=== kev is now known as Guest24274 === marco-traveling is now known as marcoceppi === CyberJacob is now known as CyberJacob|Away === kevin is now known as Guest44429 [07:26] bigjools: let me just look up the change that's needed... [07:26] jtv1: bigjools: Did you guys manage to run the script end-to-end? [07:26] jtv1: thank you sir [07:26] rvba: I'm running it now. [07:27] rvba: I think jtv did, I have been out all day so only just getting up to speed [07:27] It's actually imported an image already. [07:27] It takes a while, obviously. === jtv1 is now known as jtv === CyberJacob|Away is now known as CyberJacob [07:27] jtv: are you using the int-fix-boot-imgs patch? [07:28] No, not yet. [07:28] Wanted to see the "before" state first. [07:28] Okay. [07:28] bigjools: edit mirror_info_for_path [07:28] policy = policy_read_signed [07:28] becomes: [07:28] policy = lambda content, *args, **kwargs: content [07:28] thanks [07:28] That'll teach it. [07:29] now I get ValueError: No JSON object could be decoded [07:29] Hwrg [07:30] Source URL is http://maas.ubuntu.com/images/ephemeral-v2/daily/streams/v1/index.json ? [07:30] I should get used to calling it "source path" really. [07:30] - path: "http://maas.ubuntu.com/images/ephemeral-v2/daily/streams/v1/index.json" [07:31] release "trusty", arch "i386", subarches "generic"? [07:31] well it has * * * at the moment [07:31] All as strings? [07:32] Because we originally had subarches: ['*'] in the example, but it should be a string. [07:32] Double-quoted, I think. [07:32] (Not that it's likely to matter) [07:32] subarches is a list. [07:32] all strings [07:32] doesn't make any different [07:33] I am running straight from the integration branch here, you must have other changes [07:34] Nope. But I did have that error earlier... I wonder if it's just its way of saying "I didn't find anything matching your requirements." [07:34] Try setting arch/subarches/release. [07:35] rvba: did subarches become a list? In the integration branch as I have it (pulled earlier today) it was still a comma-separated string. [07:35] ok how are you running the script? [07:35] I use bin/py scripts/maas-import-pxe-files [07:36] jtv: AFAIK, it's been a list since the beginning of the week… let me check… [07:36] it's currently not a list [07:36] bigjools: ah-HAH. That'll import the globally installed maas modules, won't it? [07:36] Try running sudo PYTHONPATH=./src scripts/maas-import-pxe-files [07:37] rvba: no, the pserv.yaml in the branch used to say "[*]" but that didn't actually work. [07:37] Him, something is wrong then. [07:37] subarches = Set(if_missing=['*']) [07:37] That's how it's defined. [07:37] s/Him/Hum/ [07:38] When you pasted examples for Julian yesterday, you had subarches as a single comma-separated string. [07:38] jtv: I don't have globally installed modules :) [07:42] rvba, jtv: ok so I am not sure what else I need to do here to get the script working [07:42] * bigjools dist-upgrades [07:42] Careful! [07:43] Remember the old and new pserv.yaml are mutually incompatible. [07:43] If you have an old pserv.yaml with a new codebase, or vice versa, the upgrade hangs. [07:43] jtv: I am not using any packages [07:43] OK [07:43] this is completely from branch [07:43] on a fresh canonistack instance [07:45] Not promising it'll help me understand, but do you have a traceback? [07:45] jtv: http://paste.ubuntu.com/7129350/ [07:45] thx [07:45] looks like it barfs on the index [07:45] Most likely, yes. [07:45] bigjools: I'm running the script now… [07:46] For me it completed the download, then broke because it didn't have the int-fix-boot-imgs branch yet. [07:46] Which I'm trying now. [07:46] bigjools: that looks like a failure I had earlier. [07:46] And I'm pretty sure I fixed it by changing pserv.yaml. [07:46] I am going to have to break for a bit shortly, I am home alone with twins [07:46] there's only so much kids tv