[07:11] <darkmanta> hi everybody, I'm new here, I'd like to pose some questions about maas. how to do it?
[07:12] <bigjools> ask away
[07:14] <darkmanta> thank you. I'm trying MAAS on Virtualbox virtual machines. One of the problems I have met is that ALL ephemeral .tgz look crc-corrupted. I've tried to download the images through maas service and wget but the result is always the same
[07:15] <bigjools> darkmanta: is this the latest maas in saucy by any chance?
[07:17] <bigjools> https://bugs.launchpad.net/maas/+bug/1240652
[07:21] <darkmanta> I've tried 12.04.3, 12.10, 13.04 versions (desktop and server versions) for the virtual maas server. the result is always the same. can't complete the 'import boot images' procedure from web interface and from maas-import-pxe
[07:21] <bigjools> log please
[07:31] <darkmanta> i'm not on the right PC so I can't send the logs. but if anybody tries to download and check the ephemeral image...
[07:38] <bigjools> darkmanta: we see this only on saucy - it's the bug above
[07:40] <darkmanta> ok thank you
[07:40] <darkmanta> at the moment I'm unable to try MAAS at all.
[07:44] <darkmanta> trying to start a client machine not by PXE but through a ubuntu boot cd, following the MAAS procedure menu, fails again. the procedure ends up with a machine reset.
[07:45] <darkmanta> bigjools thank you for helping me. do you know a configuration that works well?
[07:49] <darkmanta> at the moment it seems I've tried all possible combinations of ubuntu versions, server/desktop included...
[08:00] <darkmanta> anybody knows a MAAS configuration that simply works?
[08:04] <rvba> darkmanta: the only configuration with a known blocking problem is 13.10.  And a fix is coming up soon.  If you have problems with another versions, it's not normal, but we need the error log to help you as it's probably related to your particular setup.
[08:05] <gmb> rvba, allenap: Can you take another quick look at https://code.launchpad.net/~gmb/maas/fix-stray-newlines-bug-1187851/+merge/191611 please?
[08:06] <gmb> Now complete with isatty() goodness.
[08:09] <darkmanta> rvba: thank you. I've set up a 12.04.03 virtual machine following the procedure at http://maas.ubuntu.com/docs/install.html
[08:11] <darkmanta> the problems i've found are: maas-install-pxe-images always fails with 404 file not found
[08:12] <darkmanta> 2) client PC never receive PXE booting image
[08:13] <darkmanta> 3) client PC booting by CD procedure always fail with a system reboot after selecting the IP of the MAAS server
[08:13] <rvba> All of the problems seem related to the import script not running correctly.
[08:14] <rvba> This smells like bug 1212434 but this bug has been fixed.
[08:15] <rvba> So I'd like to know which version you're using exactly (apt-cache policy maas) and the error message you're getting when you run the import script.
[08:15] <allenap> gmb: You can avoid the buf.seek(0); buf.read() with buf.getvalue().
[08:16] <gmb> allenap: You remind me of Benji sometimes. And I mean that as a compliment.
[08:16] <allenap> gmb: Otherwise, looks grand :)
[08:16] <gmb> Thanks
[08:16] <allenap> gmb: I don't know benji well enough to know :)
[08:16] <darkmanta> yes, iv'e patched the import script by removing the non-AMD63/non-i386 images downloads so the script continues with ephemeral images. but after the download these images always fail due CRC-error
[08:17] <gmb> allenap: I learnt more about Python from getting him to review my code than I have anywhere else… Looks like you're a more than worthy successor :)
[08:17] <rvba> darkmanta: CRC error?  That, as far as we know, is only happening on Saucy.
[08:18] <rvba> darkmanta: if you're using Saucy as the host of the MAAS server and seeing a checksum error, then there is a workaround (which is nothing more than that fix that is coming up shortly).
[08:18] <darkmanta> try to download the ephemeral images by wget... you can't expand it because of crc-error
[08:19] <rvba> I just downloaded all the images on a test system and all went fine.
[08:20] <rvba> darkmanta: can you confirm you're seeing the error described on bug 1240652 ?
[08:23] <darkmanta> the problem i have found seems different. anyway i see there is a patch committed 12 hours ago. I will try again to see if it's reproducible
[08:24] <rvba> darkmanta: I just added a comment on that bug.  It shows how to apply the fix manually if you don't want to wait for the package to be published :).
[08:26] <darkmanta> rvba: thank you so much
[08:26] <rvba> Welcome.
[09:25] <baffle> Hi, does MAAS acutally support defining multiple DHCP ranges on one controller that are not directly connected? I.e. so that I can PXE boot servers on different L2 networks, that have an L3 router with dhcp/tftp/etc proxying?
[09:27] <rbasak> I think that could be made to work. I assume that you'd make everything route right on the MAAS server(s), and provided that you proxy everything, what support would MAAS need for that? Everything would be transparent to it, right?
[09:29] <baffle> rbasak: Yes, everything would be transparent. I have this running using TheForeman right now, and have various manual setups earlier.
[09:29] <baffle> rbasak: The only "obstacle" is that MAAS requires an interface. And also seem to only allow one subnet pr. interface. :)
[09:30] <rbasak> What makes you say that MAAS only allows one subnet per interface?
[09:30] <baffle> rbasak: Guessing. :)
[09:30] <baffle> rbasak: It looks like tht from the GUI at least.
