[11:08] Morning [11:08] should squid-deb-proxy automatically be configured with the prefix from which your machines will be coming from? === cmagina_away is now known as cmagina [15:06] rvba: howdy!! [15:07] roaksoax: hi Andres [15:07] rvba: So I was wondering. I need to install 2 different commissioning images [15:07] rvba: or the same image in two different [15:07] rvba: directories, but it doesn't let me [15:08] rvba: so i want to install images in commissioning/ and xinstall/ dir [15:08] it doesn't let me due to saying something like "these ephemeral images have already been installed" [15:09] or similar [15:09] What command are you running exactly? [15:10] rvba: let's say I do: maas-provision install-pxe-image "--arch=$arch" "--subarch=$subarch" "--release=$release" --purpose="commissioning" --image="$tmpdir" [15:10] rvba: let's say I do: maas-provision install-pxe-image "--arch=$arch" "--subarch=$subarch" "--release=$release" --purpose="xinstall" --image="$tmpdir" [15:10] (see the different purpose) [15:11] $tmpdir is the same image dir. It fails to install into xinstall because it already has installed the images into commissioning [15:11] rvba: which is right cause it doesn't make sense to copy the same images to a different tftp directory [15:12] rvba: however, we are adding a different purpose (xinstall) for installation using an ephemeral image (fast path installer) [15:12] rvba: so, i was wondering if it would make sense to add a parameter like --symlink="place-to-symlink-to" [15:13] rvba: so that: maas-provision install-pxe-image "--arch=$arch" "--subarch=$subarch" "--release=$release" --purpose="commissioning" --image="$tmpdir" --symlink="xinstall" [15:13] roaksoax: Where is the code for maas-provision again? [15:14] rvba: that's src/provisioningserver (maas-provision is just a wrapper that uses src/provisioningsrever) [15:14] roaksoax: ah right, let me have a look at that code. [15:14] GOOD MORNING [15:16] rvba: https://code.launchpad.net/~andreserl/maas/maas_fast_path_installer --> this is what I'm adding so far [15:16] bigjools: morning! [15:16] bigjools: us time this week? [15:16] roaksoax: the path used to store the image already seems to include the "purpose" ('/'.join([arch, subarch, release, purpose])) [15:17] roaksoax: yeah I am in Austin [15:18] rvba: yes. So if I try to install an image dir to purpose 'commissioning' and then the same image dir to purpose 'xinstall', it will fail due to have 2 identical dirs (which is expected). However, I need to access the same commissioning images using a different purpose [15:18] rvba: so I was wondering if it was makes to add the --symlink option that basically does "if I install image with purpose commissioning, symlink that purpose dir into XYZ" [15:18] bigjools: cool! [15:20] roaksoax: maybe I'm missing something but what I see in the code is that the dirs should not be identical if when you install the same image to another "purpose". [15:22] rvba: Let's look at it another way. I need to use the 'commissioning' images for a different purpose ('xinstall'). [15:22] roaksoax: too cool, it's bloody freezing :) [15:23] so I need to symlink /var/lib/maas/tftp//generic//xinstall to use /var/lib/maas/tftp//generic//commissioning [15:23] rvba: ^^ [15:23] Ah I see. [15:23] rvba: for that I'd like to add --symlink="purpose", which will do that (when using maas-provision install-pxe-image [15:23] Right now you only have the option so re-install it right. [15:23] ? [15:23] bigjools: heh I don't think is that bad is it? [15:24] roaksoax: it depends on your relative outlook :) [15:24] roaksoax: the part I don't get is why you get the error you're getting when you're simply trying to install the same image for a different purpose. This should work all right. [15:25] roaksoax: having a symlink instead of duplicating the file is nice. But it should be just an optimisation over installing the same image twice. [15:26] rvba: yes that's what I want to do instead of installing the same image twice [15:26] bigjools: heh.. I live in Florida so I feel you :) [15:27] roaksoax: EXACTLY! [15:27] bigjools: it's 16C here and its cold for me [15:28] rvba: the error when trying to install $imagedir into 'xinstall' purpose *after* installing them into 'commissioning' purpose is: http://pastebin.ubuntu.com/1559858/ [15:29] rvba: but anyways I guess I'll simply add the --symlink option [15:29] since that does make more sense [15:29] roaksoax: Sounds like an useful optimisation. [15:29] rvba: cool thanks [15:29] roaksoax: but you should not be getting that error. [15:29] roaksoax: you in Miami? [15:29] bigjools: yes I am :) [15:30] warm :) [15:30] indeed :) [15:30] i love the weather [15:30] roaksoax: I'm looking into your error now… [15:30] rvba: yeah but i guess it is comparing all the already installed dirs to check if there's one installed already [15:30] rvba: cool thanks [15:31] roaksoax: it's not that clever :). I suspect it's failing because of something else. [15:33] :) === matsubara is now known as matsubara-lunch [15:38] roaksoax: when you install an image, the source gets cleaned up afterwards. That's why the second run fails: it does not find the image to import. [15:39] roaksoax: if you duplicate the directory containing the images first and run the second import on the copy, it will work. [15:39] roaksoax: note that the symlink thing won't solve that pb :). [15:40] rvba: ah I see that now [15:41] clear [15:41] ls [15:41] err [15:41] heh :) [15:41] rvba: good catch though! [15:41] * roaksoax needs some coffee apparently [16:55] rvba: I have a question... say i have 2 node controllers and 1 region. In order for me to deploy 2 different networks (on of each of the cluster), do I still need to be accessible to the region controller? === matsubara-lunch is now known as matsubara [16:57] roaksoax: not sure I understand the question… but every cluster needs to be able to contact the region if that's what you're asking. [17:01] rvba: so the clusters/region communicate through 192.168.0.1/28, while I want each cluster to deploy on nodes using 192.168.1.0/24 [17:01] and 192.168.2.0/24 [17:01] roaksoax: that's fine, so the clusters will have 2 interfaces. [17:01] rvba: so they don't have to be able to contact the region then [17:02] Yes they do [17:02] rvba: so all the deployed nodes have to have 2 interfaces then [17:04] roaksoax: Either that or you need to add the appropriate routes so that they can contact the region through the cluster. [17:04] right [17:22] bigjools: there's no merge bot for packaging.precise.sru ? [17:22] roaksoax: no, because it's not a recognised release branch [17:22] it's an integration branch [17:22] so you need to merge manually [17:22] and push up [17:22] ok cool [17:22] thanks :) [17:22] I turned on appendrevisions for it so it should be safe [17:23] cool! [17:32] rvba: btw.. it is now the cluster that runs the tftp server and stores the images ? [17:32] roaksoax: yes [17:32] ok cool === freeflyi1g is now known as freeflying [19:24] bigjools: btw... i was wondering whether you guys were planning to SRU the kernel_opts tag stuff to precise branch [19:30] t/win 9 [19:47] matsubara: ping? [19:47] roaksoax, hi [19:47] matsubara: howyd!! quick question. Have you played with tags much? [19:47] roaksoax, yes, a bit. How can I help? [19:48] matsubara: a ny ideas? ubuntu@maas:/usr/share$ maas-cli admin tag update-nodes "xinstall" add=node01 [19:48] { "removed": 0, "added": 0 [19:48] } [19:51] roaksoax, did you create the tag xinstall first? [19:51] yes [19:51] also the doc says: The nodegroup parameter, which restricts the operations to a particular nodegroup, is optional, but only the superuser can execute this command without it. [19:52] maybe you need the nodegroup parameter as well since you seem to be running with the ubuntu user [19:52] matsubara: yeah: http://pastebin.ubuntu.com/1560577/ [19:53] or node01 is not matching any node in your cluster. are you sure that's the system_id for the node? [19:54] matsubara: ah!1 that's why then [19:54] thanks for the info :) [19:54] no problem :-) [20:25] roaksoax: no plans to backport any tags [20:29] bigjools: ok === matsubara is now known as matsubara-afk [21:51] bigjools: why do I need maas-dns installed in order for the UI to not fail: http://paste.ubuntu.com/1560879/ [22:02] j/win 14 === lborda_ is now known as lborda [23:39] roaksoax: well that sucks. Ask rvb about it, he wrote the start up wsgi stuff. I am a little surprised that this happens but it might be an untested setup :( [23:54] bigjools: yeah anyways, I already filed a bug. I'll check with him tomorrow