2017-04-01 23:12
Hello, how can I run the vagrant/compose setup with DHCP?

2017-04-01 23:27
vagrant as the admin or nodes? I could never figure out how to make the vagrant respect DHCP boot

2017-04-01 23:28
@chilicat what are you trying to do w/ Vagrant? We don't use it that much

2017-04-01 23:28
not sure what you mean with admin or nodes... DR runs on the base node as docker container

2017-04-01 23:28
are you running the base via Vagrant?

2017-04-01 23:28
yes

2017-04-01 23:29
(btw: your admin password issue may just be complexity requirements - the password must be alpha and numeric)

2017-04-01 23:29
yes the password must be "non-simple" just wanted to report that the ui doesn't seem to tell you :)

2017-04-01 23:30
ah, will add some guideance

2017-04-01 23:30
I could never find a way for a Vagrant node to DHCP boot - it seems like there's a very strong assumption that you start with a working image.

2017-04-01 23:31
so, you'd have to mix and match VirtualBox VMs & Vagrant to simulate boot. In that case, I'd recommend using 100% virtual box.

2017-04-01 23:31
in vagrant there is a docker setup - maybe my question is not vagrant specific. I want to add the dhcp service to my setup (its off by default)

2017-04-01 23:31
do I use ./init_files.sh ?

2017-04-01 23:31
if you use run-in-server.sh then add --con-dhcp --con-provisioner

2017-04-01 23:33
the recommended approach is to rerun run-in-server with the added commands and it will update the configs.

2017-04-01 23:33
I will try that.

2017-04-01 23:33
does it also use the docker compose files? or is that a different setup?

2017-04-01 23:33
digitalrebar/deploy/compose

2017-04-01 23:35
run-in-system runs the Ansible digitalrebar.yml file

2017-04-01 23:35
that file then runs the command: "./setup.sh --tag {{ dr_tag }} {{ dr_workloads | default([]) | join(" ") }}" to correctly configure docker-compose config

2017-04-01 23:36
then rebar runs with docker-compose -up

2017-04-01 23:36
obviously, there's a lot of other setup steps, but that's the key elements of the DR config

2017-04-01 23:37
I see, yea it all runs to the setup.sh. I try my luc

2017-04-01 23:37
thanks

2017-04-01 23:37
So, YES. Everything ultimately runs to docker-compose.

2017-04-01 23:38
even the core/tools/docker-admin.sh scripts are doing the same thing. They just assume that you've already got the pre-reqs.

2017-04-01 23:39
we

2017-04-01 23:40
if all you need is DHCP/Provision.

2017-04-01 23:40
preview github.com/rackn/rocket-skates

2017-04-01 23:41
I also need the deployment mechanisms put a slim DHCP/PXE/TFTP in go sounds good

2017-04-01 23:42
is there any content management for DR?

2017-04-01 23:46
We use deployments to manage parallel application deployments and then track the state of them using the roles system. Not exactly what I'd consider content management. Do you have a specific use case or platform in mind?

2017-04-01 23:48
In a on-premise deployment we must deliver all our software packages (rpm, zip, docker containers, isos). I was just wondering if there was something where you manage the content and associate content with a deployment.

2017-04-01 23:48
I little like Katello

2017-04-01 23:49
Ok was able to add the dhcp service... vagrant uses quickstart.sh which also accepts the parameter you gave me.

2017-04-01 23:50
glad the vagrant is working! it's been a while since I was testing it

2017-04-01 23:50
I had some issue with the swapfile... I was switching back to a older ubuntu - but else it works

2017-04-01 23:51
got it, there are a few ways to handle that, yes. we've done "air gap" deploys where there's no external connectivity. Your request is a variant. takes some twiddling. @galthaus would be a better person to answer.

2017-04-01 23:51
I saw the issue, thanks for opening it - that helps me know what to check'

2017-04-01 23:53
since you can add your own roles, you could just inject a few bash or ansible commands that would pre/set the environment. also, the proxy & mirrors are configurable so you can point those to internal resources

2017-04-01 23:53
One more question, is it possible to use DR when you have absolutly no control over DNS and DHCP. I know the functionality is very limited in this case but it would be still possible to deploy software on existing machines,

2017-04-01 23:55
100% yes. that's pretty normal scenario. We would just need forward or next boot configured for DHCP (e.g.: we work w/ infoblox). For DNS, that's configurable too. Can an own, delegate, drive or ignore DNS. Pretty flexible about that.

2017-04-01 23:57
this is old, but mostly right > http://digital-rebar.readthedocs.io/en/latest/deployment/old/external-services.html?highlight=dns

2017-04-01 23:58
I found an out of date page about DHCP that I'll fix. please disregard > http://digital-rebar.readthedocs.io/en/latest/faq/dedicated_dhcp.html?highlight=dhcp

2017-04-01 23:58
thanks.

2017-04-01 23:58
call my vms pxe request works... just TFTP times out :)

2017-04-01 23:59
ah... that's sadly common. TFTP and Docker port mapping are not friends.

2017-04-01 23:59
that's a primary reason why we're doing Rocket Skates as a stand alone binary

2017-04-02 00:00
the last line is: "TFTP..."

2017-04-02 00:00
maybe it is doing something?

2017-04-02 00:00
it's going to be inconsistent because the TFTP ports are being assigned into ranges that Docker is unhappy about

2017-04-02 00:00
if you scroll back in this chat history, you'll see dicussions about this.

2017-04-02 00:02
which container has the TFTP server?

2017-04-02 00:03
provisioner?

2017-04-02 00:03
yes

2017-04-02 00:05
sorry, gtr... the TFTP discussion from earlier was on 2/16 w/ mech422 and @galthaus

2017-04-02 00:07
but dns is also in a the low range... that is not a problem for docker?

greg
2017-04-02 01:08
It is really a protocol problem.

greg
2017-04-02 01:09
Make sure you have tftp nat modules loaded into your kernel.

2017-04-02 01:20
I guess that would be "modprobe nf_nat_tftp"

2017-04-02 01:20
have to try that tomorrow.

2017-04-02 01:20
thanks for your heklp

greg
2017-04-02 01:20
yeah

2017-04-02 03:46
Figured out what the problem was. The vagrant box has two interfaces where the eth1 is in the shared network. But "Next Server" option is set to the ip address of the eth0.

greg
2017-04-02 13:23
Awesome

2017-04-03 17:42
What would y'all use for a bunch of metal other than OpenStack?

2017-04-04 00:42
Kubernetes

2017-04-04 00:43
Or just Ansible if you had a app already

2017-04-04 15:02
I mean to provide VMs. I need a whole bunch of discrete network stacks to prove out architectures. Can kvm-slave.sh work across metal?

greg
2017-04-04 15:06
currently - no. Do the VM sets need to have isolated networks?

2017-04-04 15:27
Nope.

2017-04-04 15:27
A "nice to have" option.

greg
2017-04-04 15:28
hmm - okay - I don't have a good option for any answer currently.

greg
2017-04-04 15:30
One is to do the k8s/flat hack. Take a set of machines, run docker on them with sub ranges and a super net. Than use tools/kvm-slave to start vms on that docker bridge, but that is ugly and fraught with peril.

zehicle
2017-04-04 16:12
there was one request that would treat each node as a dedicated cloud (and this isolated). You could use Rebar to build a node islands that would then be isolated.

zehicle
2017-04-04 16:13
cross node networking is always intersting - that's why SDN layers get involved

2017-04-06 20:05
"interesting" as in, omg, encapsulation. :)

2017-04-07 02:38
Is this no longer kosher? How to get more detailed debug info? rebar --debug nodes create '{"name"; "os1.newgoliath.com", "bootenv": "local"}' 2017/04/06 22:36:01 Talking to Rebar with https://127.0.0.1 (rebar:reebaar) 2017/04/06 22:36:01 Unable to create new node: Expected status in the 200 range, got 400 Bad Request

greg
2017-04-07 02:42
Rebar nodes update name json

greg
2017-04-07 02:42
Assuming os1 exists

2017-04-07 02:43
fresh install. no nodes

2017-04-07 02:44
rebar whoami 2017/04/06 22:37:44 Cannot determine what node this is: Expected status in the 200 range, got 404 Not Found

2017-04-07 02:45
just he phantom

2017-04-07 02:46
hostname -f os1.newgoliath.com

2017-04-07 02:48
duh, I see it. semicolon

2017-04-07 02:51
Thank you, Greg!

greg
2017-04-07 02:52
:slightly_smiling_face: no problem, Judd!

greg
2017-04-07 02:52
Glad to see you back hanging around

2017-04-07 02:53
I've gotta do SOMETHING with all this gear I have laying around.

greg
2017-04-07 02:53
:slightly_smiling_face:

2017-04-07 14:43
By default there's a router defined at 192.168.124.11, but I don't see any interfaces with that address. Can I ignore?

greg
2017-04-07 14:59
That is the Forwarder container's IP on the docker bridge. It allows you to use the tools/kvm-slave script.

greg
2017-04-07 15:00
In forwarder mode, you can bridge a second interface into the docker bridge to use the 124 network with external hosts.

greg
2017-04-07 15:01
You could reinstall with HOST mode (--access HOST) That will use the machine host networking for accessing out to machines. You would then create an admin network for what you already have with the external routers.

2017-04-07 15:02
what I already have?

greg
2017-04-07 15:07
I suspect that you took the defaults which is FORWARDER mode.

greg
2017-04-07 15:08
We are right now contemplating getting rid of this confusion, but that isn't ready yet.

2017-04-07 15:09
That's correct. I'm trying to plan a new install. I have 4 R410s. They have 2 nics, publically routable. I'm considering running one as the Rebar admin + whatever workloads. Is that doable?

2017-04-07 15:09
Or is there a happier path?

2017-04-07 15:10
Well, 1 nic on each is publically routed.

greg
2017-04-07 15:10
Yes - You probably want HOST mode with the internal nic as your internal pxe network.

