[04:26] <rawang> hello everyone, with maas 1.4, how to specify a specific version of kernel that nodes will use? include enlist, commision, and deployment stage?
[04:30] <rawang> Configure MaaS to use specific image for PXE
=== sfeole` is now known as sfeole
[06:00] <jtv> rawang: for now it's always just the latest available kernel for the release series.
[06:01] <jtv> (Though we do plan to change that)
[06:02] <rawang> jtv, i have check the precise boot image, it's still 3.2 kernel (12.04.1)
[06:03] <jtv> So you want to run with a newer kernel?
[06:16] <rawang> jtv, yes, want to run 3.11 kernel
[06:21] <jtv> I suppose a workaround for deployment would be to install that kernel on the system and reboot.
=== CyberJacob|Away is now known as CyberJacob
[07:24] <rawang> jtv, the problem is the server need 3.11 kernel for nic driver, but during enlist and commission and installation, it alway use 3.2 kernel, right?
[07:25] <jtv> Yes...  you should be able to commission with a newer series, but installation would be harder.
[07:26] <jtv> That's where it comes down to custom kernel images, which we haven't done yet.
[08:08] <rawang> kentlin, ^^
[08:27] <rvba> gmb: so there is no support for CONNECT so request (<2) apparently cannot do proper https proxying… but in maas-test we don't need "proper" proxying (i.e. we don't care if the proxy is used as a relay for the request)… so there is still hope…
[08:27] <rvba> requests*
[08:28] <gmb> rvba: Are you thinking a the scenario where we pre-fill the proxy cache and then only use it when running in no-internet-access-mode?
[08:29] <gmb> s/a the/of a/
[08:29] <gmb> Or am I misunderstanding?
[08:29] <rvba> gmb: no, I'm thinking about using the proxy as a relay where the *proxy* does the https connection.
[08:30] <gmb> Aaah. Interesting.
[08:30] <rvba> Not sure it's possible… but I'm looking into it…
[08:30] <gmb> Excellent, thanks.
[08:32] <rvba> btw, I've got 2 branches up for review: https://code.launchpad.net/~rvb/maas/az-api-list-2/+merge/198922 and https://code.launchpad.net/~rvb/maas/admin-method-2/+merge/199049
[08:32] <gmb> rvba: I'll take a look after I've had some more coffee
[08:33] <rvba> The second one is a clean-up jtv might be interested to review ;)
[08:33] <gmb> rvba: Are you suggesting this setup: simplestreams - (http) -> proxy - (https) -> interweb ?
[08:33] <rvba> gmb: exactly
[08:33] <gmb> rvba: That could work. We can't specify https_proxy *at all* for simplestreams, or else it breaks.
[08:34] <gmb> If you give it an http url by accident (I just did) you get
[08:34] <gmb> simplestreams.util.SignatureMissingException: No signature found!
[08:34] <rvba> gmb: even if it works, it will require monkeypatching I'm afraid.
[08:34] <gmb> Yuck
[08:39] <gmb> I like a library that uses exclamation marks in its error messages.
[08:39] <gmb> "Oh no! Calamity!"
[08:39]  * gmb is clearly wearing his early morning grumpy trousers; goes to get coffee
[09:59] <rvba> gmb: about the https proxy issue, I had another idea : simplestreams can use requests or urllib/urllib2 (see the top of the file /usr/share/pyshared/simplestreams/contentsource.py).  Maybe we can simple monkey patch simplestreams very lightly to force it to use urllib/urllib2 even if requests is available et… voilà!
[10:04] <gmb> rvba: Ooh, that's filthy. Especially considering that we're shelling out to simplestreams, not using it as a library...
[10:04] <rvba> gmb: ah!  This probably means we can't do that then :/
[10:05] <gmb> rvba: Yeah... Using it as a library would be non-trivial now.
[10:07] <rvba> gmb: maybe it would be simple.  It's worth a shot: it might be just a matter of calling the right uvtool method.
[10:08] <rvba> jtv: I'll take it
[10:08] <gmb> rvba: Well, let's get the AZ stuff done first and then we can take a look at it — at worst, it's a nice run-up-to-Christmas project for me and allenap.
[10:09] <rvba> gmb: k
[10:13] <rvba> gmb: another (much simpler) solution is this: if the sync step fail, continue anyway.  If there is a real problem (i.e. if the image is not cached already) the next step (creating the VM) will fail anyway.
[10:14] <gmb> rvba: Yeah, that might not be such a terrible way to deal with it.
[10:19] <allenap> gmb, rvba: Why are we shelling out to simplestreams btw?
[10:19] <gmb> Let's move this over to #maas
[10:19] <rvba> allenap: we're shelling out to uvtools.
[10:19] <gmb> Haha
[10:19] <gmb> Sorry
[10:19] <rvba> gmb: :)
[10:20]  * gmb has had it beaten into him
[10:20] <allenap> rvba: Ah, okay.
=== jamespag` is now known as jamespage_realme
=== cmagina_ is now known as cmagina
=== CyberJacob is now known as CyberJacob|Away
=== hazmat` is now known as hazmat
=== Daviey_ is now known as Daviey
=== jamespag` is now known as jamespage
[17:19] <roaksoax> rvba: ping
=== CyberJacob|Away is now known as CyberJacob
=== CyberJacob is now known as CyberJacob|Away
[23:50] <_bjorne> someone now why im alwas get this answer in apache logfile: /MAAS/metadata//2012-03-01/user-data HTTP/1.1" 404 200 "-" "Python-urllib/2.7" on the second and so on.. the first node is function not the others.
[23:53] <_bjorne> 1st node GET /MAAS/metadata/curtin/2012-03-01/user-data HTTP/1.1" 200 33593 "-" "Python-urllib/2.7
[23:54] <_bjorne> im mean this: "GET /MAAS/metadata//2012-03-01/user-data HTTP/1.1" 200 12311 "-" "Python-urllib/2.7