| === CyberJacob is now known as CyberJacob|Away | ||
| === CyberJacob|Away is now known as CyberJacob | ||
| === freeflying is now known as freeflying_away | ||
| === CyberJacob is now known as CyberJacob|Away | ||
| === freeflying_away is now known as freeflying | ||
| jtv | bigjools: I'm thinking to address the issue by accepting zero as a vlan tag, but normalising it to None. | 03:13 |
|---|---|---|
| bigjools | jtv: evil, but.... Django | 03:14 |
| jtv | In a word. | 03:14 |
| jtv | bigjools: by the way, have you seen those over-frequent DHCP config rewrites (and dhcpd restarts) lately? Or did they go away? | 03:38 |
| bigjools | jtv: I have not checked, I can do so in a bit | 03:39 |
| jtv | I'd be grateful if you could. | 03:39 |
| bigjools | will build a new package for my hp | 03:39 |
| jtv | And by the way, if you have time for a review... https://code.launchpad.net/~jtv/maas/zero-vlan-tag-is-not-unique/+merge/205073 | 03:41 |
| bigjools | jtv: I don't suppose I can persuade you to go and edit the last migration that set this up? | 03:57 |
| jtv | You can, if you want to. | 03:58 |
| bigjools | Having multiple migrations for the same new model is ugly as hell | 03:58 |
| jtv | I went with the safer option... If you want me to play dangerous with the schema, sure. :) | 03:58 |
| bigjools | meh | 04:01 |
| jtv | bigjools: updated. | 04:10 |
| bigjools | jtv: cheers | 04:16 |
| jtv | bigjools: one thing I didn't get around to bringing up... vlan tags: hex or dec? | 04:47 |
| bigjools | jtv: either I guess, prefix with 0x if you want hex | 04:52 |
| jtv | If you want to get fancy, I say: separate card! | 04:57 |
| bigjools | jtv: your signal fix has not worked | 05:53 |
| bigjools | still getting lease write jobs after probes | 05:53 |
| jtv | Arrrrrgh. | 06:04 |
| jtv | O! | 06:04 |
| jtv | Wait... | 06:04 |
| jtv | No. For a moment I was hopeful that I'd simply forgotten to remove something. | 06:09 |
| jtv | bigjools: can you verify that you're running a version where src/maasserver/dhcp_connect.py connects dhcp_post_change_NodeGroupInterface to a _field change_ on NodeGroupInterface, not to a save? | 06:10 |
| bigjools | jtv: I was running the daily build | 06:11 |
| bigjools | 1.4+bzr1853+dfsg+1891+231~ppa0~ubuntu14.04.1 | 06:11 |
| bigjools | so r1891 | 06:11 |
| bigjools | I powered down the machine now so can't check code easily | 06:11 |
| jtv | From trunk, right, not 1.4? | 06:11 |
| bigjools | yup | 06:11 |
| jtv | Ohdearohdearohdear | 06:12 |
| jtv | Hmm... was there really never a test for "does nothing if fields don't actually change"? | 06:13 |
| * jtv writes one | 06:13 | |
| bigjools | :) | 06:13 |
| jtv | Nope, nope, that passes. | 06:15 |
| jtv | So something else is triggering that signal. | 06:15 |
| * jtv throws some more hate at this use of signals | 06:15 | |
| jtv | bigjools: I do see that leases are uploaded every minute as per the celerybeat schedule... | 06:25 |
| bigjools | jtv: yes and then it writes dhcp and restarts server | 06:26 |
| jtv | Could you try it with a longer unconditional-dhcp-lease-upload schedule time? | 06:26 |
| bigjools | can't today will come back tomorrow | 06:26 |
| jtv | OK. | 06:26 |
| jtv | It's just that this may be a side effect of the leases upload. | 06:27 |
| jtv | Which happens once a minute. | 06:27 |
| bigjools | jtv: I am offski. Speak tomorrow. | 06:32 |
| jtv | nn bigjools | 06:34 |
| === CyberJacob|Away is now known as CyberJacob | ||
| === CyberJacob is now known as CyberJacob|Away | ||
| rvba | jtv: What I'm seeing is that the *DNS* config still gets rewritten every minute. | 09:10 |
| jtv | Different! | 09:11 |
| jtv | Julian said he saw the DHCP probe followed by the DHCP config rewrite. | 09:11 |
| jtv | He was running from package, so he wasn't looking at the "make run" output which combines all logs. | 09:12 |
| rvba | I don't see that. I see the DHCP probe followed by the DNS config rewrite. | 09:12 |
| jtv | In the same log? | 09:12 |
| rvba | I'm also running from a package. | 09:12 |
| jtv | I'm just making guesses about what he might have seen... | 09:12 |
| rvba | jtv: hang on, I forgot to look at one log file… | 09:13 |
| rvba | No trace of the DHCP rewrite. | 09:14 |
