:26
mathiazkirkland: eucalyptus.conf stays as a conffile01:27
* kirkland reads01:27
mathiazkirkland: euca_conf modifies eucalytpus.local.conf01:27
mathiazkirkland: the sysadmin is advised to edit eucalytpus.local.conf01:27
kirklandmathiaz: and what is eucalyptus.conf seeded with on new install?01:27
mathiazkirkland: eucalyptus.local.conf is actually a configuration file01:28
mathiazkirkland: eucalytpus.conf is a conffile, ie is part of the debian package01:28
mathiazkirkland: it's content is the default eucalytpus.conf with all the options set correclty01:28
mathiazkirkland: eucalyptus.local.conf is installed by the maintainer script (thus is a configuration file, but not a conffile)01:29
nurmimaybe s '01:29
nurmimistype, nvm01:29
kirklandmathiaz: basically what i put in /usr/share ?01:29
mathiazkirkland: nothing01:29
mathiazkirkland: the current eucalyptus-default.conf is /etc/eucalytpus/eucalyptus.conf01:30
mathiazkirkland: and the current eucalyptus.conf is the base for eucalyptus.local.conf01:30
kirklandmathiaz: right01:30
mathiazkirkland: euca_conf modifies eucalyptus.local.conf (during pkg upgrade as well)01:30
kirklandmathiaz: gotcha01:31
mathiazkirkland: and since eucalyptus.local.conf is *not* a conffile, there shoudn't be any prompt on upgrades01:31
kirklandnurmi: you're good with this?01:31
nurmikirkland: yep!  that sounds great to me01:31
mathiazkirkland: now - there is a special case to upgrade from karmic01:31
mathiazkirkland: since eucalyptus.conf will have changed between karmic and lucid01:31
mathiazkirkland: (well technically it is different on every system since the list of NODES is local)01:32
nurmikirkland: i'll make sure '--upgrade-conf' is working properly, over the next week, if this is the decided upon plan01:32
mathiazkirkland: so to avoid a prompt on upgrade from karmic, the presinstall script should move the existing eucalytpus.conf to eucalyptus.local.conf01:33
mathiazkirkland: so that dpkg installs the new eucalyptus.conf file01:33
kirklandmathiaz: right, i already hacked it that way in today's upload01:33
kirklandmathiaz: slangasek advised me as such01:33
mathiazkirkland: and then --upgrade-conf should be run as part of the normal upgrade process in the maintainer script01:33
kirklandmathiaz: basically, if upgrading, and eucalyptus.conf exists, and .local doesn't, it's moved to .local01:34
* mathiaz nods01:34
kirklandnurmi: what's the full syntax for --upgrade-conf?01:34
kirklandmathiaz: always --upgrade-conf ?01:34
mathiazkirkland: nurmi: and so we rely on euca_conf --upgrade-conf to mangle properly the local configuration file01:34
kirklandmathiaz: on any upgrade?01:34
nurmikirkland: euca_conf --upgrade-conf <old-config-file>01:34
mathiazkirkland: yes01:34
nurmikirkland: however, I am going to need to modify it to work in the new environment (with two config files)01:35
kirklandnurmi: and it does it work in-place ?01:35
mathiazkirkland: nurmi: unless euca_conf --upgrade-conf is *not* idempotent01:35
nurmimathiaz: i will make it so01:35
mathiazkirkland: mysql-server does the same thin01:35
mathiazkirkland: thing01:35
mathiazkirkland: the upgrade script is always run01:35
kirklandmathiaz: nurmi: okay, i think i know what needs to be done01:35
kirklandit will be a late night, but i'll get it done tonight01:36
kirklandnurmi: ping me if you have a rev you want me to merge01:36
nurmikirkland: mathiaz: for now, we should omit --upgrade-conf until we have a chance to test it, I think01:36
kirklandnurmi: as i'll depend on your work too01:36
mathiazkirkland: great - I'll setup a local version of the setup01:36
kirklandnurmi: agreed01:36
kirklandmathiaz: cool, let me know what you find01:36
mathiazkirkland: to make sure things are working correclty01:36
kirklandmathiaz: i'll be online for another 5 hours, probably01:36
nurmikirkland: cool, i'll start working on it tonight, but I have some other obligations and will probably not be able to finish01:37
kirklandnurmi: okay01:37
kirklandnurmi: thanks for irc'ing01:37
nurmikirkland; anytime - thank *you* guys for working to get this all right01:37
mathiazkirkland: right - so the key part for now is to have euca_conf modify eucalytpus.local.conf rather than eucalyptus.conf01:37
kirklandmathiaz: that's trivial01:38
mathiazkirkland: while waiting for the upgrade-conf the be tested01:38
kirklandmathiaz: i already changed it to edit eucalyptus-local01:38
mathiazkirkland: oh - and the upstart jobs need to be updated01:39
mathiazkirkland: to source both configuration file01:39
mathiazkirkland: and eucalyptus.conf should *not* source eucalytpus.local.conf01:39
mathiazkirkland: since it may break the C parser01:39
nurmibrb guys01:41
kirklandmathiaz: right01:42
zig_Is there anything wrong with only having LVM2 LV's?02:22
zig_And no /boot outside the LVM or anything?02:22
=== jono_ is now known as jono
jtajizig_: with grub2 it works02:25
zig_Yeah, I tested it and it works like a dream with Ubuntu 9.10 server02:26
zig_/dev/sda & /dev/sdc => /dev/md0 => LVM => root & swap LV's02:26
zig_A very elegant setup for a software RAID1 system, I think.02:26
jtajiit's been good on my thinkpad with 9.1002:27
zig_I'm pleased.02:27
zig_Second question: Have you dont an ubuntu server installation USB drive?02:27
jtajiluks on lvm with no separate /boot.. nice02:27
zig_Or rather, USB stick?02:27
zig_I didn't bother with a crypt setup.02:28
jtajiI have not messed around with bootable USB yet.. actuallly02:28
jtajiyeah I only use it because it's a laptop02:29
zulsmoser: are you using vmbuilder for the ebs stuff?02:32
smoserthat is the goal. really, the goal is to basically take root filesystem that we create for local-store instances and dump it onto a ebs volume, create a snapshot and register.02:33
zulso do you do it on an ami or on the build machine?02:35