2017-04-07 15:11
Seems like it.

greg
2017-04-07 15:11
You want to set the external IP to your node's intenral IP.

greg
2017-04-07 15:11
Just a second - Let me find an example.

greg
2017-04-07 15:12
```curl -fsSL https://raw.githubusercontent.com/digitalrebar/digitalrebar/master/deploy/quickstart.sh | bash -s -- --con-provisioner --con-dhcp --admin-ip=1.1.2.3/24 --access=HOST```

greg
2017-04-07 15:12
Make 1.1.2.3/24 the CIDR IP form of your internal network.

greg
2017-04-07 15:13
The API/UI will listen on all by default interfaces so you can have external inet access.

greg
2017-04-07 15:13
That is the quickstart form.

2017-04-07 15:13
I tried to get all fancy, and confused myself.

2017-04-07 15:13
off to try the quickstart.

greg
2017-04-07 15:13
See what happens . :slightly_smiling_face: I'll be here

2017-04-07 15:13
<3

2017-04-07 15:14
I love the working name "Rocket Skates," btw.

2017-04-07 15:14
Very Wile E. Coyote.

greg
2017-04-07 15:15
That is Victor. I just renamed it last night to be more aligned. :disappointed: still hidden in places though

2017-04-07 15:34
If all my hosts already have public and private networks, then DHCP is for networks that DR will create?

2017-04-07 15:35
--con-dhcp ^

greg
2017-04-07 15:35
well - in this case, more thought is required.

greg
2017-04-07 15:35
To provision the machines, you would need to PXE somehow. Is there a DHCP server already on the internal net.

2017-04-07 15:36
None that I cant switch off.

2017-04-07 15:36
Nope, none.

greg
2017-04-07 15:36
Do the machines PXE on that net?

2017-04-07 15:36
Nope.

greg
2017-04-07 15:36
hmm

2017-04-07 15:37
I already have them all setup with Centos7, networked, all manual.

2017-04-07 15:37
Well, I setup pxe for a little bit, but shut it all down.

greg
2017-04-07 15:37
Soooo - you could set up like a cloud

2017-04-07 15:38
They're fat boxes. Lots of cores, ram, disk.

greg
2017-04-07 15:38
Yeah

greg
2017-04-07 15:39
Ok- so if you look at deploy/add-from-ssh.sh

2017-04-07 15:40
yes

2017-04-07 15:40
I've read through it all.

greg
2017-04-07 15:41
okay the scripts/join_rebar.sh has what you want to do, but it hasn't been used for a while.

greg
2017-04-07 15:41
so it is stale.

2017-04-07 15:42
I see.

2017-04-07 15:42
A freshening would be work.

greg
2017-04-07 15:42
First obvious thing is remove :3000 from everything.

2017-04-07 15:43
Oy, it assumes rebar1

greg
2017-04-07 15:43
well - yeah. -- sorry.

2017-04-07 15:44
:shipit:

2017-04-07 15:44
the quickstart was upset that my internal network is 192.168.1.0 it really wants 99

greg
2017-04-07 15:45
yeah - though it turns out you don't need it because you don't need DHCP

2017-04-07 15:46
What's the "passed in IP?" the node to be added?

greg
2017-04-07 15:47
The IP of the machine itself. This is for multihomed hosts or cloud hosts

greg
2017-04-07 15:47
cloud hosts don't know their public ips in all cases, but need to be registered with it.

2017-04-07 15:47
Ah, where the proper IP cannot be assumed.

2017-04-07 15:48
behind, NAT

greg
2017-04-07 15:48
yes

greg
2017-04-07 15:48
or for AWS they don't know their public ip, but it is the only path in.

2017-04-07 15:49
Sure.

2017-04-07 15:49
I killed all the containers and re-ran the quickstart.

2017-04-07 15:49
Now it's upset that I changed code in the repo. :-)

2017-04-07 15:50
holy cow, SMB is installed?!

greg
2017-04-07 15:51
windows install support.

2017-04-07 15:51
Seems kinda the wrong thing to expose on every interface by default.

greg
2017-04-07 15:52
probably a good request. We are mostly ending up at most places a single interface box that handles this stuff with forwarders to it.

2017-04-07 15:52
understood

2017-04-07 15:52
I'll roll with it for now. :-)

2017-04-07 16:03
running ./join_rebar.sh 192.168.1.10 192.168.1.10 fails

2017-04-07 16:03
I wanna add the admin node as a workload destination. But it seems like a dumb idea, because rebar's taking up so many ports. Should prolly be a VM.

greg
2017-04-07 16:04
yeah

2017-04-08 12:42
I'm trying to make a nice VM for dainty little rebar, but I'm having trouble routing out from the VM. HALP, my head is getting bloody from banging: http://stackoverflow.com/questions/43294216/bridged-libvirt-vms-cannot-route-out

2017-04-08 14:21
Nevermind - seems like I do need the NAT.

2017-04-09 21:41
If I don't see and DHCP subnets in my GUI, did install fail or something?

2017-04-09 22:01
how do I run the role to install raid-tools?

2017-04-09 22:05
Oh, nevermind. :-) I guess I need some nodes first.

2017-04-09 22:58
The hint at the quickstart should probably say if my_key.key should be the public or private key.

2017-04-09 22:59
If I mkdir -p the director path to the keyfile to pre-seed, then quickstart fails on "directory detected"

2017-04-09 23:16
during install, rule-engine:master goes up and down a lot... expected? a793838d8e11 digitalrebar/rule-engine:master "/sbin/docker-entr..." 2 minutes ago Restarting (1) 12 seconds ago compose_rule-engine_1

2017-04-10 01:35
:-)

2017-04-10 01:36
I've got VMs and Metal all booting off my Rebar on a VM.

greg
2017-04-10 02:08
Nice!

2017-04-10 02:09
Yes, happy dance.

2017-04-10 02:10
But I don't have any BMC ports on my metal plugged in.

2017-04-10 02:10
no ipmi.

2017-04-10 02:10
Will rebar fail over to logging into sledgehammer and rebooting?

greg
2017-04-10 02:15
Hmmm - I'm not sure. I think it will attempt to talk. Did you create a BMC network? It may not because there isn't a network to use.

2017-04-10 02:27
Good work @newgoliath !

2017-04-10 02:29
Thanks, @zehicle .

2017-04-10 02:30
@galthaus - I didn't create a BMC network.

2017-04-10 02:36
doing an 'Install OS' just gives a false positive in the Deployment. Sledgehammer never reboots.

2017-04-10 02:37
does it think the machine if off?

2017-04-10 02:37
nope

2017-04-10 02:38
On ready

2017-04-10 02:38
I suspect it's trying to use the BMC to reboot and fails. so it knows the machine is not off

2017-04-10 02:39
I'm not sure where hammer errors would get surfaced. I suspect if you try to power off from the UX or API then you'll get an error message

2017-04-10 02:40
the reinstall combines a few steps together. it's possbile that we're eating the error

2017-04-10 02:43
No response to a "cycle" click.

2017-04-10 02:45
rebooting the sledgehammer via ssh

2017-04-10 02:46
I've got a serial console on it, watching it boot.

2017-04-10 02:46
for your configuration you may need to manually set the nodes to use the SSH hammer.

2017-04-10 02:47
or mess w/ the hardware profiles so that they do not recognize your gear as a known type

2017-04-10 02:52
too bad sledgehammer doesn't send output to the serial console by default. I'd love to watch it boot.

2017-04-10 02:55
rebooting sledgehammer via ssh returns the node up, but no ssh.

2017-04-10 02:56
there's interaction between the provisioner & hammer so that it knows that you've started a reboot cyckle

2017-04-10 02:56
unless nextboot changes, you'll keep going back to sledgehammer

2017-04-10 02:57
sledgehammer is up - but ssh port is closed.

2017-04-10 02:57
host pings ok

2017-04-10 03:00
If I destroy the deployment and send it back to system, you think it will open up ssh?

2017-04-10 03:00
I have no idea - very strange for the image to have ssh

2017-04-10 03:01
It's always reaching back to rebar for instructions, right?

2017-04-10 03:15
@zehicle how do I change to the SSH hammer?

2017-04-10 03:16
I think you can nodes update X set {"hammer":"ssh"} or similar.... checking

2017-04-10 03:22
@zehicle interestingly, it's got "quirks ipmi-dell-dedicated-nic and ipmi-nodelay"

2017-04-10 03:29
hammer is not an attrib.

2017-04-10 04:17
sorry, it's actually a subtype nodes/#/hammers

2017-04-10 04:17
or just api/v2/hammers

2017-04-10 04:17
I think you can delete the offending hammer(s)

2017-04-10 04:18
leave SSH in place (assuming it's there) - @VictorLowther will have better suggestions

2017-04-10 04:19
I seem to have changed it, because there's a whole new complaint:

2017-04-10 04:20
rebar-access role: RuntimeError: Did not create remote_tmpdir on da4-ba-db-3e-93-ac.local.neode.org for some reason! (ssh: connect to host 192.168.1.110 port 22: Connection refused )

zehicle
2017-04-10 04:22
could be a template generation problem

zehicle
2017-04-10 04:23
from the provisioner - these things are all connected together

2017-04-11 17:46
All, we've got the Digital Rebar Provision code ready for community feedback! We'd love for you all to take a look and give it a try. It's just DHCP/PXE intended as a cobbler replacement that will be the v3 DR provisioner.

2017-04-11 17:46
https://robhirschfeld.com/2017/04/11/provision-preview/

wdennis
2017-04-12 22:06
Hey @zehicle - what needs to be done with Provision to get it to use an outboard DHCP server? Same stuff as Cobbler (set relevant next host etc?)

greg
2017-04-12 22:30
Yes - You need to have your DHCP point to the IP of provision as the next server. Specify option 67 as lpxelinux.0 (bootfile)

greg
2017-04-12 22:31
You cna then use --disable-dhcp on the command line when you run dr-provision

2017-04-12 23:09
http://digital-rebar.readthedocs.io/en/latest/deployment/old/external-services.html ctrl+f pxelinux

2017-04-12 23:09
Should that be lpxelinux.0?

2017-04-12 23:10
since I don't see pxelinux.0 in `./.cache/digitalrebar/tftpboot/` and nothing in digitalrebar/digitalrebar that indicates it's used

2017-04-12 23:11
I do notice that discovery is just a symlink to its parent directory though, does it matter if it's specified in the dhcp server's config?

greg
2017-04-12 23:16
difference between dr-provision and digitalrebar.

greg
2017-04-12 23:17
dr-provision's tftpboot directory has it all in the top directory.

2017-04-12 23:18
those are the digitalrebar docs aren't they? I'm using digitalrebar/digitalrebar, not digitalrebar/provision

greg
2017-04-12 23:19
Sorry - Thought you were referring to @wdennis above.

greg
2017-04-12 23:20
@Iae - it does not matter, I think. They both reference the same layout.

greg
2017-04-12 23:20
The main thing is that lpxelinux.0 can get to the pxelinux.cfg/default file or appropriate per node config files

2017-04-12 23:21
ok, but lpxelinux.0 has to be specified in the dhcp server config and not pxelinux.0

greg
2017-04-12 23:29
Yes - the DHCP server needs to have the EXTERNAL_IP of DigitalRebar as the next server in the DHCP response and option 67 (bootfile) needs to be lpxelinux.0 (for legacy BIOS boots).

greg
2017-04-12 23:30
The digitalrebar dhcp server and dr-provision use go template expansion to inspect the incoming packet to determine what file should be sent. This can be seen in the UI if you configure an admin network.

greg
2017-04-12 23:30
Anyway, dinner and soccer for a few hours.

wdennis
2017-04-13 16:41
So, looking to trial Provision on the same server that I have full DR running on -- I've downed DR by running 'docker-compose stop' in the deploy/compose dir - that's all that's needed to bring DR down, correct?

greg
2017-04-13 16:46
yes

wdennis
2017-04-13 16:50
Cool. Now I can curlbash the Provision into the same 'digitalrebar' directory, or should I use another dir above that?

greg
2017-04-13 16:51
different dir

greg
2017-04-13 16:52
use the --isolated flag.

greg
2017-04-13 16:52
it will keep everything in that directory for now.

wdennis
2017-04-13 16:52
Cool, thx

greg
2017-04-13 16:53
np

wdennis
2017-04-13 16:55
Getting a 'No package p7zip-full available' error

greg
2017-04-13 16:55
what OS?

wdennis
2017-04-13 16:55
This on CentOS 7.x

greg
2017-04-13 16:55
try: yum install -y p7zip

wdennis
2017-04-13 16:56
Yup, that worked

greg
2017-04-13 16:56
sigh - it is p7zip-full on ubuntu/debian

greg
2017-04-13 16:56
just p7zip on centos

greg
2017-04-13 16:56
fixing now.

wdennis
2017-04-13 16:57
Love this differing pkg names across distros thing :-/

greg
2017-04-13 16:58
yeah- I did a replace and was too aggressive.

greg
2017-04-13 16:59
in about 15 minutes, all the packages and tree will be updated. It will work. Since you already have it installed now, it should be fine to continue.

wdennis
2017-04-13 16:59
How to continue?

greg
2017-04-13 16:59
Rerun

wdennis
2017-04-13 16:59
Ok

greg
2017-04-13 16:59
It tests for the package and skips if already there.

greg
2017-04-13 16:59
the binary that is.

wdennis
2017-04-13 17:04
Is the '--data-root=' line in readthedocs QuickStart incorrect now?

greg
2017-04-13 17:05
It should be. The install script should have kicked out an example.

wdennis
2017-04-13 17:05
References "discovery-load.sh" instead of "digitalrebar"

greg
2017-04-13 17:05
Line right above it

wdennis
2017-04-13 17:06
It did, and I went with the outputted example

greg
2017-04-13 17:06
okay so is it running? :slightly_smiling_face:

greg
2017-04-13 17:06
export RS_KEY=rocketskates:r0cketsk8ts

greg
2017-04-13 17:06
./drpcli users list

greg
2017-04-13 17:07
./drcpli prefs list

wdennis
2017-04-13 17:07
Yes running

greg
2017-04-13 17:07
Should be able to do:

greg
2017-04-13 17:07
in browser

greg
2017-04-13 17:08
https://<ip>:8092/ui/?token=rocketskates:r0cketsk8ts

wdennis
2017-04-13 17:08
Yup dricli cmds work

wdennis
2017-04-13 17:08
*drpcli

greg
2017-04-13 17:08
then run: tools/discovery-load.sh

greg
2017-04-13 17:09
That will put the pieces in place to do discovery.

wdennis
2017-04-13 17:09
Cool

greg
2017-04-13 17:09
Did the browser thingee work?

wdennis
2017-04-13 17:10
Hold on, let me try

wdennis
2017-04-13 17:12
No, URL not working

wdennis
2017-04-13 17:13
I do see ports 8091/8092 listening on server

greg
2017-04-13 17:13
https://<ip>:8092/swagger-ui

wdennis
2017-04-13 17:14
Nope

wdennis
2017-04-13 17:15
Firewall maybe? Will nmap DR server

wdennis
2017-04-13 17:15
Yup, only SSH port open...

greg
2017-04-13 17:16
okay - are these local firewall rules on your box?

greg
2017-04-13 17:16
Err - dr server?

wdennis
2017-04-13 17:17
Yup - firewalld

greg
2017-04-13 17:17
okay - Note to self update docs to talk about open port reqs.

greg
2017-04-13 17:17
I indirectly list them in the docs, but ...

wdennis
2017-04-13 17:17
Stopped that, will try again

greg
2017-04-13 17:18
you will need to open ports, udp 67, udp 69, tcp 8091, tcp 8092

greg
2017-04-13 17:18
The last two are configurable (well all are configurable, but tftp and dhcp WIGGGGG OUTTTTT if you change then from 69 and 67 respectively).

wdennis
2017-04-13 17:19
Ok, tried swagger, got UI but error on screen "Can't read from server. It may not have the appropriate access-control-origin settings."

greg
2017-04-13 17:19
sigh - need to adjust that. The text box has a URL it defaults to localhost or something like. Put the IP of the Host in that field instead.

wdennis
2017-04-13 17:20
Regular (?) UI page working

greg
2017-04-13 17:20
be sure to add ?token=rocketskates:r0cketsk8ts

wdennis
2017-04-13 17:21
Ok, in

greg
2017-04-13 17:21
There should have been a little dialog for you to enter your username/password or token.

greg
2017-04-13 17:21
It is small. This is working in progress.

wdennis
2017-04-13 17:22
Do I have to declare a subnet if not doing DHCP thru Provision?

greg
2017-04-13 17:22
oh -- yeah.

greg
2017-04-13 17:22
you should be okay.

greg
2017-04-13 17:22
Make sure preferences are set to sledgehammer for default and discovery for unknown.

wdennis
2017-04-13 17:22
So, don't have to, right?

greg
2017-04-13 17:23
correct.

greg
2017-04-13 17:23
You need to have next server point to your IP and bootfile should be lpxelinux.0

wdennis
2017-04-13 17:23
Yup, SH in defaultBootEnv

greg
2017-04-13 17:24
in the bootenvs section discovery, local, and sledgehammer should be available.

wdennis
2017-04-13 17:24
Let me mod my routers DHCP settings

greg
2017-04-13 17:24
Does your DHCP server respond with DNS settings like Domain Name and DNS server?

wdennis
2017-04-13 17:25
Nothing in 'local' showing...

greg
2017-04-13 17:25
Yeah - it should be just available.

wdennis
2017-04-13 17:26
Yes, DHCP server does provide those (it's a pfSense box.)

greg
2017-04-13 17:26
okay - then you are even betterer

greg
2017-04-13 17:26
Another note to self - document external DHCP server case.

greg
2017-04-13 17:26
We need DNS domain name when we make up a name.

greg
2017-04-13 17:30
so - I'm walking you through discovery, but you can also add machine directly as well.

greg
2017-04-13 17:30
If you know that a machine is coming in as an IP. You can create a machine by IP and tell it the bootenv to run when contacted. That works as well.

wdennis
2017-04-13 17:32
Ok, time to boot some bare metal and see what happens :)

greg
2017-04-13 17:33
okay then :slightly_smiling_face:

wdennis
2017-04-13 17:33
When sledgehammer runs, where does it store the discovery info?

greg
2017-04-13 17:34
Okay - so dr-provision is much slimmer. Its purpose is just provisioning and installing OS. It is not meant to inventory and so forth. That is the next step in our integration path. Provision will hook into DigitalRebar for that path.

greg
2017-04-13 17:35
It will create a machine in dr-provision, but it won't have the full inventory data like DR does.

wdennis
2017-04-13 17:35
Ah, I see

wdennis
2017-04-13 17:37
So am waiting on PXE boot now... Dell PE R510 boot is sloooooow

2017-04-13 17:37
Time to feed the :bear:!

wdennis
2017-04-13 17:39
Loading sledgehammer/.... :)

wdennis
2017-04-13 17:40
Hmmm... failed on loading second-stage initramfs

wdennis
2017-04-13 17:41
wget: can't connect to remote host (IP): connection refused

greg
2017-04-13 17:41
Does it say what ip?

greg
2017-04-13 17:41
Also need another param


greg
2017-04-13 17:42
Did you start drprovison with static ip param

wdennis
2017-04-13 17:43
Yes

greg
2017-04-13 17:43
Is 1.148 the drop

wdennis
2017-04-13 17:43
Yes, that's the Provision svr

greg
2017-04-13 17:44
okay

greg
2017-04-13 17:44
why :80? hmm


greg
2017-04-13 17:48
okay in can you cat the tftpboot/pxelinux.cfg/default file

wdennis
2017-04-13 17:50
Does not exist

greg
2017-04-13 17:50
okay - we should have default. That is what is being strange, I think.

greg
2017-04-13 17:51
looking at it now.


greg
2017-04-13 17:52
yeah - I recreated it here. I guess we just broke something. Recently - More unit tests to add.

wdennis
2017-04-13 17:52
Always more tests :)

greg
2017-04-13 17:57
try this:

greg
2017-04-13 17:57
./drpcli prefs set unknownBootEnv ignore

greg
2017-04-13 17:57
./drpcli prefs set unknownBootEnv discovery

greg
2017-04-13 17:58
See if that creates the file.

wdennis
2017-04-13 18:00
Nope

greg
2017-04-13 18:00
ok


greg
2017-04-13 18:18
I'm silly - it isn't rendered that way anymore .

greg
2017-04-13 18:18
sigh


greg
2017-04-13 18:24
It think I see the busted-ness. We switched to inline rendering and it isn't working on a shift.

greg
2017-04-13 18:24
We don't use the filesystem for content anymore. Should add preview for bootenvs and templates.

greg
2017-04-13 18:26
NOtes So far: 1. Document Port requirements for DRP 2. Document DHCP requirements when not using DRP as DHCP (option 15 (discovery), bootfile, nextserver). 3. Bug in changing unknown bootenv and rendering content. Changed ignore to discovery and ignore is still being served.

wdennis
2017-04-13 18:30
So, at this moment, busted & cant continue?

greg
2017-04-13 18:32
hmm - try this.

greg
2017-04-13 18:33
``` ./drpcli prefs list ./drpcli prefs set unknownBootEnv ignore curl http://127.0.0.1:8091/pxelinux.cfg/default ./drpcli prefs list ./drpcli prefs set unknownBootEnv discovery curl http://127.0.0.1:8091/pxelinux.cfg/default```