| rvba | (I'm using the daily package.) | 09:14 |
| jtv | So... progress. | 09:14 |
| rvba | jtv: Something else (but related): have a look at the apache log: http://paste.ubuntu.com/6884158/. As you can see, we first query ".../interfaces?op=list" and then we get redirected to "/interfaces/?op=list". If we could issue the right request from the start (with the '/' at the end), we would avoid one redirect every minute. This is a detail of course but I thought it was worth mentioning. | 09:15 |
| jtv | Yes... you can always file a tech-debt bug. :) | 09:17 |
| rvba | heh :) | 09:17 |
| rvba | jtv: src/maasserver/dns_connect.py says saving a NodegroupInterface triggers a full DNS rewrite… | 09:21 |
| jtv | Yes, DNS rewrites are more easily triggered. I was just looking for a DHCP rewrite... | 09:21 |
| rvba | bigjools: can you double check that you're still seeing the DHCP config being rewritten every minute? | 09:40 |
| rvba | bigjools: I just tested the daily package and I don't see that. I see the *DNS* config is still being rewritten every minute. | 09:41 |
| bigjools | rvba: no I can't, sorry. But I was using the daily and it was definitely doing it on the cluster celery. | 09:41 |
| bigjools | probe » config write » restart dhcp | 09:42 |
| rvba | I don't see that. And I've got a DHCP server configured and running. | 09:43 |
| bigjools | ok I'll check again tomorrow | 09:44 |
| bigjools | too late for me now | 09:44 |
| rvba | Okay. | 09:44 |
| bigjools | eating with plate on my lap and the laptop on the arm of the chair here | 09:44 |
| jtv | A recipe for success... | 10:11 |
| rvba | jtv: bug filed: https://bugs.launchpad.net/maas/+bug/1276985 | 10:14 |
| ubot5 | Launchpad bug 1276985 in MAAS "The "Edit cluster interface" page shows an undocumented "Foreign dhcp ip" field." [Medium,Triaged] | 10:14 |
| rvba | bigjools: ^ (with screenshot ;)) | 10:14 |
| jtv | Thank you ubot5 | 10:14 |
| bigjools | rvba: grar, docs | 10:15 |
| bigjools | rvba: why do you keep talking about editing it? You can't edit it. | 10:15 |
| bigjools | rvba: oh I see the page | 10:16 |
| bigjools | my bad. WTF. That didn't used to be there. | 10:16 |
| rvba | bigjools: :) | 10:17 |
| * jtv is glad to see that the screenshot was of some use. | 10:19 | |
| rvba | jtv: time for a tiny review? https://code.launchpad.net/~rvb/maas/fix-none-vlan/+merge/205119 | 10:54 |
| jtv | Sure | 10:55 |
| rvba | Thanks. | 10:56 |
| === marcoceppi_ is now known as marcoceppi | ||
| tomixxx3 | hi, how can i completely remove maas? i have executed sudo "apt-get purge maas ; sudo apt-get autoremove" and rebooted the server, but after reboot, i am still able to open the MAAS dashboard... | 13:57 |
| tomixxx3 | ok, i have it already | 14:06 |
| tomixxx3 | hi, after executing "$ maas-cli maas node-groups import-boot-images" i get 'HttpResponse' object has no attribute '_is_string' | 14:31 |
| tomixxx3 | does anyone know what's the problem here? | 14:31 |
| gmb | tomixxx3: Can you try running `maas-cli refresh` and see if that fixes it? | 14:44 |
| tomixxx3 | gmb: hmm, now i have already started sudo maas-import-pxe-files | 14:44 |
| tomixxx3 | gmb: should i abort that? | 14:44 |
| tomixxx3 | gmb: furthermore, i have executed sudo add-apt-repository cloud-archive:tools and sudo apt-get-update before installing MAAS but it seems i always get version 1.0... | 14:45 |
| gmb | tomixxx3: No, just let it run | 14:45 |
| gmb | tomixxx3: What does `apt-cache policy maas` return? | 14:46 |
| tomixxx3 | gmb: ah, it says 1.4+bzr16934+ | 14:47 |
| tomixxx3 | gmb: but i see not the very same output and screens like described here: http://maas.ubuntu.com/docs/install.html#post-install | 14:47 |
| gmb | tomixxx3: Okay, what *do* you see? | 14:47 |
| tomixxx3 | gmb: slightly different screens and output most of the time ;) | 14:48 |
| tomixxx3 | gmb: for example, i cannot enter the IP address at start, i need to call dpgk-reconfigure... afterwards | 14:48 |
| tomixxx3 | gmb: on start i get a (wrong) IP assigned | 14:48 |
| tomixxx3 | gmb: furthermore, "--password" is missing in the createadmin statement | 14:48 |
| tomixxx3 | last but not least, i cannot execute maas-cli maas node-gruops import-boot-images | 14:49 |