[09:31] <baffle> rbasak: But I guess I could just be wrong.
[09:31] <rbasak> I would expect the MAAS server to be on an L3 network that can route to all the others, and for you to sort that routing out, together with the network configuration on the MAAS server itself.
[09:31] <rbasak> So the MAAS server would have a single IP, and everything would route to that.
[09:32] <rbasak> I don't see why the MAAS GUI would need to be any different to normal.
[09:32] <rbasak> Alternatively you could have separate region controllers but I'm not familiar with that side of MAAS.
[09:32] <baffle> rbasak: Currently in the GUI I need to specify an interface and the IP MAAS should have on this interface.
[09:33] <rbasak> I don't understand the details then, sorry. But I don't see it as a big hurdle unless I'm missing something.
[09:35] <baffle> rbasak: I think you understand it; Everything network-wise works as it should (dhcp-requests arrive at the IP for the MAAS server) but configuration-wise there doesn't seem to be any way to define subnets without also configuring the interface on the server.
[09:36] <baffle> ISC DHCP handles stuff perfectly, as long as it has proper configuration. :)
[09:36] <rvba> baffle: you can always config the dhcp server managed by maas manually.
[09:39] <baffle> rvba: Doesn't MAAS use omapi etc to control dhcp/dns?
[09:39] <rvba> baffle: it does, but this doesn't mean you cannot change the configuration manually.
[09:40] <rvba> baffle: well, you cannot configure it any way we want.  For instance MAAS relies on the dhcp information it has to figure out to which cluster a connecting node belongs.
[09:41] <rvba> s/way we want/way you want/
[09:41] <baffle> baffle: Yes, I would assume I would have to add the DHCP scopes into MAAS someway. I just can't really see where. :-/
[09:42] <rvba> Right, I think you've hit something that MAAS does not currently support.
[09:42] <baffle> I kindof assumed this would be a quite normal usecase, to be honest. :-) Is everyone using a dedicated cluster controller for every subnet/rack of servers?
[09:43] <rvba> Yeah, that's the current recommended way to do this.  But we have plans to extend this.
[09:45] <baffle> rvba: Hmm, should be fairly straight forward...
[09:45] <rvba> Yes indeed.
[09:49] <baffle> Instead of having an "Interfaces" section under "CLuster Controller" it should just say "Networks". And the "interface" part of a network definition (I.e. wich physical network interface + IP address to configure (should be network address if no interface "selected")) should be optional. And it would be nice to have an option to override the "DNS zone name" pr. Network, instead of having one master zone.
[09:49] <baffle> Would also be nice to actually query/browse the current leases on DHCP as well. :)
[09:50] <rvba> baffle: true… could you please put that in a bug?
[09:51] <baffle> Actually, an "advanced" field where you can specify extra options for the dhcp-scope would be nice too.. :)
[09:51] <baffle> Sure.
[09:51] <rvba> Ta.
[10:05] <baffle> https://bugs.launchpad.net/maas/+bug/1241473
[10:05] <baffle> Does this seem OK?
[10:08] <baffle> Hmm. I'm a bit at a loss now, I guess I'll just jump on Cobbler instead for now.. I use TheForeman currently, but it is also a bit lacking and has tight Puppet integration. Razor seems to be in a state of flux, so that probably can't be used as well. Ideas?
[10:12] <baffle> Hmm, why does maas use pxelinux instead of ipxe? Is there compatability issues?
[10:14] <baffle> For previous setups, I've always loved how ipxe supports things like http:// and variables... I.e. having it chainloading deployment configuration from something like http://deployserver.com/deploy?mac=$mac is always nice. :)
[10:19] <rbasak> ARM compatibility with ipxe would be harder, but I don't think that's the reason since pxelinux was used before then.
[10:28] <baffle> rbasak: That's true. I see there is no ARM port for iPXE yet.
[10:29] <baffle> So I guess the current model is to dynamically create pxelinux.0/<macaddress> files?
[10:54] <rbasak> Right
[20:28] <Nik_> marcoceppi: remember when I was saying that LSHW was out of sync with maas so tags wouldn't apply... I found out why - We added disks to the machines and the machines needed to be recommissioned so that lshw output would be picked up for purposes of tagging.
[20:29] <Nik_> This could technically be a bug or an enhancement that can be submitted. Not sure how to check...
[20:29] <Nik_> http://askubuntu.com/questions/353814/maas-out-of-sync-with-lshw ^
[20:37] <fizzi_> Hi all
[20:37] <fizzi_> a question ...
[20:37] <fizzi_> is possible to install maas + nodes on vmware esx 4/5 ?
[20:39] <Nik_> Not an expert, but I don't see why not
[20:40] <Nik_> I think there's only a technicality regarding pxe booting a VM
[20:40] <Nik_> so I see why there may be issues
[20:40] <fizzi_> the pxe works fine
[20:40] <fizzi_> but when i start machine from maas
[20:40] <Nik_> Maas assumes IPMI also
[20:41] <fizzi_> the machine dont start
[20:41] <Nik_> for bare metal
[20:41] <Nik_> Yeah, it uses IPMI to boot up metal
[20:41] <Nik_> Hmm I guess I dunno
[20:44] <marcoceppi> Nik_: fizzi_ there is "vMAAS" (Virtual MAAS) that might help
[20:44] <marcoceppi> Nik_: thanks for the information, probably an enhancement, but open a bug against maas anyways