smoseryou have to do it on an instance02:38
smoserthat part sucks02:38
smoseri mention in the spec the complications of many more moving parts02:39
zulwell now you get to do what we talked about at the last server sprint02:41
=== mrmojo is now known as billynom
=== erichammond1 is now known as erichammond
=== dantaliz1ng is now known as dantalizing
=== erichammond1 is now known as erichammond
=== erichammond1 is now known as erichammond
=== erichammond1 is now known as erichammond
jamilI have us robotics 56k message modem, How to install driver at ubuntu 8.10???04:33
=== rberger_ is now known as rberger
mcarsecan anyone help with ubuntu 9.10 and vmware server?06:01
Parabolawhats your question06:01
mcarsei can't get the vmware tools to compile in a vmware 9.10 guest06:02
Parabolaokay, what error are you getting06:02
mcarseit is not on the officially supported list for vmware server 2.006:02
mcarsewhen it is trying to build the kernel modules, it exits with Error 206:03
mcarsefailed to insert into kernel06:03
Parabolaoh my06:03
mcarseI have done this countless times without error on 9.0406:03
Parabolawill you be around tomorrow? its 1am and i need to get to bed, however i can test this from work :)06:04
Sam-I-Amyou might want to try the openvm-tools06:04
Sam-I-Amor whatever its called now...06:04
Parabolastill called that i think06:04
mcarseParabola: I can be back on tomorrow yep06:04
Sam-I-Amthey're basically vmware's tools... only some folks outside of vmware update them more quickly to support newer stuff06:04
Parabolaalright, i'll be on off and on during the day from 7am est to 4pm est06:05
mcarseis it not best to use the vmware ones though06:05
Sam-I-Amthey are the vmware tools, more or less06:05
Parabolashouldnt make a difference06:05
mcarseParabola: I will be on around 9 est06:05
Sam-I-Amsince the vmware tools are open source06:05
mcarsemaybe I am confusing openvm-tools with something else.06:06
Sam-I-Amtheres even a nice interface using the module installer thing... cant remember its name06:06
mcarseDo you think Vmware will release an updated Server where the tools support 9.10?06:07
mcarseI am patient I can wait.06:07
Sam-I-Amthey're just a tad slow06:07
mcarseslow is one word i guess.06:07
mcarsesoon to be 3 months behind.06:07
Sam-I-Amthey'll roll a patch level which will include it most likely06:07
Sam-I-Amso pay attention to randomly incrementing build numbers06:08
Sam-I-Amsometimes they dont tell you about added OS support06:08
* Sam-I-Am wanders off to sleep06:09
_rubenSam-I-Am: open-vm-tools is actually maintained by vmware themselves08:06
ploumCan anybody with an LDAP server help me to confirm this bug :09:44
uvirtbotLaunchpad bug 506317 in python-ldap "ldap.schema.urlfetch doesn't work anymore since slapd.d migration" [Undecided,New]09:44
ploumit's a matter of launching 3 lines of Python09:44
uvirtbotNew bug: #423527 in libvirt (main) "Should suspend or shut down domains on system shutdown" [Medium,Triaged] https://launchpad.net/bugs/42352709:57
uvirtbotNew bug: #506328 in squid-langpack (main) "Please sync squid-langpack 20091005-1 (main) from Debian unstable (main)." [Wishlist,New] https://launchpad.net/bugs/50632810:17
=== freeflyi1g is now known as freeflying
uvirtbotNew bug: #506316 in eucalyptus (main) "[lucid] New config file system fails to support node registration" [High,Fix committed] https://launchpad.net/bugs/50631610:51
=== giLL0r_ is now known as giLL0r
underdevHi! Does anyone know how to run a server in the background that will persist after the shell is closed?11:41
cemcunderdev: if that server doesn't have any -d (as in daemon) switch, or config file option, you could try runinning it with '&' at the end, like ./server_daemon &11:42
underdevcemc- ty11:44
cemcunderdev: if you go with the '&' one, you might want to redirect output to some logfile, like ./server_daemon > /var/log/logfile.log 2>&1 & (if you need the output)11:45
=== dendrobates is now known as dendro-afk
=== deegee_1 is now known as deegee
ragsIs there a way I can configure directory level quota's?13:06
ttxzul: i've split the canonical-app-support work items to alpha3... please doublecheck, if there is anything you plan to complete by Thursday, please move them back to alpha213:21
zulttx: ack13:22
ttxzul: also for seeds I still assume that it will be completed by alpha213:22
ttxso I didn't defer any item to alpha313:22
zulttx: k13:23
zulttx: umm we are still discussing packages for the server-lucid-seeds packages13:25
ttxzul, mathiaz: but the discussion should end before thursday, right13:26
ploumdoes anybody have an idea why  phpldapmyadmin is displayed recursively when I install it ?13:26
ploum(fresh phpldapadmin install !)13:26
zulttx: afaik yes13:26
zulautofs5 is done on our side now pitti just has to flick the switch13:30
jiboumansttx, zul: mathiaz confirmed to me that he'd close the dicussions on monday actually13:30
zuljiboumans: ah good13:31
jiboumansduring our 1:1 last friday13:31
zulso we just have to wait for him to wake up ;)13:31
ttxjiboumans: so that kinda spills over as well.13:31
jiboumansttx: yesterday's monday13:31
ttxjiboumans: ah13:31
ttxthat's good :)13:31
jiboumansttx: best follow up with him on that then, as i haven't seen a conclusion mail yet13:31
ttxI don't mind an exception for something requiring wider consensus/discussion, but that should be the exception rather than the rule13:32
jiboumansand there was some traffic overnight still13:32
jiboumansttx: agreed13:32
zulim still trying to beat ctdb into shape13:32
Italian_Plumber3how do I audit my startup process, to see if, or make sure that, one service is started before another?  I need to do that with NTP and networking.14:12
cemcItalian_Plumber3: I think ubuntu takes care of that. take a look in /etc/network/if-up.d/ntpdate14:19