greg
2017-04-13 18:44
The discovery file should look like this: ``` DEFAULT discovery PROMPT 0 TIMEOUT 10 LABEL discovery KERNEL sledgehammer/708de8b878e3818b1c1bb598a56de968939f9d4b/vmlinuz0 INITRD sledgehammer/708de8b878e3818b1c1bb598a56de968939f9d4b/stage1.img APPEND rootflags=loop root=live:/sledgehammer.iso rootfstype=auto ro liveimg rd_NO_LUKS rd_NO_MD rd_NO_DM provisioner.web=http://127.0.0.1:8091 rs.api=https://127.0.0.1:8092 IPAPPEND 2 ```

greg
2017-04-13 18:45
The 127.0.0.1 is because my request came from 127.0.0.1

greg
2017-04-13 19:00
I'm now wondering if the tftp error was an issue. You could also make sure that your pref is discovery and reboot.

greg
2017-04-13 19:00
See if we hit the tftp error again.

wdennis
2017-04-13 19:33
@wdennis uploaded a file: https://rackn.slack.com/files/U416T0AAX/F4ZP4RLRM/image_uploaded_from_ios.jpg and commented: Looks good?

greg
2017-04-13 19:35
yeah if provisioner.web=http://127.0.0.1:8091 on the second to last line

wdennis
2017-04-13 19:36
Should that not be the public IP?

greg
2017-04-13 19:36
We build the reply based upon the dest IP.

greg
2017-04-13 19:36
run it with your node's iP.

greg
2017-04-13 19:36
curl http://node's IP:8091/pxelinux.cfg/default

greg
2017-04-13 19:37
it should pop out with a different IP.

wdennis
2017-04-13 19:40
Node being the Provision server, or the client server?

greg
2017-04-13 19:40
provisioner server

wdennis
2017-04-13 19:41
Ok, did change the URL

greg
2017-04-13 19:42
Reboot the node and look for the tftp error in the drp log

wdennis
2017-04-13 20:07
The PXE client boot did fail again in the same way- that was expected?

greg
2017-04-13 20:29
hmm - I guess it is good, but was hoping not.

greg
2017-04-13 20:34
sorry for the delay. @wdennis - can you run a command on the system stuck in ash

greg
2017-04-13 20:34
cat /proc/cmdline

greg
2017-04-13 20:34
grep -o 'provisioner.web=[^ ]*' /proc/cmdline

wdennis
2017-04-13 20:36
Will do when I can in a bit - taking kidddo to karate :)

greg
2017-04-13 20:36
np

wdennis
2017-04-14 00:31

wdennis
2017-04-14 00:32
here you go, @greg

greg
2017-04-14 00:35
Thx. If you added /24 to the command line of drprovison please stop the server remove it and restart.

wdennis
2017-04-14 00:38
I did? will do.

greg
2017-04-14 00:42
Another thing to add to the checks on start up :slightly_smiling_face:

wdennis
2017-04-14 00:46
did the curlbash output say to put ?/24? at the end of the IP, or did I just goof?

wdennis
2017-04-14 00:46
readthedocs site does not have that

greg
2017-04-14 00:59
It should not. The readthedocs should not have it either. I bet it might have. My bad.

