[14:46] <l1l> any idea why im getting a 403 when trying to commission a node?
[16:14] <jefferai> My nodes, on commission, are getting a bunch of 500 INTERNAL SERVER ERROR lines near the end, and the logs say that it's due to failed authentication
[16:14] <jefferai> In browsing around the web it seems like this might happen if the clocks are off
[16:14] <jefferai> but they clocks are in sync
[16:14] <jefferai> Although the guest (KVM, running on the same node) is using UTC, and the host isn't
[16:15] <jefferai> but it should be able to account for timezones properly, right?
[16:20] <jefferai> Well, setting host to UTC made no difference
[16:20] <jefferai> This is on Trusty, trying to deploy a Trusty image
[16:20] <jefferai> I can hit the endpoint with my browser, so it's up
[16:23] <jefferai> Oh, doh
[16:23] <jefferai> I was seeing an exception related to me trying to access it in my web bwoser
[16:23] <jefferai> in fact, it's an "Integer out of range" exception
[16:24] <jefferai> http://paste.kde.org/p2xfi9ukg
[18:04] <l1l> is it possible to use maas without juju? is maas responsible for handing the SSH keys to the nodes? finally got the node to deploy and install ssh.. but can't ssh to the machine using the user "ubuntu"
[18:28] <jtv> l1l: yes, you can use maas without juju.
[18:28] <jtv> When you allocate a node, maas installs your ssh key on the node — assuming you tell it what your ssh key is first, of course.  :)
[18:29] <jtv> So you _should_ have been able to ssh in, e.g. after clicking "Start node" in the UI.
[18:30] <jtv> You do need to wait for the OS to install and reboot — though we're planning to integrate that part into the starting process more sensibly.
[18:36] <l1l> it went through the os install, then prompted for additional stuff.. I checked SSH there and it installed, then rebooted
[18:36] <l1l> however, getting access denied with the ubuntu user
[18:36] <jtv> It shouldn't have asked you for anything.
[18:36] <jtv> What did it prompt you for exactly?
[18:37] <jtv> You're talking about the console on the node itself, right?
[18:37] <l1l> No, talking about ssh'n to the node from the maas machine (where the ssh key was created)
[18:38] <l1l> Though I can't get logged in directly on the console either. It prompts for the password on the ubuntu user
[18:38] <jtv> And you uploaded your public ssh key, right?  Do you have that key on the machine from which you're trying to ssh, and on the account you're running in when you're trying to ssh?
[18:39] <jtv> Oh, it's not the ssh prompt saying "I can't verify this signature myself, please make sure this is the right machine we're talking to"?
[18:39] <l1l> Yes. I uploaded the ssh key via the maas web UI. Logged into the maas machine where the key was created and ssh ubuntu@x.x.x.x
[18:40] <l1l> It did prompt that, to which I answered Yes to accept.. then the password prompt
[18:40] <jtv> So presumably ssh was using the private key that matches the public key that you uploaded.  That sounds right.
[18:41] <jefferai> jtv: I'm getting tracebacks at the end of a node commissioning...any thoughts? http://paste.kde.org/p2xfi9ukg
[18:41] <jtv> jefferai: that's a new one on me.  :(
[18:42] <jefferai> Fresh install of maas on trusty, with a trusty node being provisioned
[18:42] <jtv> This should be pretty straightforward — we're used to that just working.  Do you have any custom commissioning scripts?
[18:43] <jtv> l1l: the first prompt would be normal, but the second one...  it's not a client-side password that you put on the key pair?
[18:43] <jtv> Probably a stupid question, but just walking through the possibilities.
[18:45] <jtv> jefferai: I'm afraid I'm not actually here, technically, and need to go do other things.  But it sounds as if maybe "lshw" produced some wildly improbable numbers.
[18:45] <jtv> jefferai: could you try running "lshw -xml" on the node, and looking through such things as memory to look if maybe it's reporting bytes where it should be reporting megabytes or something?
[18:46] <jtv> jefferai: it could be specific to the kind of machine you're running on.
[18:46] <jefferai> it's a kvm virtual machine
[18:46] <jtv> so there may easily be something weird then.
[18:46] <jefferai> not sure how I'd run on lshw on the node though, as I don't know how to ssh in during the install and at the end of the install it shuts itself down
[18:47] <jtv> Could you just define a VM with the same parameters?
[18:47] <jefferai> oh, sure
[18:47] <jefferai> not necessarily using the same kernel and such though
[18:47] <jtv> That's a pity.  It looks as if there's something weird in, probably, the memory or storage size; or just maybe an insane CPU count.
[18:48] <jtv> Maybe it's just expressed in a different way than we're used to parsing, and causing a weird value.
[18:49] <jtv> If memory or storage is blank, or left out, or NaN, that should be OK.  But if it's a huge number, kaboom.  Or _maybe_ finding unexpected text such as "lots" might produce something like this.
[18:51] <jtv> l1l: the rsyslog for that node, in /var/log/maas, might have something helpful.  Also try the -v option to ssh, just to make sure that it's really finding your keypair.  What should have happened is that you uploaded your public key, and ssh uses your private key.
[18:51] <jtv> Also, another box to check: make sure the ssh key is registered for the same maas user identity which allocated the node.
[18:52] <jtv> (Although maas won't actually let you allocate a node, at least through the UI, if you haven't uploaded a key).
[18:53] <jtv> And now, I really must run.  I'll leave IRC open so l1l & jefferai, do please ping me as you learn more.  There may be lessons in there for improving maas, or its documentation, or whatever.
[18:54] <jefferai> jtv: nothing odd - 2G ram, 4 cpu, 20G (virtual) hard drive. I'll let you know if I find anything
[18:57] <magicrob1tmonkey> i had a problem in a similar area because lsblck reports a device as cciss!c0d0
[18:58] <magicrob1tmonkey> and its at /dev/cciss/c0d0
[18:58] <magicrob1tmonkey> it only affected curtin though
[19:00] <jefferai> hm...maybe I need to e.g. not use virtio
[19:00] <jefferai> maybe it's not understanding /dev/vda
[19:00] <jefferai> or something
[19:09] <l1l> thanks jtv, finally got in! must have been something silly I overlooked
[20:08] <jtv> magicrob1tmonkey, jefferai: we use xpath to extract the information from xml output...  I don't _think_ that “!” needs to be an entity code, but I could be wrong...
[20:09] <jtv> l1l: it may be that the node simply wasn't ready yet.  We just switched to using the Curtin (“fast-path”) installer by default; otherwise, you may speed up installation by making nodes use that instead of the “default” installer.
[20:09] <jtv> You can do that en masse on the nodes listing by checking off all or some nodes, selecting “Use fast installer” from the top-left dropdown, and hitting Go.