| melmoth | is rebooting a zookeeper node on a maas installation supposed to work ? | 00:17 |
|---|---|---|
| melmoth | i did it on 2 sepearate installations, and after a reboot, the agent-state of the zookeeper node is being kept on "not-started" state | 00:18 |
| lifeless | thats more a #juju question, sorry! | 00:18 |
| lifeless | SpamapS: ^ | 00:18 |
| melmoth | zookeeper runs, but it s log does mention thing likes "xpiring session 0x139bcf841700004, timeout of 10000ms exceeded" | 00:18 |
| SpamapS | as long as the address doesn't change, you should be able to reboot node 0 | 00:20 |
| SpamapS | agents should poll/reconnect eventually | 00:20 |
| melmoth | thanks. | 00:20 |
| smoser | roaksoax, around ? | 00:21 |
| === torment is now known as Discharge | ||
| === Discharge is now known as torment | ||
| smoser | jtv, https://code.launchpad.net/~smoser/maas/packaging.lp1049177/+merge/124083 | 00:31 |
| smoser | roaksoax, ^ for you too. | 00:34 |
| === matsubara-afk is now known as matsubara | ||
| bigjools | smoser: I'll look | 00:38 |
| bigjools | smoser: do you need something to uninstall all that when the package is removed? | 00:43 |
| smoser | well ther ei sonly the one thing installed, and its will be removed just as any other file | 00:45 |
| smoser | the one thing ... /etc/apparmor.d/dhcpd.d | 00:45 |
| bigjools | smoser: ok - I didn't know if it was automatically removed or not | 00:46 |
| bigjools | smoser: can you do a packaging.precise merge as well please | 00:46 |
| smoser | yeah, its just a file. | 00:46 |
| smoser | ackaging.precise will depend on sru of the isc-dhcp fix also. | 00:47 |
| bigjools | yeah | 00:47 |
| smoser | and i'd like to wait until this at least is baked a bit before uploading that. | 00:47 |
| bigjools | yup | 00:47 |
| smoser | tomorrow i'll try to do a pakaging merge for precise though and just have the mp ready | 00:48 |
| bigjools | cool | 00:49 |
| jtv | Thanks smoser! | 02:19 |
| bigjools | hello jtv | 02:23 |
| jtv | Hi | 02:23 |
| jtv | bigjools: saw that email where DEFAULT_MAAS_URL turns out to tie into the UI somehow. Damnation. | 02:30 |
| bigjools | jtv: yes - I asked diogo exactly how it breaks | 02:30 |
| bigjools | damnation indeed | 02:30 |
| jtv | No idea where that happens. | 02:30 |
| bigjools | back to the drawing board for you :) | 02:30 |
| matsubara | bigjools, jtv: if I dpkg-reconfigure the package and choose 192.168.21.1 (the internal interface where pxe /provisioning runs) then the web server stops listening on 10.98.0.13 | 02:31 |
| jtv | Hi there matsubara | 02:31 |
| matsubara | jtv, hey | 02:31 |
| bigjools | hey matsubara | 02:32 |
| jtv | Hmmm... this sounds as if the connection between UI and the setting is made in the packaging somewhere. | 02:32 |
| bigjools | that's a bizarre thing for the web server to do | 02:32 |
| bigjools | sounds like packaging | 02:32 |
| jtv | Yup. | 02:32 |
| jtv | It should definitely be listening on both interfaces. | 02:32 |
| bigjools | not sure why we need to tell apache where to listen, all interfaces seems more reasonable to me | 02:32 |
| jtv | Because it must service both you (the human) and the workers/nodes in the MAAS. | 02:32 |
| bigjools | yeah | 02:32 |
| matsubara | bigjools, jtv: when I run dpkg-reconfigure maas, it stops the apache server but doesn't restart it again | 02:34 |
| bigjools | matsubara: haha! | 02:34 |
| jtv | That does kind of explain... | 02:35 |
| matsubara | in any case it looks like a bug, reconfigure should restart apache, shouldn't it? | 02:35 |
| bigjools | yes | 02:35 |
| bigjools | not sure why it needs to restart it at all | 02:35 |
| matsubara | it stops it for some reason :-) | 02:36 |
| bigjools | if you restart it, does the UIU and everything else work? | 02:36 |