wdennis
2017-04-14 01:03
Unfortunately, I typed in something in ash that caused it to coredump? And I don?t have remote power control on that server I was using :disappointed:

wdennis
2017-04-14 01:04
hangs head in shame

wdennis
2017-04-14 01:14
This look like the problematic bit in `install.sh`:

wdennis
2017-04-14 01:14
```[dradmin@dr-admin ~]$ /sbin/ip -o -4 addr show scope global |head -1 |awk '{print $4}' 192.168.1.148/24```

greg
2017-04-14 01:14
Yeah I lifted to much code. Argh.

wdennis
2017-04-14 01:15
That?s why the community needs to test :slightly_smiling_face:

greg
2017-04-14 01:16
Well thanks and sorry.

wdennis
2017-04-14 01:16
np

wdennis
2017-04-14 01:17
Unfortunately unless I run into work (~15 min drive) the game?s over until Mon? We?ll see what the wife thinks?

wdennis
2017-04-14 01:18
There?s a bit of other work (wiring work) I could stand to do there?

greg
2017-04-14 01:18
I'm in no hurry. Marital harmony is important.

wdennis
2017-04-14 01:19
yes it is :slightly_smiling_face:

wdennis
2017-04-14 01:20
So may I ask what?s involved with installing a distro image and doing kickstart/preeseed?

greg
2017-04-14 01:20
Cd assets

greg
2017-04-14 01:21
Drpcli bootenvs install bootenvs/ubuntu16.04.yml

greg
2017-04-14 01:21
Wait awhile

greg
2017-04-14 01:22
Then change machines bootenv to Ubuntu. And then reboot node

greg
2017-04-14 01:22
Wait Awhile. When machine bootenv goes to local done

greg
2017-04-14 01:22
The yemplates and bootenvs work mostly like DR

greg
2017-04-14 01:23
Please forgive phone typing

greg
2017-04-14 01:24
Examples in assets dir


wdennis
2017-04-14 13:47
And I see the machine in the UI

wdennis
2017-04-14 13:50
API Help links in the UI are broken...


greg
2017-04-14 13:54
Yeah!!!!!

2017-04-14 13:55
I'll fix those links! Thanks.

wdennis
2017-04-14 14:14
Now into the Ubuntu 16.04 install...

wdennis
2017-04-14 14:19
Looks like we have a successful installation :)

greg
2017-04-14 14:19
Nice!

wdennis
2017-04-14 14:20
When the machine boots post-install, does it communicate back to Provision and set BootEnv to be 'local'?

greg
2017-04-14 14:20
yes

greg
2017-04-14 14:21
no - local is set as last step of install in ks or preseed

greg
2017-04-14 14:21
ignore the yes

wdennis
2017-04-14 14:22
Yup, I see local now

wdennis
2017-04-14 14:23
Cool

greg
2017-04-14 14:23
I think for ubuntu - rocketskates/r0cketsk8ts

greg
2017-04-14 14:23
default login.

wdennis
2017-04-14 14:23
Not the 'rebar' one?

greg
2017-04-14 14:24
I think I changed it.

greg
2017-04-14 14:24
You can change it by add params (either to the machine or globally).

greg
2017-04-14 14:24
Each bootenv defines a set of parameters that can be injected.

greg
2017-04-14 14:24
some are required and some are optional. These are optional (with defaults).

greg
2017-04-14 14:25
You can specify the user (for ubuntu) and the password hash.

greg
2017-04-14 14:25
Those are next on my doc list.

wdennis
2017-04-14 14:25
Where to set param's?

greg
2017-04-14 14:25
drpcli params list

greg
2017-04-14 14:25
those are global (it is empty by default).

greg
2017-04-14 14:26
drpcli machines update <uuid> <json blob of parameters>

greg
2017-04-14 14:26
I should make that better one day soon.

wdennis
2017-04-14 14:26
Cannot seem to log in with the rocketskates user/passwd (or the rebar one)

greg
2017-04-14 14:27
rebar/rebar1

greg
2017-04-14 14:27
:slightly_smiling_face:

greg
2017-04-14 14:27
I don't remember - I'll have to check.

wdennis
2017-04-14 14:28
Nope :)

greg
2017-04-14 14:28
What stupid thing did I do.

greg
2017-04-14 14:29
rocketskates is the user

greg
2017-04-14 14:29
rocketskates

greg
2017-04-14 14:30
r0cketsk8ts

wdennis
2017-04-14 14:31
Does not work...

greg
2017-04-14 14:31
okay sigh arg

greg
2017-04-14 14:32
Time to fix it and put it in the docs.

greg
2017-04-14 14:37
Just a second


wdennis
2017-04-14 14:37
Where do I see the .Param 's used?

greg
2017-04-14 14:39
in this case, it is unset and uses the default in the file.

greg
2017-04-14 14:40
```./drpcli params create '{ "Name": "provisioner-default-password-hash", "Value": "$6$5trJ0SAGobo9YE1X$r31iOEokINeaYTCbtyfpCZU6wgKoK7Tr3mc7CIFl7hDPP6LNPkUveg3hB2vasE.H5IBwvW5qBK8aQz5imnp8J0" }'```

greg
2017-04-14 14:40
That will create a new global parameter which represents r0cketsk8ts

greg
2017-04-14 14:40
You could try: R0cketSk8ts

greg
2017-04-14 14:41
Anyway, the parameter call will set the global.

wdennis
2017-04-14 14:41
Nada

greg
2017-04-14 14:42
then do: drpcli machines update <uuid> '{ "BootEnv": "ubuntu-16.04-install" }'

greg
2017-04-14 14:42
reboot the node and then log in when it is done.

greg
2017-04-14 14:43
Wait

wdennis
2017-04-14 14:43
Oh that'll make the node do a reinstall?

greg
2017-04-14 14:43
RocketSkates is the passsword

greg
2017-04-14 14:43
Yes

wdennis
2017-04-14 14:43
Yuuuuup :)

greg
2017-04-14 14:43
Okay - I'm gonna doc that - it was in our git commit log history.

wdennis
2017-04-14 14:44
Lol

wdennis
2017-04-14 14:46
OK, this is good... Now, how to spec differing preseed templates for different nodes?

wdennis
2017-04-14 14:47
We have groups that want specific disk partitioning, etc

greg
2017-04-14 14:48
YOu need custom bootenvs for each one.

greg
2017-04-14 14:48
So you would create template for each type/group - then a bootenv the refs that template. Then set node's bootenv to that bootenv.

wdennis
2017-04-14 14:49
I try to do as little as possible in kickstart/preseed, and really config node via Ansible, but disk partitioning isn't something that can. E done that way :)

greg
2017-04-14 14:49
Yep - our philosophy as well.

wdennis
2017-04-14 14:50
Any ideas about pulling in config blocks within templates?

greg
2017-04-14 14:50
Well , two thoughts. One is parameters with inject.

wdennis
2017-04-14 14:50
Like you can still use the same basic template, but the disk partitioning can be a sub-template if you will...

greg
2017-04-14 14:51
In theory, the templates can contain other templates (from a golang template spec perspective, but I don't think we've hooked that together). So one day that way.

wdennis
2017-04-14 14:51
Also would handle multi-disk setups

greg
2017-04-14 14:51
Today, you could build a bootenv with a big string explosion and put that as a parameter. The parameter could then be injected by node type.

wdennis
2017-04-14 14:53
Define "big string explosion" please :)

wdennis
2017-04-14 14:54
I'd hate to have to make/maintain 'n' templates that are basically 90%+ same, with just minor changes

greg
2017-04-14 14:55
give me a second

wdennis
2017-04-14 14:55
Cobbler has the concept of 'snippets' that are config blobs that can be put into a static template kickstart

greg
2017-04-14 14:56
Yeah - template in template will be the way of that.

greg
2017-04-14 14:56
We just aren't quite there yet.

greg
2017-04-14 14:57
Instead you would need this:

greg
2017-04-14 14:58
Say I had this in my preseed:

greg
2017-04-14 14:58
```d-i clock-setup/ntp boolean false {{end}} {{if .ParamExists "operating-system-disk"}} d-i partman-auto/disk string {{ .Param "operating-system-disk" }} {{else}} d-i partman-auto/disk string /dev/sda {{end}} d-i partman-auto/method string lvm d-i partman-lvm/device_remove_lvm boolean true d-i partman-lvm/device_remove_lvm_span boolean true d-i partman-auto/purge_lvm_from_device boolean true d-i partman-md/device_remove_md boolean true d-i partman-lvm/confirm boolean true d-i partman-lvm/confirm_nochanges boolean true d-i partman-lvm/confirm_nooverwrite boolean true d-i partman-auto-lvm/guided_size string max d-i partman-auto-lvm/new_vg_name string {{ .Machine.ShortName }} d-i partman-auto/choose_recipe select custom_lvm d-i partman/confirm_write_new_label boolean true d-i partman/choose_partition select finish d-i partman/confirm boolean true d-i partman/confirm_nooverwrite boolean true d-i partman/auto expert_recipe string \ custom_lvm:: \ 500 50 1024 free $iflabel{ gpt } $reusemethod{ } method{ efi } format{ } . \ 128 50 256 ext2 $defaultignore{ } method{ format } format{ } use_filesystem{ } filesystem{ ext2 } mountpoint{ /boot } . \ 10240 20 10240 ext4 $lvmok{ } mountpoint{ / } lv_name{ root } in_vg{ {{ .Machine.ShortName }} } method{ format } format{ } use_filesystem{ } filesystem{ ext4 } . \ 50% 20 100% linux-swap $lvmok{ } lv_name{ swap } in_vg{ {{ .Machine.ShortName }} } method{ swap } format{ } . {{if (and (eq "ubuntu" .Env.OS.Family) (lt "12.10" .Env.OS.Version))}} d-i live-installer/net-image string {{.Env.InstallUrl}}/install/filesystem.squashfs```

greg
2017-04-14 14:58
I could replace it with this:

greg
2017-04-14 15:00
```d-i clock-setup/ntp boolean false {{end}} {{ .Param "my-disk-layout-string" }} {{if (and (eq "ubuntu" .Env.OS.Family) (lt "12.10" .Env.OS.Version))}} d-i live-installer/net-image string {{.Env.InstallUrl}}/install/filesystem.squashfs```

greg
2017-04-14 15:00
Then do:

greg
2017-04-14 15:01
./drpcli params create '{ "Name": "my-disk-layout-string", "Value": "custom stuff here - no params though" }'

greg
2017-04-14 15:01
That would be the default

greg
2017-04-14 15:01
Then for each machine that needed custom from default.

greg
2017-04-14 15:02
./drpcli machines update <uuid> '{ "Params": { "my-disk-layout-string": "other custom form" } }'

greg
2017-04-14 15:02
This is giving me some good ideas.

greg
2017-04-14 15:02
for some features.

greg
2017-04-14 15:03
make those commands easier

greg
2017-04-14 15:03
template in template

greg
2017-04-14 15:03
docs

wdennis
2017-04-14 15:07
Totally should work on that Tmpl-in-tmpl stuff :)

wdennis
2017-04-14 15:08
And all this work will port into future DR?

greg
2017-04-14 15:09
Yeah - eventually, the plan is to have this replace the DHCP/Provisioner in DR with a real provider.

greg
2017-04-14 15:09
The hope is to start with this to get moving - it should be simpler to start using and it will work on multiple sites without the back haul inet perf/bandwidth hit.

greg
2017-04-14 15:10
Local control and all that.

wdennis
2017-04-14 15:10
It's the gateway drug for full-on DR ;)

greg
2017-04-14 15:11
well - we dream

wdennis
2017-04-14 15:32
So the templates are Golang "text/template" ?

greg
2017-04-14 15:39
yes

greg
2017-04-14 15:43
@wdennis - you want something like:

greg
2017-04-14 15:44
{{ .Template (.Param "disk-template") }}

greg
2017-04-14 15:44
Where .Param would be a string parameter that is the name of a template to include in this template.

wdennis
2017-04-14 15:45
Can the resulting master template be rendered to see what it will produce? (Not a Go dev...)

greg
2017-04-14 15:45
Yeah - that was another idea that I was thinking about. A template preview option.

greg
2017-04-14 15:46
bootenv preview option as well.

wdennis
2017-04-14 15:46
Great idea

greg
2017-04-14 15:50
You can open issues in github for things you see as enhancements as well. I'm doing these now.

wdennis
2017-04-14 15:50
Yup, will do

greg
2017-04-14 16:04
okay - I think I created the ones that we talked about.