Italian_Plumber3i'm not using ntpdate I'm using npt (the server)14:21
smosergood morning all.14:22
smoserttx, ping14:22
smoserbug 50633214:23
uvirtbotLaunchpad bug 506332 in ec2-init "Can't login with SSH to Lucid UEC image" [High,New] https://launchpad.net/bugs/50633214:23
ttxsmoser: yo14:23
smoserwhere is that one? its fixed?14:23
ttxsmoser: no it isn't14:23
ttxI was wondering what broke it14:23
Xpistos|workHi everyone14:24
smoserhm... i havent tried 20100112, maybe something busted outside ec2 init14:24
ttxsmoser: fwiw it still boots and starts an SSH server14:24
cemcItalian_Plumber3: scroll further down, there is a bit where it's restarting ntpd, too14:24
smoser20100111.1 is the first first with new ec214:24
Xpistos|workCan someone help me with a message I am getting when I try to apt-get upgrade?  "The following packages have been kept back:   linux-generic-pae linux-headers-generic-pae linux-image-generic-pae"  I am no sure why anything is being held back?14:25
RoyKhttp://www.daily.pk/norway-time-hole-“leak”-plunges-northern-hemisphere-into-chaos-14311/ <-- whee - it's all our fault!14:26
ttxsmoser: let me know how I can best assist you on that one. Try on EC2 ? Try 20100111.1 on UEC ?14:26
smoseroh, i see. so ssh is up but you can't log in. its probably timing, as i dont know what changed between 2010111.1 and 20100112 (nothing from me)14:26
ttxsmoser: haven't tried 20100111.114:27
ttx20100108.1 works14:27
ttxsmoser: want me to try 20100111.1 on UEC ?14:27
smoserah. is uec easily installable? should i install it locally?14:28
ttxsmoser: it's installable, but maybe not the best use of your time14:29
ttxsmoser: what did you tested so far ? 20100111.1 on EC2 ?14:29
ttxsmoser: ok, then try 20100112 on EC2, I'll try 20100111.1 on UEC14:30
smoser'euca-get-console-output stops before "ec2: Generating public/private rsa key pair." on the affected images.'14:30
smoserthats a bug, there wont be the ec2 fingerprints to console at the moment.14:30
ttxsmoser: ok, known bug ?14:30
smoseri should open a bug, but yes, i was aware that i'd not added that into ec2-init 0.5.x14:31
smoseri just verified that 20090112 booted on ec2 and i reached it14:31
ttxsmoser: downloading 20100111.1 for UEC testing14:33
smoseri really dont expect any difference there.14:33
Malekohi. whats the simplest way to track down the time that a server spends to reboot14:33
ttxsmoser: well, there is a difference between 20100108.1 and 2010011214:34
smoserwell yes. but ec2-init 0.5.0 went into 20100111.114:34
ttxsmoser: you think it's the ec2-init stuff that landed in 20100111.1 ?14:34
smoserso theres a big difference.14:34
smoserwell right now i'd think thats the smoking gun14:34
ttxsmoser: i'll narrow the regression to 20100111 -> 20100111.1 just to be sure14:35
ttxsmoser: you need a UEC to debug this ? Do you have a karmic UEC leftover ?14:36
smoseri think i might have lucid alpha114:36
smoserwhich ~ karmic, right?14:36
ttxsmoser: yes14:36
ttxanything that saves you from reinstalling it is good14:37
smoseri'll resurect it and see.14:37
ttxsmoser: if I can do anything apart from this narrowing-down, let me know14:37
ttxI'll have a look at ec2-init to see if anything rings a bell14:37
ttxsmoser: kvm should be pretty close anyway14:38
smoseri had been testing in kvm. i can test it too.14:38
ttxsmoser: also see slangasek message from ~3 hours ago on #ubuntnu-devel14:39
ttxabout "lucid-server-amd64.tar.gz"14:39
ttxinstead of "lucid-uec-amd64.tar.gz"14:40
smoseri dont see anything on -devel14:41
ttxsmoser: see pm14:42
bogeyd6is it even possible to burn an iso from a network share?14:46
kworksure it is14:46
kworkwill it work :P? who knwos14:47
uvirtbotNew bug: #506317 in python-ldap (main) "ldap.schema.urlfetch doesn't work anymore since slapd.d migration" [Undecided,Invalid] https://launchpad.net/bugs/50631714:51
=== robbiew_ is now known as robbiew
Xpistos|workCan someone help me with a message I am getting when I try to apt-get upgrade?  "The following packages have been kept back:   linux-generic-pae linux-headers-generic-pae linux-image-generic-pae"  I am no sure why anything is being held back?15:00
RoyKXpistos|work: just apt-get remove them and add new replacements. being held back means apt/dpkg can't do an automatic upgrade for some reason. removing and re-adding replacements usually work15:03
Italian_Plumber3what is the best way to automatically keep the clock updated on an ubuntu server?15:03
Xpistos|workRoyK: So remove linux-generic-pae and then reinstall it?15:04
ParabolaXpistos|work - use apt-get dist-upgrade15:07
Parabolathat will install thoes for you15:07
Xpistos|workParabola: which is prefered upgrade or dist-upgrade?15:07
Parabolai personally do dist upgrade15:08
Parabolabut that will upgrade the kernel if its availible :)15:08
Parabolajust upgrade is "safer" however if you need thoes packages updated15:08
Parabolathen dist-upgrade is the way to go, you could also block kernel upgrades, and then use dist-upgrade15:09
Parabolawould be a little safer15:09
Parabola(again i've been using dist-upgrade on my webservers for sometime, never had an issue)15:09
JanCwell, upgrading kernels might be useful in case of security issues  ;)15:09
psyferrehey folks, does anyone know if there's a stripped down package of gnome that can be installed on ubuntu server?  I'd like to use gnome once in a while, but i don't need open office, etc that comes with the ubuntu-desktop package...15:17
deianyone here familiar with dns?15:25
uvirtbotNew bug: #506464 in squid (main) "snmp not enabled in squid" [Undecided,New] https://launchpad.net/bugs/50646415:26
cemcdei: somewhat15:27
deion my dns server15:27
deifor my dns server rather15:27
dei1: Set Subnet with Registrar (i.e. NS8.blahblah.com equals 149.242.440.1015:28
dei2: Created Zone at above ip15:28