that will sate them [07:48] Script running okay thus far (here is my pserv.yaml http://paste.ubuntu.com/7129354/) [07:48] Okay, it failed in the end because I don't have the changes from int-fix-boot-imgs. [07:48] jtv: same as what you got. [07:48] bigjools: the "boot" part of my pserv.yaml is http://pastebin.ubuntu.com/7129356/ [07:49] Now going to try with the new branch, which includes some label stuff so it'll find a new reason to break. [07:51] Running the int-fix-boot-imgs branch (plus Scott's fixes which are only in trunk)… [07:52] uec2roottar running… [07:52] Script finished running without errors. [07:52] * rvba checks the output… [07:52] so I have a pristine int branch with rvba's pserv.yaml and I get the json decode error [07:53] And then... commissioning, deploying with fpi, and deploying without fpi! [07:53] bigjools: what about the policy patch? [07:53] I have that [07:53] that's it [07:53] you guys must have other changes [07:53] Did you try the pserv.yaml segment I pasted? [07:53] I used rvba's [07:54] who said his was working [07:54] Output looks sensible but I don't see the label in the images' path http://paste.ubuntu.com/7129368/ [07:54] it was the same as mine [07:54] I have similar results, basically what I expected. [07:54] Now to work the label in there, and make sure it's not "daily". [07:55] Well, it should be whatever is in simplestreams' data. [07:55] eventually yes but it's broken ATM [07:56] Actually, in the simplestreams data I'm seeing paths like precise/amd64/di/20101020ubuntu136.15/raring/generic/di-initrd [07:57] ...which are nothing at all like what we have. [08:05] bigjools: I remember the weird thing when I got that error was that the JSON was not empty. [08:05] jtv: the new list_boot_images() in int-fix-boot-imgs seems to work. Of course, right now it returns the empty list because the label is not yet part of the path. But I did manual test (adding the label as part of the images' path manually) and list_boot_images returned something sensible. [08:06] That's good. I'm having some more trouble getting mine to work. :( [08:07] Looks like I'm missing an update somewhere... simplestreams looks for /index.json/streams/v1/index.sjson [08:07] — so it's not doing the "URL to the index plesae" yet. [08:08] Merging update from latest integration branch. [08:10] Success! [08:15] With that diff (it includes a couple of changes from the int-fix-boot-imgs branch) I get the layout we want: http://paste.ubuntu.com/7129439/ [08:16] "this" [08:17] Do you get "release" and "alpha-2" as the labels? Or always "daily"? [08:18] * rvba checks [08:19] I downloaded only trusty/amd64/generic and for that image the label is daily. [08:19] Yay! [08:19] * rvba includes saucy, re-runs the script. [08:19] We'll need to hard-code that to "release" [08:20] But that's fine — the important thing is that the label is in the dir structure. [08:20] I don't understand… shouldn't we just expect Scott to fix that? [08:21] * jtv invites rvba to ponder that option for a bit [08:21] saucy→daily [08:22] Yeah. [08:22] If it gets fixed before release, then yes, we can just pretend "daily" is a workable label. But otherwise... [08:26] For some reason memory usage seems to have jumped. My little server is getting really slow. [08:31] Review needed: supporting both the old and the new config items, to avoid upgrade breakage. https://code.launchpad.net/~jtv/maas/support-new-pserv-config-items/+merge/212110 [08:36] jtv: why do you want to make that part of trunk MAAS now? Shouldn't we land this with the rest of the int. branch… if and when we do so? [08:37] It can complicate testing. I want to land it in both. [08:41] New diff with the tgt_entry() method fixed http://paste.ubuntu.com/7129525/ [08:42] err: http://paste.ubuntu.com/7129527/ [08:48] rvba: does that include my int-fix-boot-imgs branch by the way? [08:49] My version adds "label" in