| gmb | tomixxx3: Okay, one thing at a time. | 14:50 |
| gmb | tomixxx3: Have you created an admin yet? | 14:50 |
| tomixxx3 | gmb, i have done this installation for mulitple times now | 14:50 |
| tomixxx3 | gmb: i can create an admin, but i have to add "--password=xxxx" | 14:51 |
| tomixxx3 | thats not the problem | 14:51 |
| gmb | tomixxx3: So which of the things that you listed as a problem is actually the problem? :) | 14:51 |
| tomixxx3 | gmb: none of them | 14:52 |
| tomixxx3 | gmb: ah, when i set the http proxy in the maas dashboard, save, and delete the proxy again, it seems that the proxy does not appear even if it is not displayed anymore | 14:53 |
| tomixxx3 | gmb: but that's a looooooooong story | 14:53 |
| tomixxx3 | gmb: but the reason why i reinstall maas at the moment | 14:53 |
| tomixxx3 | another thing: nodes are not able to download packages | 14:55 |
| tomixxx3 | but maybo this is because of my tricky architecture | 14:55 |
| tomixxx3 | -appear, +disappear | 14:56 |
| tomixxx3 | but i know, we don't live in a perfect world ;) | 14:56 |
| gmb | tomixxx3: I'm not sure that this is something that we can solve easily over IRC — partly because you'll have to explain the whole architecture to me. Can you file a bug for the proxy issue please (https://bugs.launchpad.net/maas/+filebug) with instructions for reproducing the problem? | 14:58 |
| gmb | tomixxx3: Usually with nodes not being able to download packages that's a routing issue | 14:58 |
| gmb | tomixxx3: (Although they should just use the proxy if they can) | 14:59 |
| tomixxx3 | in short the architecute is: 2 nodes + 1 maas server connected via switch. additionally, a 2nd network interface in the maas-server connects the server to an external (university) network which is more or less a black box to me ;) | 14:59 |
| tomixxx3 | but i need the connection to the university in order to get internet-access - at least - at the maas-server | 15:00 |
| gmb | tomixxx3: Okay, that's not all that tricky; it's just that the nodes on the private / MAAS-managed network need to be able to get out to the internet *via* the MAAS server. | 15:00 |
| tomixxx3 | yeah... i have not achieved this milestone yet | 15:01 |
| tomixxx3 | however, i have tried a lot of things, i guess | 15:01 |
| gmb | tomixxx3: Two things you can do (assuming DHCP is set up correctly in MAAS; it should be from what you've said): | 15:01 |
| tomixxx3 | yep, dhcp works | 15:02 |
| gmb | tomixxx3: Okay, so, gimme a sec here... | 15:02 |
| tomixxx3 | k | 15:02 |
| tomixxx3 | gmb: altough, the problem is, also my university has a dhcp proxy which wants to give all my beautiful nodes some beuatiful IP :-) | 15:02 |
| tomixxx3 | therefore, i have two network interfaces | 15:03 |
| gmb | tomixxx3: Ah, okay, so you're *not* having to route all traffic out through the MAAS server then? | 15:03 |
| gmb | The nodes have two NICs each? | 15:03 |
| gmb | And one of those NICs on each node is connected to the university network | 15:04 |
| gmb | The other to MAAS | 15:04 |
| tomixxx3 | gmb: yes, they have fixed IPs from uni | 15:04 |
| tomixxx3 | gmb: but the problem is, if they get Ip from university and dhcp from university, they try to download images from university instead from maas server | 15:04 |
| gmb | tomixxx3: So. There's no need to plug them in to the university network *if* we configure the MAAS server to act as a gateway. | 15:05 |
| gmb | That should solve both that problem. | 15:05 |
| tomixxx3 | gmb: that's right. therefore i have 2 network interfaces at the server | 15:05 |
| tomixxx3 | gmb: the private network works together with the private maas dhcp | 15:05 |
| tomixxx3 | gmb: only problem left: how do the nodes are able to download their packages? | 15:05 |
| gmb | tomixxx3: Well, once the MAAS server is configured properly as a getaway (i.e. IPTables is set up, IP forwarding is turned on and so on) the nodes will be able to connect to the outside world through the MAAS server — essentially as though it were just a firewall. | 15:06 |
| gmb | Have you touched IPTables on the MAAS server already? | 15:06 |
| gmb | s/getaway/gateway/ | 15:07 |
| tomixxx3 | i have alredy tried sth with iptables but it did not work | 15:07 |
| tomixxx3 | but maybe i just misconfigured sth | 15:07 |