2017-04-14 18:57
Is there public documentation for connecting the docker dhcp container to the "Real" network so it can get and response to helper requests hitting the routable ip already? Easy as just making docker bridge in to "eth0" on the host?

greg
2017-04-14 19:00
If you are using FORWARDER mode, yes, you can bridge the eth1 into docker0. Then the 192.168.124.0/24 network will be served on that interface as well.

2017-04-14 19:01
Running host mode in theory. forwarder simpler?

greg
2017-04-14 19:01
If you are using HOST mode, the DHCP server is already listening on the interfaces. You may need to create a specific admin network for that interface.

2017-04-14 19:01
I see the host binds on the host

2017-04-14 19:01
udp 0 0 192.168.122.1:53 0.0.0.0:_ 1400/dnsmasq udp 0 0 0.0.0.0:67 0.0.0.0:_ 1400/dnsmasq

2017-04-14 19:01
k

2017-04-14 19:01
specifically bind it to "ens3" ?

greg
2017-04-14 19:02
Actually, you just have to create the network in the UI or CLI.

greg
2017-04-14 19:03
with the proper ranges and it will just start listening on that interface.

greg
2017-04-14 19:03
That may not make complete sense.

2017-04-14 19:03
What does DR consider proper ranges? maybe thats the issue

2017-04-14 19:03
does it support dhcp-relay with "External" ranges?

greg
2017-04-14 19:03
Yes

greg
2017-04-14 19:04
It will handle both.

2017-04-14 19:04
hm. the default admin-internal scope is there, I'll fix it up

greg
2017-04-14 19:04
OR create a new one.

greg
2017-04-14 19:04
either way.

greg
2017-04-14 19:04
make sure to get the router at the bottom of the page

2017-04-14 19:05
Is "conduit" the key field that matters

2017-04-14 19:05
not really clear

2017-04-14 19:06
or enable bridge/put in ens3

greg
2017-04-14 19:07
conduit is an abstraction for that.

greg
2017-04-14 19:07
and it is for the client (not the server).

greg
2017-04-14 19:07
So, it is used to force the network to a specific interface on the client.

greg
2017-04-14 19:07
For admin, you can use the dhcp conduit.

greg
2017-04-14 19:08
it will put the network on the interface that dhcp booted/

2017-04-14 19:08
k

greg
2017-04-14 19:08
You can also specify 1g0 or 10g2 or ... a speed and an index.

greg
2017-04-14 19:08
1g0 = first 1g capable interface.

2017-04-14 19:11
Is there something to tickle to make the new network show up in dhcp subnets

greg
2017-04-14 19:12
refresh the UI. that page doesn't always reload.

greg
2017-04-14 19:12
Also. make sure you gave the category in the network as admin.

greg
2017-04-14 19:12
group should be something else.

2017-04-14 19:12
didn't :) changing

2017-04-14 19:13
shows now

greg
2017-04-14 19:13
:slightly_smiling_face:

2017-04-14 19:14
would the dhcp container be the appropriate place to do a tcpdump

greg
2017-04-14 19:15
in host mode, no. The dhcp container is a host networking container. So the host is sufficient.

greg
2017-04-14 19:15
you can also so the DHCP logs

2017-04-14 19:15
yea, those are clean/quiet

greg
2017-04-14 19:15
cd digitalrebar/deploy/compose

greg
2017-04-14 19:16
docker-compose logs -f dhcp

greg
2017-04-14 19:16
okay

2017-04-14 19:16
Looks like all api requets

2017-04-14 19:16
requests*

greg
2017-04-14 19:17
okay - so - we aren't seeing the DHCP requests.

greg
2017-04-14 19:17
hmm

2017-04-14 19:17
how do I verify this is host mode

greg
2017-04-14 19:17
tcpdump or the log

2017-04-14 19:17
this was my deploy cli ./run-in-system.sh --deploy-admin=local --access=host --admin-ip=10.62.7.140/26 --con-dhcp --con-provisioner

2017-04-14 19:17
but it still sat forever on the ansible forwarder task

2017-04-14 19:18
skipped host

greg
2017-04-14 19:18
host should be HOST

greg
2017-04-14 19:18
you are in forwarder mode. We have silly args.

2017-04-14 19:18
case check?

2017-04-14 19:18
lol k

2017-04-14 19:18
rerolling, that makes sense for what I'm seeing then :)

greg
2017-04-14 19:19
Yes, please.

2017-04-14 19:19
tx rob, will bbiab

greg
2017-04-14 19:19
really - this is Greg. :slightly_smiling_face: our slack to gitter app bounces through rob.

2017-04-14 19:20
AH woops lol

greg
2017-04-14 19:20
np - all good.

2017-04-14 19:29
dhcp_1 | 2017/04/14 19:28:47 Recieved DHCP packet: type Discover xid 0x8 ciaddr 0.0.0.0 yiaddr 0.0.0.0 giaddr 10.62.0.1 chaddr 70:10:6f:bc:4f:4c much bettah

2017-04-14 19:29
tx again

greg
2017-04-14 19:29
cool

2017-04-14 19:39
cool @crafty_house_twitter

2017-04-14 22:21
newly deployed digitalrebar system, working with crafty. we are not seeing the nodes added via pxe after dhcp subnet was created. any ideas. nothing in the logs but this:

2017-04-14 22:21
dhcp_1 | 2017/04/14 19:25:30 Recieved DHCP packet: type Discover xid 0xa1be8f5e ciaddr 0.0.0.0 yiaddr 0.0.0.0 giaddr 10.62.16.1 chaddr 9c:dc:71:4a:fa:c0 dhcp_1 | 2017/04/14 19:25:30 xid 0xa1be8f5e: Starting processing: 2017-04-14 19:25:30.948700271 +0000 UTC dhcp_1 | 2017/04/14 19:25:30 xid 0xa1be8f5e: Config lock acquired: 2017-04-14 19:25:30.948703801 +0000 UTC dhcp_1 | 2017/04/14 19:25:30 xid 0xa1be8f5e: relay from 10.62.16.1 (hops 1) dhcp_1 | 2017/04/14 19:25:30 Discover xid 0xa1be8f5e: No subnet for leases

2017-04-14 22:22
any ideas...

greg
2017-04-15 03:59
Is your admin network ranges include in 10.62.16.1?

greg
2017-04-15 03:59
@csayler

2017-04-15 22:11
Just try out the standalone Provision. Looks great.

2017-04-15 22:12
I was able to pxe boot a machine, now I was wondering if the machine should appear in machines list (drpcli machines list) ?

2017-04-16 08:29
@zehicle I'm trying to test drp and i blocked, when i try to write the user and passwd. i put rocketskates:RocketSkates And nothing happens. No more login.

2017-04-16 08:44
Sorry, it works with https://<ip>:8092/ui/?token=rocketskates:r0cketsk8ts .merci

greg
2017-04-16 11:52
@chilicat - what did you boot? Did you boot ignore? or Discovery? If discovery, then yes. If ignore, then no.

greg
2017-04-16 11:52
@moula - yeah - one is for dr-provision and one is for the machines that dr-provision installs.

greg
2017-04-16 11:53
More docs to come on how to change both of those.

2017-04-16 15:43
Thanks, go it working. Is it actually possible to trigger a reboot via cli of the machine?

2017-04-16 17:11
@zehicle Thank's .

zehicle
2017-04-16 18:36
@chillicat, in full DR yes. There is a node X reboot option. In DRP, no.

2017-04-16 20:59
@chilicat specially from the rebar cli -> "rebar nodes poweractions [node]" will give you the options

2017-04-16 20:59
then "rebar nodes power [node] [action]" will take that action. e.g.: reboot is the most common because it works even if we don't have OOB control.

2017-04-16 21:00
if you are looking to wipe and reconfigure the node, that's "rebar nodes redeploy [node]" or you can do a match of them using the deployment redeploy.

2017-04-17 14:01
@zehicle it's not possible to reboot a physical node after detect it, to do install system!!! i do it with maas and cloudinit.

2017-04-17 14:03
@zehicle another thing, if i shutdown my drp server, how i can be enable it automatically? thank you.

greg
2017-04-17 14:48
@moula - this is @galthaus - That is the point of dr-provision. It is simple and quick. DigitalRebar provides that level of integration today.

greg
2017-04-17 14:48
@moula - run the install.sh without the isolated flag and it will place start up scripts in place, put binaries in place, and show you the service commands to enable and start dr-provision to survive reboots.

wdennis
2017-04-17 14:49
DR team: in DRP, where can I set params for a host? I see there's a "OptionalParams" section in a given bootenv, but can these be set on a per-host basis?

greg
2017-04-17 14:49
@wdennis - they are only set on a per host or global level. I feel another helper coming on.

greg
2017-04-17 14:49
drpcli params is for manipulating global parameters

greg
2017-04-17 14:50
drpcli machines update <uuid> '{ json blob of params}' is for machine specific params

wdennis
2017-04-17 14:50
Thought so - if I do 'drpcli params list' I get empty set

greg
2017-04-17 14:51
Yes - params are free form.

greg
2017-04-17 14:51
They are deep dictionaries.

greg
2017-04-17 14:51
string -> Struct

greg
2017-04-17 14:51
where struct can be a json object, string, int, bool, ...

wdennis
2017-04-17 14:52
I think that params should be perhaps site-specific, apply to a group of hosts

greg
2017-04-17 14:52
For example, you can inject ssh keys by params. They need to be an array of objects. I'll be documenting this.

wdennis
2017-04-17 14:52
Like ntp_servers for example

greg
2017-04-17 14:52
@wdennis - that is where DR comes in. We are trying to give some function but level grouping and higher order ops to DR.

greg
2017-04-17 14:53
Well - long term vision is many dr-provision units local to regions to handle bandwidth and locality issues with a central DR driving them.

wdennis
2017-04-17 14:53
Ah

wdennis
2017-04-17 14:55
Any docs yet on providing params on a per-host basis?

greg
2017-04-17 14:56
of course not- I think we may be exposing a little too fast, but that is next on my plate. I have an internal thing I'm working on, but those docs and flows are next.

greg
2017-04-17 14:56
:slightly_smiling_face:

wdennis
2017-04-17 14:56
Trying to get to a point where I can substitute DRP for Cobbler in my environment, but missing some functionality (like profiles for groups of servers)

greg
2017-04-17 14:57
that is what we want and I need to help with that.

wdennis
2017-04-17 14:57
I could live with per-host for now

greg
2017-04-17 14:58
how do you specify the group membership and group scope?

greg
2017-04-17 14:59
need to drive for a bit. Back in a little while.

