[16:11] <tmclaugh[work]> hi, i see that runcmd runs ar an "rc.local like" level.  Is there any way to run commands before other modules?
[16:11] <tmclaugh[work]> We're having some issues with yum repos and that's causing our puppet installation to fail.
[16:12] <tmclaugh[work]> we tried adding a 'yum makecache' before the installation of puppet as a precaution but it doesn't help since the command is executed after the attempt to install puppet.
[17:27] <smoser> well, tmclaugh didn't stick around long enough
[18:49] <Chocobo> Hi.  So I am having a small problem with persistent network naming rules and I think I need to delete the persistent net name rules in /etc/udev/rules.d/.  I am thinking about creating a per-instance script to do that.  Is there a better way?
[19:05] <smoser> Chocobo, what is your hypervisor?
[19:06] <smoser> because persistent naming network rules should work correctly.
[19:06] <smoser> they basically turn off for known MAC that are virtual
[19:06] <smoser> by mac-id allocation
[19:15] <Chocobo> smoser: kvm
[19:16] <smoser> on debian or ubuntu the default mac addrs should not get perisstent rules created.
[19:16] <smoser> unless youre creating macs yourself and not doing it correctly
[19:17] <smoser> s/creating macs /creating mac addresses/
[19:18] <Chocobo> smoser: Ahh, I am using it to make a CentOS image.
[19:18] <smoser> hm..
[19:18] <smoser> i think udev has that function uptream
[19:18] <smoser> probably just too old there.
[19:18] <smoser> or maybe locally patched in ubuntu or debian
[19:19] <Chocobo> I already have the official ubuntu image.  I should check and see how it is done.   I don't see anything in /lib/udev/rules.d/75-persistent-net-generator.rules that looks like it would match the virtual nic's MAC
[19:20] <Chocobo> I acutally don't mind having persisting nic names per instance (vm's can have multiple nics)
[19:24] <smoser> Chocobo, it works right
[19:24] <smoser> mostly.
[19:24] <smoser> ie, virtio bus will do the right thing generally.
[19:25] <smoser> you can get crazy and mess it up with plug unplug and such
[19:25] <smoser> but generlly it enumerates them in a consistent way
[19:26] <smoser> Chocobo, some related bugs with info
[19:26] <smoser> https://blueprints.launchpad.net/ubuntu/+spec/t-cloud-server-cloud-images
[19:26] <smoser> shoot
[19:26] <smoser> paste fail
[19:26] <smoser> https://bugs.launchpad.net/ubuntu/+source/udev/+bug/719418
[19:30] <Chocobo> So this rule: ENV{MATCHADDR}=="?[2367abef]:*", ENV{MATCHADDR}="" <--- that would match something like fa:16:3e:c1:c0:12  correct?
[19:46] <Chocobo> How strange.  CentOS has the same rule but it still writes the rule out.
[20:06] <smoser> Chocobo, i dont know.
[20:28] <smoser> ctracey, is there some reason why you'd not want 'tee' ?
[20:28] <smoser> https://bugs.launchpad.net/cloud-init/+bug/1241701
[20:28] <smoser> as in
[20:28] <smoser> output: {all: '| tee -a /var/log/cloud-init-output.log'}
[20:29] <smoser> the package command being logged makes quite reaosnable sense
[20:30] <smoser> but is thre a reason that capturing output generically is insufficient ?