dei3: Created A record, Set NS Records, and created PTR records on Server at above ip15:28
deia record = ns8.bahblah = 149.242.440.10 [for example]15:29
deins record = ns8.blahblah.com15:29
deiptr record = 149-242-440-10.myserver.myisp.com15:29
deiis that how its supposed to be done?15:30
bogeyd6dei would you like me to pastebin my dns setup?15:30
deiim running powerdns15:31
deiits functional..15:31
bogeyd6well you are taking the right steps15:32
deii mean15:32
deidig @ns0.osmiumdefensesystems.com any osmiumdefensesystems.com15:32
deithat works15:32
deiits just15:33
deiwhen im doing squish, for example15:33
deisometimes you see stuff like this15:33
deiReferral: com is at L.GTLD-SERVERS.NET - IP not supplied15:33
deibut it will say15:33
deiReferral: com is at NS0.OSMIUMDEFENSESYSTEMS.COM - IP not supplied15:33
deiAsking A0.ORG.AFILIAS-NST.INFO ( for paschool.org (type A)15:35
deiReferral: paschool.org is at ns0.osmiumdefensesystems.com - IP not supplied15:35
deiReferral: paschool.org is at ns8.osmiumdefensesystems.com - IP not supplied15:35
henriquev"update-rc.d my-script 2" => I get this: update-rc.d: warning: /etc/init.d/plifk missing LSB information15:37
henriquevand it doesn't do anything else... Like something is wrong15:37
subit's just a warning. did you confirm that it made the appropriate symlinks?15:39
henriquevit didn't15:39
henriquevif I use update-rc.d my-script defaults it does15:40
henriquevthe man page doesn't help15:42
ttxsmoser: confirmed regression is between 20100111 and 20100111.115:44
smoseri think its timing15:44
smoseryour input is useful, yeah.15:44
ttxsmoser: anything I can test ?15:44
smoseri just verified that 20100112 boots fine in kvm with ec2 metadata inserted15:44
smoseri might know, yeah15:45
smoseryou can get to a 20100111 ?15:45
smoserinside it, it should have boto 1.915:45
ttxI'm in15:45
smoserpython  -c 'import boto.utils; import pprint; pprint.pprint(boto.utils.get_instance_metadata())'15:46
smoserpastebin that output if you can15:47
smoserhttp://pastebin.com/f7939e1a7 is what i get15:48
smoserso metadata service in uec is not identical to ec2. i'm using that instance metadata call to get the keys15:49
ttxsmoser: I'd say, I'm not surprised15:50
smoseri think i can work around.15:50
ttxsmoser: sure, but don't forget to file a bug against eucalyptus15:50
smoseryeah, i should have expected that. boto just such a nice job of crawling the data to a object15:50
ttxsmoser: please update the bug to "In Progress" with a comment about the lead you're following now15:51
ttxsmoser: that's workaroundable in time for alpha2, right15:52
smoseryeah, i think so15:53
ttxsmoser: let me know as soon as you respin an image and I'll validate the fix15:53
smoserwhat euca version is that?15:54
ttxThe current, lucid one15:54
ttxmathiaz: just checking, you have 5 work items left, 4 in uec-testing. All those should still get completed for alpha-2, you don't expect any of them requiring to defer to alpha-3 ?15:56
mathiazttx:  * Automate UEC upgrade of existing 9.10 CLC+Walrus+CC+SC / NCs: TODO15:57
mathiazttx: ^^ probably defer to alpha315:57
mathiazttx: and running functional test suites require a working UEC which have may have today15:58
ttxmathiaz: it's working now15:58
ttxmathiaz: minus a small bugfix I committed that didn"t land on the ISO yet15:58
mathiazttx: great - I'll reinstall my UEC setup later today then15:58
ttxmathiaz: to work around it, just add NODES="" to /etc/eucalyptus/eucalyptus.local.conf before starting to add nodes15:59
ttxeuca_conf can't modify a valud that's not tere in the first place15:59
ttxmathiaz: if you netboot, you should get the fixed version16:00
mathiazttx: right16:00
mathiazttx: I do package installation - so 1.6.2~bzr1124-0ubuntu3 should be installed16:00
ttxmathiaz: yes.16:00
mathiazttx: great - so I should be able to run the functional tests later today16:01
ttxmathiaz: ok, just ack with jib the deferred items and move those you can't do to alpha3 on the whiteboard by the end of your day16:01
mathiazttx: okidoki16:02
zulmathiaz: hi mathiaz, do you have a list of packages we can demote to universe?16:05
mathiazzul: already done16:29
mathiazzul: FYI http://wiki.ubuntu.com/LucidServerSeeds16:30
zulyou filed them in launchpad?16:30
zulmathiaz: ah ok16:31
=== fahadsadah is now known as Guest14190
ikthusHi all :) it's me17:14
ikthusI want to install ubuntu server on an IBM 84805AX, but i dont see any hard drive in the description of the bios, but there is already a system installed in this machine.17:19
ikthussee desc here http://servers.alege.net/IBM-eserver-xSeries-205-8480-84805AX.html17:19
smoserttx, see bug 50633217:21
uvirtbotLaunchpad bug 506332 in ec2-init "Can't login with SSH to Lucid UEC image" [High,In progress] https://launchpad.net/bugs/50633217:21
=== [1]cab938 is now known as cab938
uvirtbotNew bug: #504368 in qemu-kvm (main) "sdl window intermittently scales instead of resizing" [Undecided,New] https://launchpad.net/bugs/50436817:32
smosermathiaz, kirkland anyone able to test https://bugs.launchpad.net/ubuntu/+source/ec2-init/+bug/50633217:51
uvirtbotLaunchpad bug 506332 in ec2-init "Can't login with SSH to Lucid UEC image" [High,In progress]17:51
mathiazsmoser: I should be able to17:51
smoseri've tested locally, seen it fail and then succeed, but i'd like someone else to if possible.17:52
mathiazsmoser: I've just installed a UEC setup17:52
mathiazsmoser: and I'm about to register the latest image17:52
mathiazsmoser: latest *lucid* image17:52
smoserwell, you wont be able to ssh to it :)17:52
mathiazsmoser: would that work for the test?17:52
smoseryou can verify it fail17:52
smoserthen fix by mount -o loop <img> /mnt && install-package-into-image && umount && register modified image17:53
smoserthat should show you fail and then fix17:53