wdennis
2017-04-17 15:01
There would have to be something like profiles in DRP that combine a bootenv, and associated templates, and then instead of defining bootenv per host, can specify a template

wdennis
2017-04-17 15:03
Like for instance in my environment, I have many Ubuntu 16.04 profiles that only differ in things like disk partitioning, root user enabled or not, default user + passwd

wdennis
2017-04-17 15:03
95%+ of the OS install is the same, but these are crucial differences

wdennis
2017-04-17 15:05
(Maybe wrong here) in DRP, a bootenv is a single OS flavor installer map - don't want to have to have many of these to handle the >5% differences

wdennis
2017-04-17 15:06
Sorry for the tidal wave of txt here - do concentrate on driving :)

zehicle
2017-04-17 15:26
@wdennis the advanced workflow and system profiles are already in full DR. DRP is intended to be narrrowly scoped and then driven by higher level services.

greg
2017-04-17 15:32
Acutally, I think we need to document bootenvs better.

greg
2017-04-17 15:32
If we add template in template (with parameter drivers and examples), I think you have everything you need and more.

wdennis
2017-04-17 15:32
@zehicle It seems to me the value in full-on DR is to take advantage of the DR-provided orchestration of the full OS + software stack (which in something like K8s or OpenStack is quite complex.) What about use case of just install OS + use existing automation developed by end-user to get machines ready for further use (much simpler use case?)

greg
2017-04-17 15:32
The variable expansion into something more than just string replacement is pretty powerful.

wdennis
2017-04-17 15:34
Are bootenvs supposed to be like profiles?

greg
2017-04-17 15:36
bootenvs are intended to be a bootable environment that can be customized per machine.

greg
2017-04-17 15:37
This includes local disks, discovery, installation, coreos diskless, burn-in testing, whatever.

greg
2017-04-17 15:37
We intend the bootenvs to be customizable through parameters. The complexity of the templates contained in the bootenv and how they are tied to the parameters drive the complexity and the mutability of the bootenv.

greg
2017-04-17 15:38
Like you could have no params and hardcode an install. This would generate nxN bootenvs for your description.

greg
2017-04-17 15:39
Another level is to create a bootenv that takes a param that defines user info. The user info is an array of user data objects (object is uname, passhash, groups, ... whatever). Use golang template expanders to parse that list into user names.

greg
2017-04-17 15:39
The next level up is to put that into a template that preseed/kickstart template references.

greg
2017-04-17 15:40
At that point, you can set a variable on the machine that says complex_users= true and user_table=[ {}, {}, ... ]

greg
2017-04-17 15:40
This is what I need to spend time documenting.

2017-04-17 15:55
@zehicle I will try it, Merci beaucoup.

2017-04-17 15:57
@galthaus sorry . merci.

greg
2017-04-17 15:57
@moula - np.

wdennis
2017-04-17 20:13
Some DRP questions relating to host config: - How to spec initial user/passwd other than default "rocketskates" one - How to set static IP / mask / gateway - How to insert desired pubkey in root's .ssh/authorized_keys - If desired, how to set root password, and allow root access via SSH

wdennis
2017-04-17 20:13
Can these sorts of things be done?

greg
2017-04-17 20:15
well - yes. Some require more work than others. For example, static IP. dr-provision isn't managing address spaces beyond simple DHCP (and in your case not at all). The machine has an address field that can be referenced in a template to then set a static IP.

greg
2017-04-17 20:15
that way. You would have to specify the gateway and mask as new parameters.

wdennis
2017-04-17 20:47
@greg Are any of these handled thru just setting properties (params), or all take template changes?

greg
2017-04-17 20:48
most are already handled with parameters today

greg
2017-04-17 20:49
everything but static ip config is handled through parameters today.

greg
2017-04-17 20:49
and maybe root access via ssh.

wdennis
2017-04-18 01:09
@greg, any docs on setting the above params? (sorry, haven?t looked - point me to the URL if they exist plz)

greg
2017-04-18 01:19
I'm writing them tonight

wdennis
2017-04-18 01:24
Thanks - looking forward to implementing them :slightly_smiling_face:

greg
2017-04-18 06:25
@wdennis - check readthedocs with the latest (tip isn't working right). Sigh. http://provision.readthedocs.io/en/latest/doc/arch/data.html#template

greg
2017-04-18 06:26
More to come. Sorry. Hopefully tomorrow.

wdennis
2017-04-18 17:11
Don't be sorry - this looks great. Good docs take time (but do take the time! ;)

wdennis
2017-04-18 17:12
Time to get the lab together & start playing :)

greg
2017-04-18 17:27
More coming - objects mostly complete. Next up is operations and cli examples.

vlowther
2017-04-19 02:32
@wdennis templates including templates probably works now. Still need to write unit tests for it.

vlowther
2017-04-19 02:33
{{template "other" .}} Is the standard go text/template tag to use. https://golang.org/pkg/text/template/#pkg-overview

greg
2017-04-19 02:46
:slightly_smiling_face: And I need to remove the explicit line in our docs saying don't do it.

2017-04-19 23:01
``` labs-rebar:~/digitalrebar/deploy$ docker pull digitalrebar/logging:production Error response from daemon: manifest for digitalrebar/logging:production not found ``` I'm not seeing this tag on docker hub, and it looks like `run-in-system.sh` is failing when it tries to pull this

2017-04-19 23:06
never mind this

zehicle
2017-04-20 00:01
glad you got past it - let us know if you have other questions

2017-04-20 00:36
`Error: assignment to undeclared variable raw $scope.rawProfiles@https://labs-rebar.###/ux/bundle.min.js:39:8253 this.showEditNodeDialog@https://labs-rebar.###/ux/bundle.min.js:39:9149 anonymous/fn@https://labs-rebar.###/ux/bundle.min.js line 4 > Function:2:471 b@https://labs-rebar.###/ux/bundle.min.js:2:27522 Ic[b]</<.compile/</</e@https://labs-rebar.###/ux/bundle.min.js:5:4794 sf/this.$get</n.prototype.$eval@https://labs-rebar.###/ux/bundle.min.js:3:5260 sf/this.$get</n.prototype.$apply@https://labs-rebar.###/ux/bundle.min.js:3:5492 Ic[b]</<.compile/</<@https://labs-rebar.###/ux/bundle.min.js:5:4844 Pf@https://labs-rebar.###/ux/bundle.min.js:1:18758 Of/d@https://labs-rebar.###/ux/bundle.min.js:1:18707` I'm actually getting this error when I try to go edit a node

2017-04-20 00:38
as well as HSTS warnings but I don't think those are relevant

2017-04-20 00:53
I guess this is happening in firefox only

2017-04-20 00:53
surf/webkit seemed fine

2017-04-20 00:55
nvm, it's happening in other browsers, too

greg
2017-04-20 02:21
yes - we need a container rebuild for that. I think.

greg
2017-04-20 02:21
You can change ux to ux-dev and it will work around it, I believe.

2017-04-20 02:26
Yeah, I tried ux-dev and was still running into the same issue

2017-04-20 02:27
but also, question, does sledgehammer/the discovery image wipe out disks?

2017-04-20 02:27
or partition tables at least

greg
2017-04-20 02:35
yes as part of the OS provisioning step

2017-04-20 14:28
welp

2017-04-20 14:28
that makes it harder to introduce in existing environments

2017-04-20 14:29
(we used to just wipe partition tables during preseed/kickstart)

greg
2017-04-20 14:44
Only as part of a os install request

greg
2017-04-20 14:45
It is part of a role that runs in sledgehammer Sledgehammer itself doesn't

2017-04-20 14:49
hm, this is weird then

greg
2017-04-20 14:51
what are you seeing?

greg
2017-04-20 14:54
It is part of the provisioner-os-install role - see core/script/roles/provisioner-os-install/01-install-os.sh

2017-04-20 14:55
```[root@d00-25-90-59-d5-62 ~]# parted -s /dev/sda print Error: /dev/sda: unrecognised disk label Model: LSI Logical Volume (scsi) Disk /dev/sda: 2396GB Sector size (logical/physical): 512B/512B Partition Table: unknown Disk Flags: ``` two servers that I guess their owner rebooted for whatever reason look like this after booting up the discovery image - but I do see other servers that booted up into discovery that still have their partition tables intact

greg
2017-04-20 14:57
Does it have a LSI raid controller? Do you have any raid configured? Did you drive that system to any state other than just being discovered?

2017-04-20 14:57
yes, it does have an LSI raid controller - and no, I just let it be discovered

greg
2017-04-20 14:57
so - it should have touched anything.

greg
2017-04-20 14:58
should not

2017-04-20 14:58
I'm going to try to see if I can reproduce it somehow

greg
2017-04-20 14:59
oh - does the system have multiple disks? Could they be enumerating differently?

2017-04-20 15:00
the ones that are affected only have one disk configured on the RAID controller

wdennis
2017-04-20 16:03
@greg - why does DRP control the Ubuntu apt sources instead of using std upstream Ubuntu ones?


2017-04-20 16:06
that's the repo extracted from ubuntu ISOs isn't it

greg
2017-04-20 16:07
yes

greg
2017-04-20 16:07
It tries to keep traffic local to the ISO if possible.

wdennis
2017-04-20 16:26
What about after install? Shouldn't there be a switch to std Ubuntu repos?

wdennis
2017-04-20 16:27
The ISO one doesn't have security updates, or universe

greg
2017-04-20 16:27
Interesting. I thought we left them in post install. It is easy enough to change.

wdennis
2017-04-20 16:28
Found this out when I went to install some sw tools I usually put in right after install (ethtool, tree) and they weren't available

greg
2017-04-20 16:30
yeah - okay - so you probably want to delete lines 55-61 in the net-post-install.sh.tmpl

greg
2017-04-20 16:30
That will make the default sources.list file not get overwritten.

wdennis
2017-04-20 16:30
OK

greg
2017-04-20 16:31
Another alternative that might work is to change the /etc/apt/sources.list reference to /etc/apt/sources.d/drp.repo

wdennis
2017-04-20 16:31
Maybe that should be standard, or at least selectable via a property

greg
2017-04-20 16:32
Yes - issue please. Interestingly enough - since you got us off our rears to add template in template support. I'll be reworking these quite a bit to add those actions.

wdennis
2017-04-20 16:32
Yes, agree, that would be better

wdennis
2017-04-20 16:33
Will do

wdennis
2017-04-20 16:36
In other matters, any DR folk planning on going to USENIX LISA'17?

greg
2017-04-20 16:36
maybe - that is sooooooooo far away. :slightly_smiling_face:


wdennis
2017-04-20 16:37
It would be a good place to demo your wares or give a talk at

greg
2017-04-20 16:38
good to know. Rob may have gone to the 16

wdennis
2017-04-20 16:39
I may (or may not) be submitting a talk (on Logstash improvements our lab has developed)

greg
2017-04-20 16:41
cool

2017-04-20 16:54
hm, I couldn't reproduce the wiped partition table issue from what I thought might have been a catalyst

greg
2017-04-20 16:55
ok - cool - i guess.

greg
2017-04-20 16:56
- PSA - I've updated the k8s workload to the latest kargo tree. It now defaults to v1.6.1

2017-04-20 17:25
@zehicle is this something I can do about now? edit/redeploy/reboot buttons still don't seem to be functional

greg
2017-04-20 17:27
@lae - switch to cli.

greg
2017-04-20 17:27
to make sure it works.

greg
2017-04-20 17:27
on the admin node, I usually do this:

greg
2017-04-20 17:27
docker cp compose_rebar_api_1:/usr/local/bin/rebar /usr/local/bin/rebar

greg
2017-04-20 17:27
chmod +x /usr/local/bin/rebar

greg
2017-04-20 17:27
export REBAR_KEY=u:p

greg
2017-04-20 17:28
rebar nodes list

greg
2017-04-20 17:28
rebar nodes redeploy <id of node in question>

greg
2017-04-20 17:29
Are the errors in the console of the browser? or in the rebar_api container log?

greg
2017-04-20 17:29
Get some of that would be good. All of that in an issue would be better. :slightly_smiling_face:

2017-04-20 17:30
console of the browser, I wasn't seeing anything while watching `docker-compose logs`

greg
2017-04-20 17:30
okay - still the raw error?