| bigjools | UI* | 02:36 |
| jtv | I wonder where it stops apache... it seems to be only the postinst and such that mess with apache. | 02:37 |
| matsubara | yes, restarting it seems to bring the UI back up | 02:38 |
| roaksoax | matsubara: there's a bug where reconfigure does not restart apache and I'll deal with that | 02:38 |
| jtv | \o/ | 02:39 |
| bigjools | woo | 02:39 |
| * bigjools heads out to lunch | 02:39 | |
| jtv | roaksoax: out of interest, where does it stop apache? I would have expected to see that in maas.config. | 02:39 |
| jtv | (Because I'm not a packaging expert) | 02:39 |
| roaksoax | jtv: we don't stop apache2 | 02:40 |
| roaksoax | jtv: maybe it is crashing for some reason? the restart stops and start the daemon agian | 02:40 |
| roaksoax | maybe that is it? | 02:40 |
| jtv | Ohhhh | 02:40 |
| roaksoax | jtv: no error logs? | 02:40 |
| roaksoax | apache2 error logs that is | 02:40 |
| jtv | Ask matsubara :) | 02:40 |
| roaksoax | matsubara: http://paste.ubuntu.com/1201757/ | 02:40 |
| roaksoax | matsubara: try that | 02:40 |
| jtv | I guess it's just a way to restart the wsgi processes..? | 02:41 |
| roaksoax | matsubara: can you, however, directly point me to bugs or issues you find with packaging otherwise I'll lots in bugmail | 02:41 |
| roaksoax | jtv: yes | 02:41 |
| roaksoax | jtv: so dpkg-reconfigure maas, change IP should restart apache2 to pick the new IP and things should be just fine | 02:42 |
| roaksoax | matsubara: ^^ | 02:42 |
| jtv | Wait... what new IP? | 02:42 |
| jtv | Apache should be listening on all interfaces. | 02:42 |
| jtv | MAAS must be aware of the new address, but not apache, right? | 02:42 |
| roaksoax | jtv: yes, it is for maas to pick up the new IP | 02:43 |
| roaksoax | and stuff | 02:43 |
| matsubara | roaksoax, I'm looking at the apache logs. let me paste it for you | 02:43 |
| roaksoax | jtv: the dpkg-reconfigure maas sets a new ip in DEFAULT_MAAS_URL | 02:44 |
| roaksoax | s/in/for | 02:44 |
| matsubara | roaksoax, https://pastebin.canonical.com/74380/ | 02:44 |
| roaksoax | matsubara: do you have a node deployed where I can test? | 02:45 |
| roaksoax | matsubara: err log in and test? | 02:45 |
| roaksoax | that's maybe upstream error | 02:45 |
| matsubara | roaksoax, the maas server is running but there's no node deployed | 02:45 |
| roaksoax | matsubara: yeah can you tell me your steps to reproduce the issue you encountereD? | 02:46 |
| matsubara | roaksoax, sudo dpkg-reconfigure maas | 02:46 |
| matsubara | change the ip | 02:46 |
| matsubara | the apache server is stopped but not restarted | 02:46 |
| matsubara | feel free to log in the Lenovo lab | 02:47 |
| matsubara | https://wiki.canonical.com/Launchpad/QA/MAASLenovoLab | 02:47 |
| roaksoax | matsubara: yeah that's a bug | 02:47 |
| roaksoax | matsubara: there's a restart_apache2 missing on dpkg-reconfigure | 02:48 |
| roaksoax | matsubara: http://paste.ubuntu.com/1201757/ | 02:48 |
| matsubara | I see | 02:49 |
| roaksoax | matsubara: try now | 02:50 |
| matsubara | is it a known bug? | 02:50 |
| roaksoax | matsubara: yes it is a known bug, I will fix it tomorrow | 02:50 |
| roaksoax | matsubara: i was concentrated on getting the releases support into maas | 02:50 |
| roaksoax | matsubara: pastebin /etc/maas/maas_local_settings.py after the dpkg-reconfigure | 02:51 |
| matsubara | roaksoax, yes, now it restarted fine | 02:51 |
| roaksoax | matsubara: alright cool then | 02:52 |
| matsubara | roaksoax, do you still need that pastebin? | 02:52 |
| roaksoax | bigjools: btw.. if you can take a look at https://code.launchpad.net/~andreserl/maas/add_ubuntu_releases_lp1013146/+merge/124092 and tell me how is it looking, it would be great | 02:52 |