mathiazsmoser: ok - I'll do that17:54
mathiazsmoser: looking at the bug - should this be a bug against eucalyptus meta-data service?17:56
mathiazsmoser: is eucalyptus meta-data supposed to be the same as the on on EC2?17:56
smosermathiaz, you're correct, yes.17:58
smoserit should be but it isn't17:58
mathiazsmoser: right - I'll open a task against eucalyptus then17:58
smoserso work around in ec2-init for now. and should open a bug against metadata service, or at least a task here.17:58
kirklandmathiaz: were you able to take a look at my eucalyptus upload?18:01
mathiazkirkland: yop - works like a charm!18:02
kirklandmathiaz: implementing the last round of eucalyptus.conf changes18:02
kirklandmathiaz: beautiful :-)18:02
=== aubre is now known as aubre_afk
mathiazkirkland: I've got a running UEC now !18:02
kirklandmathiaz: ttx said it worked too :-)18:02
mathiazkirkland: thanks for all your hard work on this one!18:02
kirklandmathiaz: so you're testing smoser's bug for him?18:02
kirklandmathiaz: sure, think.  thank *you* for your patience :-)18:02
nurmikirkland: mathiaz: ttx: nice!18:02
mathiazkirkland: well - I think it's an issue with eucalyptus18:02
mathiaznurmi: oh - you're here18:02
kirklandnurmi: o/18:03
mathiaznurmi: bug 50633218:03
uvirtbotLaunchpad bug 506332 in ec2-init "Can't login with SSH to Lucid UEC image" [High,In progress] https://launchpad.net/bugs/50633218:03
kirklandnurmi: that'll teach you to speak up :-)18:03
nurmikirkland: hehe :)18:03
nurmimathiaz: let me take a looksee18:03
kirklandmathiaz: i have fully automated installs of all 5 of my UEC machines now18:04
kirklandmathiaz: i can trigger them from anywhere with a web browser :-)18:04
mathiazkirkland: :)18:04
kirklandmathiaz: completely hands off :-)18:04
mathiazkirkland: how do you deal with ssh public keys?18:04
mathiazkirkland: hm - never mind - it's part of the installer18:04
kirklandmathiaz: i'm planning on bringing all of it with me to PDX18:04
kirklandmathiaz: yeah18:04
mathiazkirkland: so you're planning to go through airport security with 6 computers?18:05
kirklandmathiaz: right now, i do a ssh-copy-id after first reboot from my primary desktop18:05
kirklandmathiaz: i'd like to automate that18:05
kirklandmathiaz: that's the plan18:05
nurmimathiaz: looks like smoser has a new ec2-init that18:05
mathiaznurmi: well - it's a workaround18:05
mathiaznurmi: is eucalyptus supposed to provide the same metadata as EC2?18:06
nurmimathiaz: boto 1.9 change incompatible with eucalyptus meta-data, it appears18:06
smoserboto 1.9 crawls the metadata service. Euca is close, but not the same. result is that the public-keys come out a bit different when its crawled.18:06
smoserthere are some other differences there too18:06
nurmismoser: yep, we'll take a look18:07
mathiaznurmi: ok - so can we say that it's a bug that should be fixed in eucalyptus?18:07
nurmismoser: is there a place where you have/can document the differences?18:08
mathiaznurmi: and for alpha2 we could upload a workaround in ec2-init?18:08
nurmimathiaz: not enough information yet, I need to see the bits that are actually different to determine how Eucalyptus should behave - this is frequently a 'documented spec versus implementation (boto) versus reality (AWS)' problem18:09
smosernurmi, there are 2 files there, vimdiff shows fairly well.18:09
mathiaznurmi: ok - so it seems like it won't be fixed in time for alpha218:10
nurmismoser: do we know what difference is actually breaking boto?18:10
smosernurmi, boto is just crawling the metadata service. for data. i just posted a comment with the differences.18:12
smoserregarding the public keys, there are 2 differences18:12
smoseri'd have to look more at what the web service is responding with, but my guess is that uec is listing 'public-keys/0' in the index of the metadata/18:13
nurmismoser: yep, instead of a 'list'18:13
smoserand then the similar with the public-keys18:14
nurmismoser: this looks like the API has changed, we'll have to ensure that the version of boto and Eucalyptus support the same AWS API version, which it appears is not the case18:15
smosernurmi, well, in the actual use case, ec2-init requests the 2009-04-04 api18:16
smoserhopefully they're not changing things inside an api level.18:17
smoseri can't say for certain, but you'd think that is pretty bad behavior, if one day has one set of files and thne later it has a different set18:18
smoserkind of pointless to have the versions if you're going to do that18:18
=== aubre_afk is now known as aubre
nurmismoser: of course, i'm actually trying to make sure that we can claim that EUcalyptus fully supports the API that you're using to develope ec2-init/boto18:22
nurmismoser: thanks, I've got the necessary information to figure out the problem and communicate with you guys on a solution18:23
dnaumovdoes Ubuntu automatically recognize when its being installed on an SSD an setups the partition alignment and sector/track sizes accordingly (like Windows 7 and Win2008 R2 do) or is there a lot of crazy manual fdisk trickery involved?18:23
smosergiven that attachment you should pretty easiliy be able to run your own 'python -c "import boto.utils, pprint; pprint.pprint(boto.utils.get_instance_metadata())"' and compare18:24
nurmismoser: we need to start with the spec, and then will take into account what a third-party expects (i.e., the objects you posted as part of the bug report)18:26
nurmismoser: thanks for that, those are really useful reports for us18:26
mathiazkirkland: are you able to run instances on the latest lucid UEC?18:35
mathiazkirkland: my setup isn't able to start instances - the error in nc.log is: libvirt: Failed to add tap interface to bridge 'br0': No such file or directory (code=38)18:36
kirklandmathiaz: i haven't tried today18:36
kirklandmathiaz: ttx said he was able to, though18:36