2017-04-20 17:30
raw error occurs for edit, but there are different errors for redeploy/reboot

greg
2017-04-20 17:31
awesome

2017-04-20 17:32
so I'm not too familiar with the cli - how do I change a node to change to `local` for redeployment?

greg
2017-04-20 17:32
```rebar nodes redeploy <id>```

2017-04-20 17:33
right, I did that - but does that change it to local? I still see bootenv: sledgehammer in the output

greg
2017-04-20 17:33
Will set the bootenv to discovery/sledgehammer, update the node roles run counts to 0

greg
2017-04-20 17:33
and reboot the node if possible.

greg
2017-04-20 17:34
You should be able to reboot the node now and it should go into sledgehammer.

greg
2017-04-20 17:35
redeploy in the DR sense means to redo discovery and reapply the current config

2017-04-20 17:35
ah ok

greg
2017-04-20 17:35
You may need to manually reboot the node if IPMI is NOT configured and the node is not SSH able.

2017-04-20 17:36
in the UI there's a dropdown when clicking the redeploy icon so I assumed something different

2017-04-20 17:36
ipmi's usable, I have a small tool for that

greg
2017-04-20 17:36
ah - the dropdown lets you reset the OS you want to provision.

2017-04-20 22:12
``` rebar nodes propose 2 rebar nodes update 2 '{"bootenv": "ubuntu-16.04-install"}' rebar nodes commit 2 rebar nodes power 2 reboot ``` is this the proper workflow or am I missing a step?

2017-04-20 22:18
hm, so if I run `nodes update` with either `local` or `sledgehammer` bootenv it'll change the pxelinux config file fine, but `ubuntu-16.04-install` and `centos-7.3.1611-install` don't change it at all

2017-04-20 22:37
Hey folks, if I run "join_rebar.sh" I authenticate OK, but get a * Connection #0 to host 192.168.1.63 left intact curl: (22) The requested URL returned error: 502 Bad Gateway We could not create a node for ourself!

greg
2017-04-20 22:45
@Iae - do ```rebar provisioner machines list``` and see if there are errors there.

greg
2017-04-20 22:46
@newgoliath - that script is old and fragile. It check to make sure it is NOT using 3000 as its port to connect to the admin node.

2017-04-20 22:46
It auths to rebar OK.

2017-04-20 22:47
the ansible installer isn't done yet. Maybe I'm being to eager. :-)

greg
2017-04-20 22:50
oh - patience

2017-04-21 00:41
``` root@labs-rebar:~/digitalrebar/deploy# rebar provisioner machines list 2017/04/21 00:41:31 Error listing provisioner machines: Expected status in the 200 range, got 404 Not Found ```

2017-04-21 00:42
hm :|

2017-04-21 00:43
``` provisioner_1 | [GIN] 2017/04/21 - 00:39:56 | 200 | 548.631Ás | 172.17.0.11 | GET /bootenvs/ubuntu-16.04-install provisioner_1 | [GIN] 2017/04/21 - 00:40:27 | 200 | 530.185Ás | 172.17.0.11 | GET /bootenvs/local provisioner_1 | provisioner-mgmt2017/04/21 00:40:27.680734 backend: Updating bd71b2b6-6687-4b33-9861-1148c6fc2a63 1 provisioner_1 | provisioner-mgmt2017/04/21 00:40:27.680743 backend: Updating new bd71b2b6-6687-4b33-9861-1148c6fc2a63 1 provisioner_1 | machines:bd71b2b6-6687-4b33-9861-1148c6fc2a63 is a ChangeHooker provisioner_1 | [GIN] 2017/04/21 - 00:40:27 | 202 | 138.131765ms | 172.17.0.11 | POST /machines ``` I see this behaviour in provisioner logs

2017-04-24 16:56
are profiles not usable in rebar, yet?

2017-04-24 16:56
(I don't see any results when I search for "profile" in the docs, and the parameters I'm setting in profiles aren't being inserted in the pxelinux configuration files)

greg
2017-04-24 17:04
Two different profiles.

greg
2017-04-24 17:04
Profiles described in digitalrebar provision are not in digitalrebar.

greg
2017-04-24 17:04
There are profiles in digitalrebar, but they are for node/deployment attribute overrides

greg
2017-04-24 17:05
They are not available to the provisioner.

greg
2017-04-24 17:05
This is a disconnect that will be remedied in the coming months. We are inbetween.

greg
2017-04-24 17:06
@Iae - Do you need the orchestration/hardware pieces (IPMI, BIOS, RAID) for what you are doing? If not, you may want to try digitalrebar provision. It may be more immediately applicable.

2017-04-24 17:15
They're not necessarily needed since I already have tools/processes for that. So I was attempting to use dr-provision earlier this morning, but it seems I'm having trouble possibly having it integrate with our current Infoblox setup?

2017-04-24 17:16
I could configure a subnet in dr-provision but it'd then allocate IPs itself, but if I leave subnets out, machines don't seem to boot: ``` PXELINUX 6.03 lwIP 2014-10-06 Copyright (C) 1994-2014 H. Peter Anvin et al Unable to locate configuration file ```

2017-04-24 17:16
``` dr-provision2017/04/24 17:12:26.849155 Recieved DHCP packet: type Discover xid 0x6b6d145f ciaddr 0.0.0.0 yiaddr 0.0.0.0 giaddr 0.0.0.0 chaddr 00:16:3e:07:33:21 dr-provision2017/04/24 17:12:27.829885 Recieved DHCP packet: type Discover xid 0x6b6d145f ciaddr 0.0.0.0 yiaddr 0.0.0.0 giaddr 0.0.0.0 chaddr 00:16:3e:07:33:21 dr-provision2017/04/24 17:12:29.807194 Recieved DHCP packet: type Request xid 0x6b6d145f ciaddr 0.0.0.0 yiaddr 0.0.0.0 giaddr 0.0.0.0 chaddr 00:16:3e:07:33:21 dr-provision2017/04/24 17:12:29.807382 xid 0x6b6d145f: Ignoring request for DHCP server 10.13.2.24 dr-provision2017/04/24 17:12:36.196744 sending block 0: code=0, error: TFTP Aborted ```

2017-04-24 17:18
also I guess I should probably repull dr-provision since I see that PR on profiles was just merged

greg
2017-04-24 17:25
well - you can turn off dhcp in dr-provision.

greg
2017-04-24 17:25
okay - a couple of things.

greg
2017-04-24 17:26
I've updated tip with all the stuff fomr this morning - should be out there in about 10minutes. You will need to get the tarball from the tip or use --rs-version=tip in the curl bash install command.

greg
2017-04-24 17:27
Second, you can use DR-provision with DHCP off and let infoblox drive IP assignment. You will need to set nextboot server to dr-provision and set a bootfile to lpxelinux.0 (assuming legacy intel bios machines). If you are going to do discovery, you will need to also set option 15 (domainname to something).

greg
2017-04-24 17:29

greg
2017-04-24 17:29
I need to merge configure server and deployment. They are similar but different.

2017-04-24 17:32
my Infoblox as far as I know is configured correctly for both legacy/uefi systems at the moment (well, the bootfiles are prefixed with discovery/ still, but I created a symlink in drp-data/tftpboot)

greg
2017-04-24 17:33
won't work. The filesystem isn't live like that. The non-static files are generated on the fly when requested. This allows for parameters from profiles and other things to be inject as needed and things aren't out of date..

greg
2017-04-24 17:34
I have some docs in about that, but it is still coalescing.

2017-04-24 17:35
is that a difference between DR and dr-provision? PXE boot was working correctly on DR with this configuration - I can have it modified but since I don't admin the infoblox instance, I've gotta wait until a change gets implemented before I can test things again

greg
2017-04-24 17:35
yes

greg
2017-04-24 17:36
We are attempting to handle more dynamic and richer bootenvs/templates with dr-provision.

2017-04-24 17:38
Alright, just to confirm - based on what you said this config should be fine, right? ``` subnet 10.11.110.0 netmask 255.255.255.128 { option domain-name "eng.fireeye.com"; option domain-name-servers 10.11.10.12, 10.11.10.140; option routers 10.11.110.1; server-name "10.11.110.50"; next-server 10.11.110.50; filename "lpxelinux.0"; if (substring(option vendor-class-identifier,0,9)="PXEClient:Arch:00006") { # Option filter "UEFI-BOOTIA32.efi" server-name "10.11.110.50"; next-server 10.11.110.50; filename "bootia32.efi"; } elsif (substring(option vendor-class-identifier,0,9)="PXEClient:Arch:00007") { # Option filter "UEFI-BOOTX64.efi" server-name "10.11.110.50"; next-server 10.11.110.50; filename "bootx64.efi"; } } ```

greg
2017-04-24 17:42
I think so. The substring options look okay.

2017-04-24 22:04
Alright, sledgehammer booted! :ok_hand:

greg
2017-04-24 22:08
:slightly_smiling_face: Cool

greg
2017-04-24 22:09
I'll be on phone but driving.

greg
2017-04-24 22:09
in 10 minutes.

greg
2017-04-24 22:15
on phone

zehicle
2017-04-25 01:15
@greg wrong channel :slightly_smiling_face:

2017-04-25 12:30
question, is it possible to provision bare-metal from the ground up (configure RAID, install OS, etc.) with DRB and DRB-Provision? I work at Cisco and (obviously) we use Cisco equipment. I have been playing with DRB and DRBP, so far every thing works fine. Now I'm looking for guidance on how to go from metal-to-running_app using DRB.

greg
2017-04-25 13:03
Hi @vjcubas. It probably is. We haven't done much with Cisco gear. We usually have success with discovery and OS install and beyond. There is sometime some issues with IPMI. Raid will be dependent upon what raid controller is in the gear. BIOS configuration and setting is unlikely without some additional coding. It is the most hardware specific thing. It can be added but it takes some time and direct access to a box. (vendor tools help as well).

greg
2017-04-25 13:05
With regard to DRB (DR) vs DRB-Provision (DRP). Currently, DRP is just about OS installation and discovery. It is not currently integrated with DR. So, if you are looking for Orchestration or hw manipulation, then you will want to focus on DR. If you have DR working, then you are doing well.

greg
2017-04-25 13:06
We may need to have a discussion around what you are trying to do, where you are, and end states.

greg
2017-04-25 13:07
If you want, we can talk off-line (email me at )

greg
2017-04-25 13:10
Also, what do you have running? DRP vs DR or both

2017-04-25 17:20
right now I have DRP - and I was wondering if we could run DR and DRP on the same server

greg
2017-04-25 17:20
They will fight currently. with port contentions. well - let me think a moment.

greg
2017-04-25 17:21
okay - you kinda can, but the integration would be pretty light and may not be what you want.

greg
2017-04-25 17:22
Partially, why I asked what your end goal is.

greg
2017-04-25 17:24
You could run DRP with DR on the same system if you turn off the DHCP and Provisioner server in DR. To hook, DRP to DR you would need to create a script to "join" installed nodes into DR after it was done. The machines would show up (you probably couldn't manage lifecycle directly from DR, but it would be something).

greg
2017-04-25 17:24
Acutally, this is interesting quick integration path for the short term. Hmmm .. not what I want medium to long term. Need to think.

greg
2017-04-25 17:25
The intent is DR for more full featured environments (hardware settings, workload orchestration, ...).

greg
2017-04-25 17:25
Back to your end state goals.

2017-04-25 17:42
Getting errors on "logging-client" complaining about system clock "INFO: HTTP Request Returned 401 Unauthorized: Authentication failed. Please check your system's clock". Is there a set to run ntpdate that is missing in deployment?

greg
2017-04-25 17:43
Yes - it should run ntpdate as part of sledgehammer.

greg
2017-04-25 17:43
As part of common.env, you can set an upstream ntp server. by default it will make the dr master node the time server.

greg
2017-04-25 17:44
You can check the sledgehammer log by logging into the node in question: root/rebar1

greg
2017-04-25 17:44
Then: journalctl -u sledgehammer

greg
2017-04-25 17:44
Look for ntp

greg
2017-04-25 17:44
it may give a hint at what is going on.

2017-04-25 17:46
Added 4 hosts to deployment. Only 1 worked. The date on the working system, the time is 10:30 UTC. The date on the provisioner is 3:30 PDT. The command journalctl -u sledgehammer returns "-- No Entries --"

2017-04-25 17:49
The contents of /var/chef/cache/chef-stacktrace.out starts with Net::HTTPServerException: 401 "Unauthorized"

greg
2017-04-25 17:50
I've seen that sometime the ntp server on the node can get out of sync with it self or take a long time to stablize. I'm not sure what is causing this. Sometime restarting the ntp container. will "fix" it.

greg
2017-04-25 17:50
Is this post OS install or during discovery?

2017-04-25 17:51
How do I get in docker to set the date? It is post OS install. I cannot run ntpdate on the provisioner and get an error of "25 Apr 03:49:47 ntpdate[19206]: the NTP socket is in use, exiting"

greg
2017-04-25 17:52
```ntpq -p``` to check the state of ntp on the admin node