| roaksoax | matsubara: nah! | 02:52 |
| matsubara | cool. thanks | 02:53 |
| roaksoax | matsubara: i'm sure the above fix has fixed things | 02:53 |
| roaksoax | jtv: so can we ship our own maas-dhcp upstart job then? | 02:57 |
| smoser | roaksoax, fwiw, there are python bindings to distro-info | 02:57 |
| jtv | And hi smoser | 02:57 |
| smoser | roaksoax, yeah you can do that. | 02:57 |
| smoser | (i'm not really here) | 02:57 |
| jtv | That explains. | 02:57 |
| roaksoax | smoser: python-bindings as in import DistroINfo etc etc or similar? | 02:57 |
| smoser | right | 02:58 |
| jtv | roaksoax: I have a handy example of a modified upstart job if you want it. | 02:58 |
| roaksoax | smoser: ah cool, willk use that intstead then | 02:58 |
| roaksoax | jtv: sure, i wont merge it tonight though as I'm in zombie mode | 02:58 |
| roaksoax | but will do tomorrow | 02:58 |
| roaksoax | i'd like to get this fixed tomorrow | 02:58 |
| jtv | @#$% timezones | 02:58 |
| jtv | I'd be most grateful. Hang on, I'll paste my upstart script. | 02:58 |
| smoser | roaksoax, http://paste.ubuntu.com/1201799/ | 02:59 |
| smoser | thats a patch to cobbler that i used to use it there. | 02:59 |
| smoser | might just get you started | 02:59 |
| smoser | but its very simple | 02:59 |
| roaksoax | smoser: nice, thanks! | 02:59 |
| jtv | roaksoax: I named it maas-dhcp-server — http://paste.ubuntu.com/1201804/ | 03:00 |
| roaksoax | jtv: awesome, will merge it tomorrow and test | 03:00 |
| jtv | Thanks! We'll need to make corresponding changes to trunk as well, as we need to restart dhcpd. | 03:00 |
| roaksoax | jtv: you can go ahead and make then later in your day today, so that tomorrow morning I can make the ones in packaging | 03:01 |
| roaksoax | and we should be good | 03:01 |
| jtv | \o/ | 03:01 |
| jtv | Thanks again. | 03:01 |
| smoser | jtv, that looks reasonable. | 03:02 |
| smoser | did you see my note that we will need packaging of that? | 03:02 |
| smoser | as right now maas-dhcp just has 1 file | 03:02 |
| roaksoax | smoser: i'll take care of that tomorrow morning :) | 03:03 |
| jtv | smoser: If I'm thinking of the same note, then yes. But that was quite some time ago now, so maybe there's something I'm missing. | 03:03 |
| smoser | jtv, so how are you going to configure $INTERFACES ? | 03:03 |
| jtv | /etc/default/maas-dhcp-server | 03:03 |
| jtv | Which also needs packaging. | 03:03 |
| smoser | jtv, no, my merge proposal had comment about it. | 03:04 |
| jtv | Oh | 03:04 |
| smoser | so you cant really do that | 03:04 |
| jtv | Didn't see that. | 03:04 |
| jtv | Why not? | 03:04 |
| smoser | hm.. | 03:04 |
| smoser | so /etc/default/maas-dhcp-server is going to be a conf file | 03:04 |
| smoser | and if you edit it | 03:04 |
| smoser | and then the package upgrade | 03:04 |
| smoser | it will force a prompt | 03:05 |
| smoser | er... if the package'd file changes and there were local file changes. | 03:05 |
| roaksoax | smoser: we need to handle that automatically i'm afraid | 03:05 |
| jtv | Then... can we put it somewhere else? | 03:05 |
| smoser | roaksoax, right. so /etc/default/maas-dhcp-server is not the right place. | 03:05 |
| smoser | right. | 03:05 |
| jtv | /var/lib/maas/dhcp-interfaces? | 03:05 |
| smoser | for the things that you're configuring there, what i would suggest is some config not in /etc/ | 03:05 |
| roaksoax | smoser: i already have an idea of how to handle that automatically | 03:05 |
| smoser | that maas managers | 03:05 |
| smoser | but what you *could* do is have /etc/default/maas-dhcp invoke 'maas-shell get-my-dhcp-interfaces' | 03:06 |
| smoser | or somethign like that. | 03:06 |
| smoser | but having maas edit that file i think is not right. | 03:06 |