mathiazkirkland: hm - there is a new kernel available - let me upgrade first18:37
mathiaznurmi: where is the metadata service running? on the CC or on the CLC?18:49
nurmimathiaz: CLC18:49
xteejxHey guys, anyone know anything about sudo-ldap?18:50
mathiaznurmi: and the IP address is supposed to be on the CC or CLC?18:50
nurmimathiaz: CC18:50
nurmimathiaz: if you are running a CLC and CC on different hosts, you have to set VNET_CLOUDIP="CLC-IP" in eucalyptus.conf on the CC18:51
nurmimathiaz: the CC sets up an iptables redirect for to the CLC-IP:877318:52
mathiaznurmi: ok - that's the setup I have18:52
mathiaznurmi: CLC+Walrus on one host18:52
xteejxIf there is anyone around that understands ldap and is able to help with bug 115967, please do so, it would be appreciated :)18:52
uvirtbotLaunchpad bug 115967 in sudo "sudo-ldap not working with ldaps" [Medium,Confirmed] https://launchpad.net/bugs/11596718:52
mathiaznurmi: CC+SC on another host18:52
mathiaznurmi: when you mentioned VNET_CLOUDIP="CLC-IP" - should "CLC-IP" be used? or the actual CLC IP address (X.X.X.X)?18:54
mathiaznurmi: nervermind - I've read the documentation18:55
smosermathiaz, did you verify that fix?19:08
mathiazsmoser: not yet19:13
mathiazsmoser: I'm not able to start instances on my UEC setup - debugging it19:14
mathiaznurmi: is that the iptables rules supposed to handle the metadata service on the CC:     0     0 DNAT       tcp  --  *      *     tcp dpt:80 to:
mathiaznurmi: ?19:19
nurmimathiaz: yep, that rule implies that VNET_CLOUDIP was not set (the default is to set the redirect to 'itself')19:24
nurmimathiaz: if you have changed eucalyptus.conf, then you'll have to do a 'clean' restart for the change to take effect19:24
mathiaznurmi: ok - let me double check my configuration files19:25
nurmimathiaz: iirc, the upstart script honors a "restart eucalyptus-cc CLEAN=1" style argument19:25
mathiaznurmi: I've added VNET_CLOUDIP="" to eucalyptus.local.conf19:26
mathiaznurmi: and rebooted the CC19:26
ttxmathiaz: you'd still have to use CLEAN=119:26
ttxnetwork state is preserved accross reboots19:27
mathiaznurmi: ttx: \o/ - works thanks19:27
ttxsmoser: testing your fix19:28
marks256i'm trying to compile the 2.6.31 kernel to include the Lustre FS patch, but i am getting an error: Failed to create a ./debian directory: No such file or directory at /usr/bin/make-kpkg line 1048. Any ideas on how to solve this?19:31
mathiazmarks256: I'd ask in #ubuntu-kernel19:31
=== rberger_ is now known as rberger
marks256mathiaz, ok thanks19:32
ttxnurmi: is rebooting an instance supported ?19:40
nurmittx: it should be, i havn't tried with the latest KVM in lucid recently (that is the part that varies)19:41
ttxnurmi: just tried to do a test for smoser -- I get "restarting system" as the last message in get-console-output and then nothing19:42
=== gnac_ is now known as gnac
nurmittx: if the libvirt/kvm combination supports reboot, then eucalyptus should work as well19:42
ttxnurmi: looks like it doesn't :)19:42
ttxnurmi: any way to trigger it artifically on the NC ?19:43
ttxvirsh start <instance> ?19:43
mathiaznurmi: hm - so the CLC needs also to be able to route to the public IPs of the CC19:43
mathiaznurmi: IIUC instances asking for meta-data will have their source IP set to their public IP19:43
nurmittx: if that didn't work, then I don't know another way, perhaps a 'start' (you'll have to do it fast, though, before euca marks it as terminated19:43
* ttx runs19:44
mathiaznurmi: if the CLC doesn't have a route setup to the CC things are not working correclty19:44
nurmimathiaz: the public ips?19:44
mathiaznurmi: right - there a SNAT rule on the CC19:44
nurmimathiaz: those should be on the same subnet as the CC, which also must have an address that is routable from the CLC?19:45
nurmimathiaz: nod19:45
nurmimathiaz: oh i see19:45
nurmimathiaz: you're specifying public IPs that are on a different subnet from the CC's public interface19:45
mathiaznurmi: yes19:46
mathiaznurmi: and the CLC doesn't know about it19:46
nurmimathiaz: yeah, the CLC must be able to route to the instance's public IPs19:46
nurmimathiaz: for the meta-data service to work19:47
mathiaznurmi: right - a workaround is to masquerade that traffic19:47
mathiaznurmi: adding an iptables rules that just masquerade traffic to 169.19:48
nurmimathiaz: without the source IP, though, the meta-data service will not know which data to return19:49
mathiaznurmi: ah good point19:49
nurmimathiaz: that is the only reason we don't use MASQ (which would be much simpler :)19:49
mathiaznurmi: right - I guess that documenting that configuration is enough for now19:50
mathiaznurmi: another solution would be to proxy meta-data requests via the CC at the http level19:50
mathiaznurmi: but that would require some redisgn19:50
nurmimathiaz: yep, we've considered that as well - the complexity comes from the fact that some of the data is dynamic19:51
mathiazsmoser: so the instance is stuck here in the boot process: http://paste.ubuntu.com/355668/19:52
nurmimathiaz: in that it can change at run-time (so, we have a distributed state problem, which we try to avoid whenever possible)19:52
ttxmathiaz: what did you test ?19:52
mathiazttx: the current-non-fixed image19:53
ttxmathiaz: in fact it's not stuck19:53
mathiazttx: is this what I'm supposed to see?19:53
ttxmathiaz: yes, but it's not stuck19:53
ttxmathiaz: ssh is running19:53
ttxmathiaz: we are missing the ec2: messages but that's anither bug19:54
ttxmathiaz: that's what smoser said, at least19:54
ttxsmoser: I've trouble testing19:54
smoserwe are definitely missing ssh messages.19:54
ttxsmoser: looks like I can't reboot19:54
mathiazsmoser: well - I can't connect to the instance19:54
ttx(not your fault)19:54
mathiazsmoser: the ssh client is stuck at connecting19:55