one more place. [08:49] Weird, this should include the changes you made in your int-fix-boot-imgs branch. [08:49] You say you see something missing? [08:52] jtv: btw, subarches should really be a list. It works with '*' only because a string happens to be an iterable. [08:52] Christ. [08:52] But then why does "generic" work? [08:54] It should work (as in, not crash) but not download anything. [08:54] * rvba tries [08:56] Hum, looks like the config does something funky: if I set 'subarches: "generic"' in pserv.yaml, I get 'filter['subarches']=['generic']' [08:56] If I use a list in the config, I get the same list in filter['subarches']. [08:57] I guess that's a feature. [08:59] Heaven protect us from features. [09:26] help, jtv, rvba. What are you two doing differently with your branches to mine such that I get that json error? [09:29] bigjools: I don't know really… can I get access to your canonistack instance? [09:30] rvba: one sec [10:00] jtv: when you're done with testing the diff I sent you, would you mind also getting the most recent changes from the int. branch into your int-fix-boot-imgs branch? This way, we will be able to build a package from it and test the package. [10:03] jtv: oh, looks like you did that already :) [10:03] :-P [10:10] Merging trunk has resolved the ### INVALID STRING ### stuff, but the region-worker Celery keeps spawning. [10:11] jtv: Are you integrating my diff or should I create another branch on top of yours? [10:12] rvba: I'd make it a separate branch. [10:13] jtv: all right, I'll do it. [10:13] Now, why does that diff not apply? [10:13] (I was briefly distracted by travel stuff) [10:16] jtv: this one will probably do: http://paste.ubuntu.com/7129865/ [10:18] jtv: https://code.launchpad.net/~rvb/maas/use-labels-in-import-script/+merge/212125 [10:18] looking... [10:18] "patch" is taking forever on my test machine. [10:18] The performance really dropped quite dramatically. [10:20] Arg, it’s python-librabbitmq. It causes a seqfault when starting region-worker. [10:20] segfault even. [10:20] Ouch. [10:21] allenap: didn;t we replace that package with another one?> [10:21] we did [10:21] it's not installed here [10:22] allenap: uninstall it and see what happens [10:25] Everything works :) [10:26] I have a branch to forbid that package: https://code.launchpad.net/~allenap/maas/forbidden-deps/+merge/212128 [10:26] reviewing [10:28] * bigjools AFK for 10m [10:55] jtv: +1 on https://code.launchpad.net/~jtv/maas/int-fix-boot-imgs/+merge/212082 [10:55] Thanks. [10:55] I'm already trying from a package, by the way. [10:55] jtv: https://code.launchpad.net/~rvb/maas/use-labels-in-import-script/+merge/21 is officially up for review [10:56] jtv: Same here :) [10:56] Not having much luck pxe-booting though. [10:56] I'm just getting the lab ready. [10:59] By the way, when landing into the integration branch, don't forget to mark the branches (as opposed to the MPs) as merged. [10:59] I'll review your branch. [10:59] Ta. [11:01] I have high hopes for the simplestreams cache... [11:02] I'm re-running the import script after wiping all of my boot-resources directory — *except* the cache. [11:02] Yup. Nice & fast! [11:05] jtv: where and what is caching? [11:05] bigjools: images get dowloaded into /var/lib/boot-resources/cache [11:05] ...and the rest is extracted from there. [11:07] rvba: are you getting the following error? [11:07] [2014-03-21 18:06:20,750: ERROR/MainProcess] Task provisioningserver.tasks.report_boot_images[27842890-96bb-4543-8263-1a8ef4dfdff4] raised unexpected: TypeError(' is not JSON serializable',) [11:07] I thought I'd fixed one of those, but... [11:07] Anyway, I'll write up a fix. [11:08] Guys, i just built maas packages with 'make package' and cluster-controller depends on 'shim-signed' package which is provided only for amd64. [11:08] fuck [11:09] uh-oh [11:09] jtv: ta, branch is landed. === strikov