| tomixxx3 | i dont know well about iptables | 15:07 |
| tomixxx3 | i would appreciate if u could offer me some iptables statements just to put into "interfaces" file ;) | 15:08 |
| tomixxx3 | well, i can tell what i ve tried: "post-up iptables -t nat -A POSTROUTING -o eth1 -j Masquuerade" | 15:09 |
| tomixxx3 | i put this statement to eth0, which is connected to the nodes | 15:09 |
| tomixxx3 | and eth1 is the interface card which connects me to the university network and to the internet | 15:09 |
| gmb | tomixxx3: Okay, just working on a script for you to try. | 15:09 |
| tomixxx3 | gmb: kk, thank you, i really appreciate this! | 15:10 |
| gmb | tomixxx3: Try this script: http://paste.ubuntu.com/6885719/ | 15:18 |
| gmb | (on the MAAS server, run it as root) | 15:18 |
| tomixxx3 | kk, i do not have to modify the scrip in some way? | 15:19 |
| tomixxx3 | eth0 = private network, eth1 = university network | 15:20 |
| gmb | tomixxx3: I've already put those values in. | 15:20 |
| tomixxx3 | gmb: kk, thank you very much! | 15:20 |
| gmb | tomixxx3: No worries. | 15:21 |
| tomixxx3 | gmb: i have a question: | 15:23 |
| gmb | tomixxx3: Sure. | 15:23 |
| tomixxx3 | gmb: the dhcp packages from my maas-server will not travel to university-network? | 15:23 |
| tomixxx3 | gmb: i dont want do destroy university network | 15:23 |
| gmb | tomixxx3: If you've set up DHCP correctly it will *only* work on the private network. | 15:23 |
| tomixxx3 | gmb: even if lets say the range is 10.0.0.100 to 10.0.0.200 in my private network but also at the universtiy this range is used= | 15:24 |
| tomixxx3 | ? | 15:24 |
| gmb | tomixxx3: Well, I'd suggest using a different IP range — 192.168.1.0/24 for example — just to avoid confusion. | 15:25 |
| gmb | Otherwise you're going to have sad times if you do get it wrong. | 15:25 |
| gmb | Because you won't be able to tell which interface was on which network. | 15:25 |
| tomixxx3 | gmb: so i have to ask a network admin, if the ip range 10.0.0.100 to 200 is free? | 15:26 |
| tomixxx3 | just to clairfy | 15:26 |
| gmb | tomixxx3: No. | 15:26 |
| gmb | tomixxx3: It's a private network | 15:26 |
| gmb | You can use whatever range you like | 15:26 |
| gmb | tomixxx3: And it won't ever leak out into the outside world. | 15:26 |
| gmb | tomixxx3: Because the MAAS server is NAT'ing all the packets coming through it from the nodes. | 15:26 |
| tomixxx3 | and the server is only able to send dhcp pacakgeS WITHIN the private network | 15:27 |
| gmb | Correct. | 15:27 |
| tomixxx3 | but it would be different if i would BRIDGE both interfaces, right? | 15:28 |
| gmb | YES. | 15:29 |
| gmb | DO NOT DO THAT. | 15:29 |
| tomixxx3 | KK :-) | 15:29 |
| gmb | OR BAD THINGS (POTENTIALLY) HAPPEN | 15:29 |
| gmb | NAT on university networks is _much_ safer :) | 15:30 |
| bjf | i'm still new enough that i don't understand the stages a system goes through in maas. after i have a working MAAS server up and running and i power cycle a different node, that node netboots from MAAS and is "discovered" correct? | 16:50 |
| bjf | after "discovery" the node is powered down. next, i tell MAAS to "Commission" the node. the node is powered back up, netbooted from maas; maas runs some things on it and then powers it down again, correct? | 16:51 |
| bjf | i have a node whose status is "Commissioning" and is still powered up and has been this way since last week | 16:52 |
| === CyberJacob|Away is now known as CyberJacob | ||
| === zz_mwhudson is now known as mwhudson | ||
| bigjools | bjf: your understanding is correct. Either the power detection is not working or it didn't PXE boot. | 23:04 |
| bjf | bigjools, ok, thanks | 23:08 |
| bjf | bigjools, it's beta HW so could be several reasons why it's "stuck" | 23:09 |
| bigjools | bjf: you can check the power detection worked by going into the edit node page and see if the parameters are correct. which power type are you using? | 23:09 |
| bigjools | ah ok :) | 23:09 |
| bigjools | bjf: the other possibility is that the commissioning scripts are hung | 23:10 |
| bigjools | which is always possible on new hardware | 23:10 |
| bjf | bigjools, i'm using CDU | 23:14 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!