| jtv | That's getting a little roundabout, given that all we want here is to set a variable. | 03:07 |
| jtv | Why not just... | 03:07 |
| jtv | if [ -f /var/lib/maas/dhcp-interfaces ]; then | 03:07 |
| smoser | you can do that. but i'd not bother with 'dhcp-interfaces'. | 03:07 |
| jtv | INTERFACES=`cat /var/lib/maas/dhcp-interfaces` | 03:07 |
| jtv | etc? | 03:07 |
| smoser | but rather a shell sourceable file there. | 03:08 |
| smoser | and then you can set other things in it there also | 03:08 |
| smoser | rather than needing multiple settings potentially. | 03:08 |
| jtv | Well then we can do one better: just add a management command that prints out the interfaces. | 03:08 |
| smoser | thats what my suggestion above was | 03:08 |
| smoser | (/etc/default/maas-dhcp invoke 'maas-shell get-my-dhcp-interfaces') | 03:08 |
| jtv | Ah, then that'd be "maas" not "maas-shell," right? | 03:08 |
| smoser | sure. whatever it was. | 03:08 |
| smoser | or should be | 03:08 |
| smoser | but yes, ask maas what the value is | 03:09 |
| smoser | as it is the thing that knows | 03:09 |
| roaksoax | smoser: that's a cool idea | 03:09 |
| jtv | That's fine, yes. | 03:09 |
| smoser | anyway | 03:09 |
| smoser | bed time | 03:09 |
| smoser | good night. | 03:09 |
| jtv | nn | 03:10 |
| bigjools | so why is apache only listening on the one interface? | 03:12 |
| roaksoax | it is not, MAAS has as DEFAULT_MAAS_URL the IP of one interface | 03:13 |
| roaksoax | as a best effort attempt | 03:13 |
| roaksoax | to automatically determine what interface to be used | 03:14 |
| bigjools | ok | 03:14 |
| bigjools | to be used by what? | 03:14 |
| roaksoax | s/interface/address | 03:14 |
| roaksoax | to be used by MAAS as DEFAULT_MAAS_URL | 03:14 |
| roaksoax | in packaging that is | 03:14 |
| bigjools | I think we're confused about what it means | 03:14 |
| bigjools | as far as the code is concerned, it's the internal-facing URL for the nodes to contact for API | 03:15 |
| roaksoax | bigjools: yep | 03:15 |
| roaksoax | bigjools: i know that :) | 03:15 |
| bigjools | ok :) | 03:15 |
| bigjools | so we need to prompt for it on installation | 03:16 |
| bigjools | it only does it on reconfigure | 03:16 |
| bigjools | or is that deliberate? | 03:16 |
| roaksoax | bigjools: that's deliberate. We cannot, should not prompt for it on installation | 03:16 |
| bigjools | in any case it picks the wrong ip/interface in the qa lab | 03:16 |
| bigjools | ok | 03:17 |
| roaksoax | bigjools: that's why we display a message saying that if it is the wroing address it should be changed | 03:17 |
| bigjools | gotcha | 03:17 |
| bigjools | it's coming back to me now | 03:17 |
| roaksoax | bigjools: if we always proimpt for it, it will break MAAS installation on the CD | 03:17 |
| roaksoax | s/on/from | 03:17 |
| bigjools | yeah | 03:18 |
| roaksoax | alright | 03:19 |
| * roaksoax is off to bed | 03:20 | |
| roaksoax | goognight all | 03:20 |
| bigjools | nn | 03:21 |
| === matsubara is now known as matsubara-afk | ||
| jam | jelmer, mgz: https://plus.google.com/hangouts/_/6b0c81dbfeb8a772657f16370d380d1e26c9a4d3?authuser=0&hl=en-GB | 08:45 |
| === dpb_ is now known as Guest14449 | ||
| Daviey | roaksoax: not urgent today.. but can you explore using freeipmi-tools as a config option in maas upstream? | 12:32 |
| roaksoax | Daviey you mean instead of ipmitool? | 12:34 |
| Daviey | roaksoax: right | 12:37 |
| mgz | hm, I can't make after updating trunk because buildout is timing out trying to get repl | 13:09 |
| mgz | where do I need to remove it from to skip over that? it's not in versions.cfg and removing from buildout.cfg didn't help | 13:11 |
| mgz | ah, need to edit the makefile | 13:12 |