is now known as strikov-lunch [11:09] I hope that merely means that a build failed..? [11:09] strikov: make sure you pull the latest version of the integration branch, we just landed 2 branches. [11:19] allenap: thank you once more for fixing the pserv tests. Unless you were the one who broke them, in which case, I hate you. :) [11:19] jtv: Hehe, no I didn’t break them :) [11:19] jtv: didn't see that error… but I didn't get that far yet. I know where the problem is… but so do you :) [11:19] Yup. This is what happens when you play with itertools! [11:19] heh [11:20] allenap: your opinion required. Blake's branch seems amd64 only. [11:21] it uses shim-signed which is only on amd64 [11:21] bigjools: The UEFI one? It only works for amd64, indeed. [11:21] allenap: yeah, does it check for amd64 before using it? [11:21] it's in m-i-p-f [11:21] bigjools: Ah, I hope so. [11:21] * allenap checks [11:22] allenap: it doesn't [11:22] Dammit. [11:22] yup [11:22] allenap: I don;t know how to do an arch-specific dependency in packaging but I am going to learn quickly :) [11:23] bigjools: I’m losing you now. Why does this impact packaging? [11:23] allenap: we (where we is probably you at this point) should change mipf to check even though we're going to get rid of that script. [11:24] allenap: I added a dependency on shim-signed but that makes the package unbuildable on i386 [11:24] and it's an arch-indep package which only builds on i386 [11:24] s/i386/amd64/ ? [11:24] so the packaging should only depend on it if installing on amd64 [11:25] no i386 intentional [11:25] jtv: I'm landing a change to pserv.yaml to make subarches a list. [11:25] (that's a feature we shouldn't hide) [11:25] Very well. [11:25] Huh? This makes no sense to me. [11:25] allenap: the arch indep builders are all i386 [11:25] but the package now needs amd64 [11:25] even though it's technically arch indep [11:25] Why does the package need amd64? [11:26] because shim-signed is a binary package that only exists on amd64 [11:26] Can we change that? [11:26] I added it as a cluster controller dependency [11:26] It’s just a blob. [11:26] Afaik. [11:26] yeah I need to make it arch dependent [11:26] it is [11:27] rvba: pushed a fix to the json bug. [11:28] bigjools: Do you mean arch *in*depedent? Or arch dependent for building, so that binaries can be prepared for each arch? [11:28] jtv: ta. I see you've removed the call to chain()… without removing the import. [11:28] Ah. So I forgot to check for lint. :-) [11:28] jtv: I'll do it in the branch I'm working on now. [11:29] Ta. [11:29] Now, where is this boot images UI? [11:29] I want my list of boot images! === strikov-lunch is now known as strikov [11:30] allenap: I mean to make it so that it only depends on shim-signed if you are installing on amd64 [11:31] otherwise the dependency cannot be fulfilled [11:31] allenap, bigjools: I see this issue while trying to install package on armhf: https://pastebin.canonical.com/106925/ [11:31] strikov: can you use public paste [11:31] allenap: that paste is the result of this problem [11:31] thank you strikov [11:32] bigjools, allenap: sorry, here it is: http://pastebin.ubuntu.com/7130103/ [11:32] bigjools: That’s fixing the wrong thing, though I grant it’s probably the path of least resistance right now. There’s no reason why MAAS installed on i386 can’t netboot an amd64 system using UEFI. [11:33] allenap: very true [11:33] allenap: the path of least resistance would be to include that blob in MAAS' tree and remove the dependency, wouldn't it? [11:35] rvba: Ah, yes, though we may find that our sticky-out bits are at risk from the scythes of the security team. [11:35] Hum, good point. [11:36] blake_r’s original approach was to download from the archive. Perhaps we need to reconsider that? [11:37] he was downloading the source package [11:37] which is problematic, not everyone has sources enabled [11:38] Or if downloading directly, we'd lose built-in signature verification. [11:49] allenap: I think we need to revert blake's branch then [11:50] bigjools: Or fix it. [11:50] and fold its changes, with the shim-signed stuff, into the new script [11:50] well the new script is nearly done [11:50] let's not do the work twice [11:50] bigjools: There was a lot more work in Blake’s branch than just m-i-p-f. Do we want to revert all of it? [11:50] no not at all [11:51] just the bits that are going to break without shim-signed :) [11:52] bigjools: The rationale for reverting versus ploughing through is because it’s blocking QA? [11:52] allenap: it'll block everything, I think mipf will crash [11:53] we could comment out that one line in it [11:53] bigjools: Yeah, let’s hack round it to unblock things. [11:54] can't tell if you're serious or being sarcy... [11:56] bigjools: Serious :) Agreeing with you. [11:56] allenap: oh good :) [11:57] allenap: are you ok to do that while I unfcuk the packaging? [11:57] bigjools: I need to work around all references to the shim, right? [11:57] allenap: I think it's just one call you can comment out to the new func [11:57] see the diff to mipf [11:58] bigjools: So, don’t call update_uefi_boot_loader at all? [11:58] allenap: just don't call it at all for now to unblock and then fix later [11:59] so yes :) [11:59] allenap: also remove shim-signed from required-packages [11:59] Okeydoke. [12:00] bigjools: Is there a bug about this? [12:00] allenap: no [12:02] * allenap files one [12:04] ok packaging fix just went in [12:04] I have lots of tests to disable. [12:06] oh well :/ [12:08] bigjools: https://bugs.launchpad.net/maas/+bug/1295644 [12:08] Ubuntu bug 1295644 in MAAS "MAAS cannot be installed on non-amd64 system" [Critical,In progress] [12:11] thanks allenap [12:13] * bigjools sleeps [12:22] bigjools: Fwiw, by reusing your packaging branch every time, each revision gets tagged with the same bugs. See https://code.launchpad.net/~maas-maintainers/maas/packaging. [12:22] It also means I can’t link that branch to the bug. [12:26] rvba: Do you have time for a quick (critical) review fix? https://code.launchpad.net/~allenap/maas/disable-uefi-temporarily/+merge/212150 === cmagina-away is now known as cmagina [12:44] allenap: looking now [12:55] Could someone review my patch to allow us to use different keyrings for different sources: http://bazaar.launchpad.net/~strikov/maas/new-import-script-integration.keyring-per-each-source/revision/2155 Thanks [12:56] strikov: I'll have a look [13:02] strikov: looks good, but you need to fix the test_config.py tests (yes, we have unit-tests for this): ./bin/test.maas ./src/provisioningserver/tests/test_config.py [13:28] rvba: thanks for the review, does it look better: http://bazaar.launchpad.net/~strikov/maas/new-import-script-integration.keyring-per-each-source.2/revision/2155 [13:29] rvba: Ran 25 tests in 0.719s [13:29] OK [13:29] strikov: cool. Looks good. If testing is okay, land it. [13:30] strikov: maybe also update pserv.yaml (I know the value is the default, but it's a good example) [13:30] rvba: will do [13:35] rvba: Pushed with pserv.yaml change. Thanks. [15:08] allenap: time for a tiny review? https://code.launchpad.net/~rvb/maas/build-depe/+merge/212173 [15:09] rvba: Sure! [15:10] rvba: Oh man, the deps are not sorted. [15:10] allenap: ahhhhh! [15:11] allenap: fixed [15:12] rvba: If you have a few minutes, can you look at my rpc-ident-redux branch, https://code.launchpad.net/~allenap/maas/rpc-ident-redux/+merge/212070? You only actually need to review the interdiff, http://paste.ubuntu.com/7129526/, which is a lot shorter. [15:12] allenap: k === zchander is now known as zchander_at_home [16:00] rvba: Have you noticed shells taking a long time to exit when closing? I hit Ctrl-d in a screen “tab” and it can take 20-30 seconds for bash to exit. At least, I assume