ttxsmoser: I'm losing the instance if I reboot19:56
smosermathiaz, i dont know. can't think of why it would be stuck where it is.19:56
smoserwhere does it go to?19:56
mathiazsmoser: ok - I'm trying to figure out if ssh is running according to the message log19:56
smoserinstances in kvm reboot fine, i've verified that locally.  same is true on ec2.19:57
smoserobviously timing or something could break that.19:57
smoseri should install euc19:57
nurmismoser: maybe the libvirt abstraction?19:57
ttxsmoser: I reboot, then I get "restarting system" as the last message in get-console-output19:58
nurmismoser:ttx:mathiaz: general euca/VM debugging tool - run /usr/share/eucalyptus/gen_kvm_libvirt_xml to see the libvirt 'template' that we use to run the VM19:59
ttxmy geuss is that it's stuck at "rebooting", and never actually reboots19:59
nurmittx: there are some libvirt options to set various behaviors when libvirt recieves the 'reboot' call, playing with adding them to the template may be useful20:00
mathiazsmoser: ttx: it works!20:01
ttxsmoser: any other way to exercise the code, without rebooting ?20:01
mathiazsmoser: ttx: I can connect to the ssh daemon running on the instance20:01
smosermathiaz, was it just slow to come up?20:02
mathiazsmoser: and I see the public key denied20:02
ttxmathiaz: that's where I am20:02
mathiazsmoser: I can't say - it was an issue with my setup20:02
smoserso you see public key denied? thats not "works", though20:02
* mathiaz tries the proposed-fix20:02
mathiazsmoser: well - that's what was expected with the current-not-fixed image?20:02
smoserok. good.20:03
ttx<ttx> smoser: any other way to exercise the code, without rebooting ?20:03
smoserttx, yeah, you can mount -o loop the image and install the new pkg20:04
smoserthen register20:04
smoserand start instance20:04
smoserthat is what mathiaz is doing now20:04
smoserttx, actually, you might not need ar eboot20:04
smosertry just rm -Rf /var/lib/cloud/*, then start ec2init20:05
ttxrebooting works if I don't install the modified ec2-init20:05
mathiazsmoser: \o/20:08
mathiazsmoser: works!20:08
mathiazsmoser: I was able to login into the -fixed instance20:08
smoserwell good to have one happy customer, but why not ttx.20:08
mathiazsmoser: however the ssh public key is not available in the console.log20:09
* mathiaz files a bug20:09
racquadhi guys, I'm with trouble with .htaccess file and userdir module. .htaccess is ok in regular paths, but doesn't work at all under userdir paths20:09
ttxsmoser: maybe I miss something. I usually do miss things when I work late20:09
ttxmathiaz: could you reproduce the reboot issue ? I can reboot a pristine 20100111 instance, but if I install the new ec2-init before rebooting, then it fails20:10
mathiazttx: by rebooting you mean just logging and running the reboot command?20:11
ttxmathiaz: yes20:11
mathiazttx: how do you do on a pristine 20100111 where you can't login?20:11
ttx20100111 is fine20:11
ttx20100111.1 is broken20:12
smoserwhere are instructions for installing uec.20:12
smoseror do you think i should just try to get the alpha1 uec up20:12
ttxsmoser: you should just use your alpha1 install20:12
mathiazsmoser: I'd try alpha2 - it's working20:13
smoserif you hold your mouth right.20:13
ttxmathiaz: he already has a setup20:13
mathiazsmoser: now it depends how you plan to install it? iso?20:13
smoseryou've got a very friendly definition of "working" :)20:13
mathiazttx: reboot works for me20:13
mathiazttx: the instance has rebooted according to the console.log file on the NC20:14
ttxmathiaz: reboot what20:14
ttx20100111+new ec2-init ?20:14
mathiazttx: and I can login into ssh20:14
mathiazttx: 20100111.1 + new ec2-init20:14
ttxok then :)20:15
mathiazsmoser: I think we're good here20:16
ttxmathiaz: looks like it's worth at least an upload and an image respin20:16
* mathiaz nods20:16
ttxI'll let you guys sort it out20:16
smoseralright. please sponsor that patch then.20:16
mathiazsmoser: I'm going to upload your ec2-init20:16
ttxand pick it up for testing tomorrow morning20:16
=== dendro-afk is now known as dendrobates
smoseri'll spin an image build now20:16
smoserWarning: failed to parse error message from AWS: <unknown>:1:0: syntax error20:17
smoserEC2ResponseError: 403 Forbidden20:17
smoserFailure: 403 Forbidden20:17
smoserdoes that ring a bell ?20:17
smoserfrom euca-describe-images20:17
ttxsounds like the deadlock thing20:17
ttxtry sudo stop eucalyptus20:18
ttxsudo start eucalyptus20:18
ttxbug 44435220:19
smoserthat fixed it20:19
uvirtbotLaunchpad bug 444352 in ubuntu-release-notes "DB deadlock on reboot prevents UEC from working, temporarily - 403 Forbidden errors" [Undecided,Fix released] https://launchpad.net/bugs/44435220:19
mathiazsmoser: allright - ec2-init uploaded20:22
maestrojedI have built LAMP server with Ubuntu. I am missing a mail server that would allow me to use php's sendmail(). What would be the easiest to install/get up and running?20:34
Xodiac13i need help with the error i get i have torrentflux installed and when i try from another computer i get this error SSL received a record that exceeded the maximum permissible length.20:48
Xodiac13i had no problem with it when i was getting another server ready but now idk20:49
marks256If anyone here has complied their own kernel, what is your average compile time?20:51
Xodiac13marks256: i need help with torrentflux i get this error SSL received a record that exceeded the maximum permissible length.20:55
jpdsmarks256: https://edge.launchpad.net/ubuntu/+source/linux/2.6.32-10.14/+build/143468020:56
Xodiac13jpds: i need help with torrentflux i get this error SSL received a record that exceeded the maximum permissible length.20:56
jpdsXodiac13: I know, I saw. But I've never used torrentflux, so; no idea.20:57
Xodiac13jpds: when i try to go to port 80 thats when i get the error20:57