| smoser | roaksoax, do you happen to know why maas uses 'python-twisted' as a dependency | 13:34 |
| smoser | doku was asking me if that was necessary, versus a more fine grained list of what it actually uses. | 13:34 |
| roaksoax | smoser pserv and txlongpoll need it | 13:39 |
| smoser | but they dont need the blanket package | 13:39 |
| smoser | they need specific things it depends on | 13:40 |
| roaksoax | smoser why is that is it still uninstallable? | 13:40 |
| smoser | are you asking whether or not it is uninstallable? or asking *why* it is | 13:41 |
| smoser | i think doku probably fixed it, or if not is on it | 13:41 |
| smoser | and it will be fixed | 13:41 |
| smoser | but specifying a meta package when you dont need it is just wasteful | 13:41 |
| smoser | (not a big deal, just we probably get things we dont need) | 13:41 |
| roaksoax | python-twisted is a met package? | 13:42 |
| === hazmat is now known as 45PAA1GWO | ||
| roaksoax | smoser smoser: http://bazaar.launchpad.net/~maas-maintainers/maas/trunk/view/head:/required-packages/base | 13:55 |
| === matsubara_ is now known as matsubara | ||
| smoser | roaksoax, http://paste.ubuntu.com/1202679/ | 13:57 |
| smoser | it sure looks as "mostly meta" | 13:57 |
| smoser | http://paste.ubuntu.com/1202680/ | 13:58 |
| smoser | roaksoax, its low priority | 13:58 |
| smoser | but doku asked me | 13:58 |
| smoser | as we wouldn't have seen the uninstallable if we didn't depend on that package :) | 13:58 |
| smoser | rvba, around ? | 14:06 |
| smoser | {{kernel_params(arch="amd64") | kernel_command}} | 14:06 |
| smoser | what does that do ? in the templates | 14:07 |
| smoser | is it "call kernel_params *or* use kernel_command ?) | 14:07 |
| smoser | " | 14:07 |
| === wgrant_ is now known as wgrant | ||
| === marcoceppi is now known as marcopollo | ||
| roaksoax | smoser: hold on | 15:23 |
| roaksoax | smoser: why was python-twisted uninstallable in the first place? | 15:24 |
| roaksoax | smoser: cuase the reason why maas was uninstallable in your case was becuase a dependecy was uninstallable, not becuase it depended on a meta-package | 15:24 |
| smoser | right. | 15:25 |
| smoser | you are correct, roaksoax . python-twisted should never have become uninstallable (that was a valid issue) and prevented maas from being installed. | 15:25 |
| smoser | but it just raised the question of "why were you depending on that meta package" rather than the specific packages you need | 15:25 |
| roaksoax | smoser: now, we need to find out what dependencies of python-twisted are needed by maas | 15:26 |
| roaksoax | allenap: ^^ | 15:26 |
| roaksoax | smoser: well upstream mentioned that as a dependes, maybe they just need python-twisted-core, or maybe just need the whole of python-twisted | 15:28 |
| roaksoax | smoser: as pserv runs a twisted daemon | 15:29 |
| * allenap looks | 15:31 | |
| allenap | smoser, roaksoax: python-twisted-{core,web} is all I think for direct deps. | 15:32 |
| roaksoax | allenap: ok cool, i think we can update that then | 15:37 |
| allenap | Cool. | 15:38 |
| allenap | mgz: bzrlib.osutils.get_user_encoding() reports "ascii", but my args do seem to be encoded as utf-8. Would you happen to know if this is something my terminal has done, when I pasted in from charmap? Or a bug in get_user_encoding()? | 16:03 |
| mgz | it's based on your C locale setting | 16:04 |
| allenap | mgz: I'm tempted to use sys.getfilesystemencoding() but I'm not sure that's a good idea. | 16:04 |
| allenap | mgz: en_GB.UTF-8 | 16:04 |
| allenap | LC_ALL | 16:04 |
| mgz | okay, you probably are using bzrlib from a script but aren't calling `locale.setlocale(locale.LC_ALL, '')` | 16:06 |
| allenap | mgz: Ah ha, such are the runes needed. Thanks. | 16:07 |