it’s bash taking its time. [16:00] It’s been doing it for a couple of weeks, maybe more. [16:02] allenap: no, it clearly doesn't do that for me. [16:02] hi allenap, any idea what this is? http://paste.ubuntu.com/7131290/ [16:03] i'm getting lots of them [16:03] and a clue is that my dhcp server seems to not be working [16:10] tych0: Mmm, was there a request path anywhere near that? [16:11] not in maas.log, might be in apache's logs, though, let me look [16:12] yeah, access.log is totally empty [16:12] error.log doesn't have any timestamps close to that one [16:14] tych0: Is the DHCP server down, not giving out addresses? Or is it that leases are not getting into MAAS? (Not sure what outward indication there would be of that…) [16:15] allenap: it appears to not even be giving out addresses, although i restarted maas-dhcp-server [16:16] allenap: is there a log for it somewhere? [16:18] tych0: The leases file in /var/lib/dhcp might have something useful. Otherwise look in /var/log/syslog. [16:21] yeah, /var/lib/dhcp/dhcpd.leases is empty [16:24] tych0: Are you happy to dig around, try to find out why it’s not running, or not populating the leases file at least? [16:31] allenap: yeah, that's what i'm doing now [16:31] this might be an lxc, thing, actually [16:40] ah, nope [16:40] lxc is asking for one [16:40] it just isn't getting anything back [16:58] ah [16:58] i had dnsmasq running === cmagina is now known as cmagina-away [17:17] tych0: Phew, it’s not another MAAS bug :) === cmagina-away is now known as cmagina === cmagina is now known as cmagina-away [17:56] hi! [18:00] i am trying to follow the newbie install tutorial and getting stuck.. any pointers? http://paste.ubuntu.com/7131844/ [18:01] i guess sudo maas-import-pxe-files works [18:01] but document says thats only for sub 1.3 [19:20] Ive just setup my first MAAS server and now I have an issue with PXE booting. I can successfuly get an IP, but nodes cannot find a pxelinux.cfg [19:20] does maas create this or do I have to do it manually? [20:07] seem tobe an access problem. in maas.log getting "SuspiciousOperation: Invalid HTTP_HOST header (you may need to set ALLOWED_HOSTS):" === cmagina-away is now known as cmagina [20:36] I am trying to do the same thing ;) [20:37] running maas server on a bridge, trying to expose the dhcp and dns from my maas server without trashing my entire network [20:37] and feed into a pxe boot [21:01] humm what is "Router Ip" [21:01] the upstream gateway of mine? [21:05] I set mine the same as the maas server [21:06] im trying to bridge as well [21:10] ok I got it to PXE boot, had to reinstall and start from scratch [21:11] followed these: http://frugnul.blogspot.com/2013/04/maas-not-tutorial.html === cmagina is now known as cmagina-away [21:27] hum [21:28] by default, the little lxcbr0 bridge that juju added to my local machine, is not exposed to the world [21:28] i think I need to add my eth0 to be behind it [21:29] so that it can be exposed [21:29] but beyond what I have done before [21:30] do you have physical machines, who are they just on the same machine? [21:40] right now I have 2 phical machines [21:40] 1) desktop I am on now. want it to keep connected to current network (192.168.78) as a client.. but also want it to host a new maas network (as 192.168.4.1) [21:40] 2) a laptop I want to boot off PXE, and grab the 192.168.4.1 broadcast [21:40] and get added to maas [21:41] have a maas server up [21:41] his IP is set to 192.168.4.1 [21:41] laptop cant even ping that though [21:41] so that network is local to my box [21:42] so its like a need a bridge in front of my bridge [21:42] to tie my real eth0 to both networks [21:42] argh, networking hard === zchander is now known as zchander_ === cmagina-away is now known as cmagina === cmagina is now known as cmagina-away === cmagina-away is now known as cmagina === cmagina is now known as cmagina-away