marks256jpds, ouch. 3.5 hours. haha. ok thanks20:57
Xodiac13jpds: do you know anyone that knows about it20:57
marks256Xodiac13, google?20:57
Xodiac13mark256: i tried google see i never had this problem or if i can find a tutorial or something on how to get it working i already tried port forwording20:58
Xodiac13mark256: but then i realized its probably some config file or the certificate is giving me an error20:58
mathiazzul: what are the plan for php5 in lucid?21:01
mathiazzul: are we sticking to 5.2 or upgrade to 5.3?21:01
Xodiac13join #torrentflux21:02
marks256put a slash in front Xodiac1321:02
Xodiac13mark256: thanks21:02
=== luis__lopez is now known as luis_lopez
marks256"/join #torrentflux"21:03
Xodiac13o okay duh thanks21:03
Xodiac13no one there21:03
diagoI need to do change passwords for ldap, but I can't figure out what the salt is to do so. Does anyone know where I can find this?21:13
zulmathiaz: i would like to stick with 5.2 and the security team probaby wants to stick with 5.2 as well21:16
mathiazzul: why?21:17
zulmathiaz: i think 5.3 is still buggy21:17
zuland its still in debian experimental21:18
zulmathiaz: i wouldnt mind having it in universe if it plays nicely and if someone takes care of security for it21:22
cab938I'm running a script as sudo, but I want to execute some lines as the user who is currently executing the script21:23
cab938can I do this without hardcoding their username?21:23
qman__cab938, the only way I know of to do that without hardcoding the user is to put the 'sudo's for the root stuff inside the script, and simply not sudo the lines that don't need root21:28
qman__you could also grab the UID at the beginning, but you'd still have to put the sudo inside the script for it to work21:29
Xodiac13i need help with torrentflux i get a ssl error saying its too long21:34
uvirtbotNew bug: #506332 in ec2-init "Can't login with SSH to Lucid UEC image - uec metadata service returns a string instead of a list of string for ssh public keys" [High,In progress] https://launchpad.net/bugs/50633221:36
zulajmitch: so you are going to try to get 5.2 and 5.3 playing side by side?21:37
ajmitchI can give it a go, depending on how annoying the package renaming is21:37
zulmathiaz: ^^ there you go :)21:38
ajmitchhopefully 5.3 will actually be stable enough this time to replace 5.221:38
EtienneGok, there we are!21:43
EtienneGcontext for the rest of the channel participants: I was asking mathiaz privately about the best network topology for running UEC21:43
mathiazEtienneG: IIRC the CC is already doing NAT21:43
mathiazEtienneG: so you don't need to do anything special21:44
EtienneGmathiaz, really?  that would be awesome!21:44
mathiazEtienneG: note that the CLC will need proper routes to the CC to talk to the *public* ips of the instances21:44
mathiazEtienneG: otherwise the meta-data service won't work (I've run into this issue today)21:44
EtienneGmathiaz, that is a given, I will running the CLC and CC on the smae machine anyway21:44
mathiazEtienneG: ah - so things will be simpler then21:45
mathiazEtienneG: you don't need NAT then21:45
EtienneGmathiaz, but I would be running Walrus on a separate machine, though21:45
EtienneGso I guess tha yes, NAT will be needed for the NC to fetch the AMI.  right?21:46
mathiazEtienneG: hm - I'd suggest to run CLC+Walrus then21:46
mathiazEtienneG: some topologies haven't been tested21:46
mathiazEtienneG: yes - you're right.21:46
EtienneGmathiaz, so: (CLC + Walrus)  <--->  (CC + SC)   <---> (nc, nc, nc, ...)21:47
mathiazEtienneG: right - that's an good setup21:47
mathiazEtienneG: I'm running something similar21:47
EtienneGmathiaz, ok, i will do that then21:47
mathiazEtienneG: same setup except with only one network21:47
EtienneGmathiaz, thanks for the clarification, otherwise I would have recommended (Walrus + SC) -> (CLC + CC)  -> nc, nc, nc21:48
EtienneGwould not have worked so well21:48
mathiazEtienneG: the SC and CC need to be on the same network21:48
EtienneGmathiaz, yes, right, missed it!21:49
mathiazEtienneG: the SC uses aoe as it provides the block storage service21:49
EtienneGmathiaz, if i am not mistaken, the CLC needs to talk to the SC, right?21:49
mathiazEtienneG: https://wiki.ubuntu.com/FoundationsTeam/UECInstallerEnhancement21:51
mathiazEtienneG: ^^ has a diagram of who needs to connect to whom21:51
EtienneGmathiaz, that is terribly useful.  thanks!21:56
=== MTecknology is now known as MTeck-ricer
smosermathiaz, ttx, kirkland there is a 20100112.1 build running, it will contain the updated ec2-init22:10
mathiazsmoser: great - thanks!22:11
smoserit will also have output named lucid-server-uec-arch22:11
smoserrather than lucid-server-arch22:11
xteejxhey guys, anyone here?22:26
xteejxcan someone take a look at bug 115967 please? It affects Karmic, and may affect Lucid, both server. From a triaging point of view does it need any more debugging information? ldap doesn't have a documented triaging procedure and I've asked our Bug Control team.22:29
uvirtbotLaunchpad bug 115967 in sudo "sudo-ldap not working with ldaps" [Medium,Confirmed] https://launchpad.net/bugs/11596722:29
uvirtbotNew bug: #506667 in euca2ools (main) "euca-describe-volumes fails with  'Volume' object has no attribute 'availabilityZone'" [Undecided,New] https://launchpad.net/bugs/50666722:36
=== erichammond1 is now known as erichammond
gregcoithi all.  if "netstat -na|grep 6082" returns "tcp        0      0*               LISTEN", is that port open to remote connections?22:44
=== erichammond1 is now known as erichammond
=== A_Tuin is now known as A-Tuin_
macliHi, curious, what test automation software ubuntu server use for testing?23:06
=== alex88 is now known as alex88[sleep]
=== robbiew is now known as robbiew_

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