| === matsubara is now known as matsubara-lunch | ||
| guimaluf | anyone can tell me what can be generating this error in my node syslog? Sep 12 17:34:26 150.164.3.160 main-menu[582]: (process:2659): wget: server returned error: HTTP/1.1 500 Internal Server Error | 16:34 |
| guimaluf | Sep 12 17:34:26 150.164.3.160 main-menu[582]: WARNING **: Configuring 'network-preseed' failed with error code 1 | 16:34 |
| guimaluf | Sep 12 17:34:26 150.164.3.160 main-menu[582]: WARNING **: Menu item 'network-preseed' failed. | 16:34 |
| mgz | guimaluf: are you filing bugs about the issues you run into? | 16:41 |
| mgz | everyone: there's a really basic "what is maas" question on the openstack mailing list | 16:41 |
| mgz | I'd reply if I knew how to give a pithy answer. | 16:42 |
| guimaluf | mgz, no i'm not. I don't understand the cause of this issue. It happened out of nothing. :/ | 16:43 |
| mgz | guimaluf: if no one is around to answer on irc, it's good to follow up with either a bug report, or a message to the mailing list with more details | 16:45 |
| mgz | people are in different timezones, there's no guarentee someone who can answer your questions will be on at the same time as you | 16:46 |
| guimaluf | I know :/ | 16:47 |
| guimaluf | I'm trying to get information about my issue, but I dont know where to go :/ | 16:47 |
| mgz | so, hm, the launchpad MaaS page also suggests using askubuntu with the tag maas | 16:51 |
| mgz | but when you've got a bug, I'd just file against maas on launchpad | 16:51 |
| === matsubara-lunch is now known as matsubara | ||
| guimaluf | mgz, I found out the issue. the char 'é' inside comments on maas.pressed :/ | 18:06 |
| melmoth | what does it mean when "juju deploy" on a maas server ends up with the new machine kept in "pending state" ? | 19:49 |
| melmoth | on a working install, as soon as i deploy something, i see a new machine id, with info about it | 19:49 |
| melmoth | but here, on the real cloud i try to install, the machine is still in pending, and no new machine are set into "allocated to root" on the web page | 19:50 |
| guimaluf | melmoth, have you check the ssh-keys? in my case I'd give up to use maas + juju + openstack. | 20:22 |
| guimaluf | cause of this. all my servers kept in pending state. | 20:22 |
| === 45PAA1GWO is now known as hazmat | ||
| melmoth | yep, not the ssh key.. turned out zookeeper was having problem, did not investigate , we are on our way to a new bootstrap | 20:22 |
| guimaluf | melmoth, exactly! zookeeper problem! | 20:25 |
| guimaluf | anyone know how can I set up a smarter partition scheme when deploying a node? I've tried late-command, but isn't working. :/ | 20:29 |
| melmoth | guimaluf, http://pastebin.com/gtxGJgt3 | 20:34 |
| melmoth | this is a preseed i used on a 1 disk machine where i wanted to have a nova-volumes vg separate from the vg used by the system done at installation stage | 20:35 |
| melmoth | i guess doing something similar in the preseed file used by maas should work (but you dont have to bother creating the nova-volumes vg, the nova-volumes charms takes care of it) | 20:36 |
| guimaluf | melmoth, partman can only handle with one disk. I want to make sd[b-z] also avaible | 20:36 |
| melmoth | havent try with several disk, my main problem was that i had only 1 anyway :) | 20:36 |
| guimaluf | I had, and partman dont deal with more then one disk. also none of my recipes had been aplied | 20:38 |
| melmoth | grumble same stuff again | 20:55 |
| melmoth | ahhh, note for later: dont put 127.0.0.1 for the maas server in the juju environment.yaml :) | 21:05 |
| melmoth | things seems better now | 21:05 |
| === cinerama_ is now known as cinerama | ||
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!