greg
2017-04-25 17:52
The docker container should be using the host's clock

2017-04-25 17:53
ntpq -p returns "localhost: timed out, nothing received"

greg
2017-04-25 17:53
okay - docker ps | grep ntp

2017-04-25 17:54
c492475e0902 digitalrebar/dr_ntp:master "/sbin/docker-entr..." 4 days ago Up 4 days 0.0.0.0:123->123/tcp, 0.0.0.0:123->123/udp compose_ntp_1

greg
2017-04-25 17:55
What install mode did you use for DR?

2017-04-25 17:56
host

greg
2017-04-25 17:56
okay - cool - hmmm

greg
2017-04-25 17:56
ntpstat

greg
2017-04-25 17:56
Post Install on all 3 of the failing nodes.

greg
2017-04-25 17:57
That service is different. I think it is rebar

2017-04-25 17:57
Yes, but one post install succeeded

greg
2017-04-25 17:57
Yes - if clock was close it would be fine.

greg
2017-04-25 17:57
chef is very very time sensitive in its cert management.

greg
2017-04-25 17:58
So - my guess is that the 3 nodes didn't time sync for some reason and didn't join correctly. The logging-client role is the first chef role encountered.

greg
2017-04-25 17:58
What os?

2017-04-25 17:59
rhel 7, do I need to start ntpd on provisioner or just run ntpdate to set the time correctly for docker ntp?

greg
2017-04-25 17:59
is the provisioner time's off?

greg
2017-04-25 17:59
ntpdate -q on the admin node hsould adjust it on the docker container. The problem is it might wig out ntp for a bit.

greg
2017-04-25 18:01
on centos7/rhel7 systems, you should have a rebar service

greg
2017-04-25 18:01
systemctl status rebar

greg
2017-04-25 18:01
journctl -u rebar

2017-04-25 18:02
systemctl status rebar Unit rebar.service could not be found.

greg
2017-04-25 18:03
hmmm - did you use the centos7 ks script? it should have created:

greg
2017-04-25 18:04
/usr/sbin/rebar_join

2017-04-25 18:04
We worked last friday on the phone and setup rhel-7.3-server

greg
2017-04-25 18:04
oh - Darren.

2017-04-25 18:04
Yes

2017-04-25 18:05
/usr/sbin/rebar_join -bash: /usr/sbin/rebar_join: No such file or directory

greg
2017-04-25 18:07
okay - checking files.

2017-04-25 18:09
rebar_join on box being install returns ... /usr/sbin/rebar_join@48(): [[ -x /bin/rebar ]] /usr/sbin/rebar_join@53(): export REBAR_ENDPOINT=https://192.168.128.10 /usr/sbin/rebar_join@53(): REBAR_ENDPOINT=https://192.168.128.10 /usr/sbin/rebar_join@65(): ntpdate 192.168.128.10 /usr/sbin/rebar_join@67(): case $1 in /usr/sbin/rebar_join@71(): echo 'Unknown action to rebar_join.sh.' Unknown action to rebar_join.sh. /usr/sbin/rebar_join@72(): exit

greg
2017-04-25 18:09
rebar_join start

greg
2017-04-25 18:11
in fact, that run successfully on machine startup because you got the logging_client error.

greg
2017-04-25 18:13
You could try ntpdate -q and the retry the node role.

2017-04-25 18:19
The rebar_join start worked - now they are all finishing the install correctly. Thanks

2017-04-25 18:20
Still had problem with bios-discover step - had to "retry" on every system and it then passed that step. The ntp problem is new

greg
2017-04-25 18:24
hmm - seems like we still have some networking communication issue. Things are consistently talking it seems.

2017-04-25 18:30
Is this because we are running "tools/docker-admin-up --access HOST --no-pull" instead of "tools/docker-admin-up --access HOST"?

2017-04-25 18:32
I remember now, the --no-pull option just was used to not download images from the Internet, correct?

greg
2017-04-25 19:45
yes

greg
2017-04-25 19:45
@intendo - yes

2017-04-25 20:53
I restarted digital rebar with:

2017-04-25 20:53
cd ~/digitalrebar/core; tools/docker-admin-down ; sleep 10 ; EXTERNAL_IP=192.168.128.10/24 tools/docker-admin-up --access HOST --no-pull

2017-04-25 20:54
None of my nodes, templates, etc. are showing up from the rebar api or GUI. What did I forget to do?

greg
2017-04-25 21:10
ohh - that is destructive install. You need to reboot the nodes for them to be rediscovered and await in sledgehammer for an OS to install.

greg
2017-04-25 21:10
If you are just wanting to restart containers, ```docker-compose restart``` from the deploy/compose directory will work.

zehicle
2017-04-25 21:42
Gitter/IRC users > we can get you invites directly into this slack channel is you prefer Slack. They are all synchronized, so there's no need to switch if you are happy where you are

2017-04-25 22:06
I have resisted the slack, but... I am in so many slack channels now might as well go slack native.

2017-04-25 22:07
If you can shoot invites to mike, ben, jordan @supergiant.io

2017-04-25 22:10
@zehicle - how do I get my templates back into digital rebar? Do I have to rerun all the commands ?: rebar provisioner files upload chef-12.18.31-1.el6.x86_64.rpm to chef/chef-12.18.31-1.el6.x86_64.rpm rebar provisioner templates upload rhel-7.3-server.ks.tmpl as rhel-7.3-server.ks.tmpl rebar provisioner bootenvs create - < rhel-7.3-server.json

greg
2017-04-25 22:11
The chef should still be there in the cache directory if you used the same user. The other two commands are correct and this is @greg. :slightly_smiling_face:

2017-04-25 22:31
@galthaus, Just FYI, everything you type shows up as "Rob Hirshfeld @zehicle\n [Greg Althaus, RackN]" and then your message.

2017-04-25 22:32
Yeah - it is the side effect of the slack to gitter integration we are using.

2017-04-25 22:32
I use the slack side - all my groups are in there. It is easier for me to watch that one place.

2017-04-25 22:32
Thanks though for pointing it out. :-)

greg
2017-04-25 22:33
@galthaus will pop a notification in gitter for me. @greg will pop up the notification in slack. Now you all can hound me everywhere. :slightly_smiling_face:

2017-04-25 23:21
@greg where is the official slack, or can I get some invites sent?

greg
2017-04-25 23:21
working on it. Have to remember. :slightly_smiling_face:

greg
2017-04-25 23:26
@jordan - done, I think.

jordan
2017-04-25 23:28
has joined #community201704

jordan
2017-04-25 23:28
Thanks

greg
2017-04-25 23:28
:slightly_smiling_face:

jordan
2017-04-25 23:30
Mike is going to start working on http://packet.net integration for SuperGiant tomorrow, and likely at the same time by extension digital rebar, so you will probably see him around these parts for a while as we work through that.

jordan
2017-04-25 23:31
You know anyone else doing anything like packet? Other than softlayer - sort of - I don't know anywhere else one can pop raw hardware?

greg
2017-04-25 23:32
yeah - not sure either. We were playing with Nobis/Ubiquity at one point, but I think they got acquired. Not sure the state of their universe.

jordan
2017-04-25 23:36
Have you worked with softlayers bare metal offerings at all?

greg
2017-04-25 23:37
nope - haven't looked. Was/Is on a One Day provider thing, but not recently

jordan
2017-04-25 23:37
or well, bluemix technically I guess

2017-04-26 00:00
@galthaus, I have rerun the provisioner and I am trying to create a workload -> Install O/S but I can't see a list of the systems. How do I put the systems back in the pool? I am still logged into two of the systems. I turned off 4 systems and then turned them back on but nothing is showing up in the ux.

greg
2017-04-26 00:01
Nodes once discovered should show up in the system deployment the first time and under the nodes nav tree item.

greg
2017-04-26 00:01
If the nodes don't show up there, then they haven't been discovered (or discovered successfully).

greg
2017-04-26 00:02
Are your nodes configured to PXE boot by default?

2017-04-26 00:09
@galthaus How do you configure them to PXE boot by default? Is that set in the BIOS? The first time I turned them on, they were visible by the UX. After the install, I did the "destructive teardown" and now none of the systems are visible. Did the install change the PXE boot to default to local disk?

greg
2017-04-26 00:21
Well - if I recall, you told me that the nodes were wiped. If the disks were wiped, they'd fall back to PXE most likely.

greg
2017-04-26 00:21
You can try the IPMI tool to force a pxe boot.

greg
2017-04-26 00:22
You may have one of the command history with the power status

greg
2017-04-26 00:24
ipmitool -U <username> -P <password> -H <ip address> chassis bootdev pxe

greg
2017-04-26 00:26
In theory, this will make it consistently pxe boot.

greg
2017-04-26 00:26
```ipmitool -U <username> -P <password> -H <ip address> chassis bootdev pxe options=persistent```

2017-04-26 00:32
@galthaus how do I get the IP addresses?

2017-04-26 00:33
I can ssh to the bmc IP address but I get the SMASH console

greg
2017-04-26 00:34
Well - will need to find them, probably since the nodes are deleted. You have the bmc network. The addresses were assigned from there. They start with the start address. The networks nav in the UI should get you to the bmc network definition to see the starting address.

2017-04-26 00:57
@galthaus here is what I did to get _most_ of them set to bootdev pxe: # for ip in 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215; do echo $ip; ipmitool -H 192.168.128.$ip -U root -P rebar1 chassis bootdev pxe options=persistent; done 200 Error: Unable to establish LAN session 201 Set Boot Device to pxe 202 Set Boot Device to pxe 203 Set Boot Device to pxe 204 Set Boot Device to pxe 205 Set Boot Device to pxe 206 Set Boot Device to pxe 207 Set Boot Device to pxe 208 Set Boot Device to pxe 209 Set Boot Device to pxe 210 Set Boot Device to pxe 211 Set Boot Device to pxe 212 Set Boot Device to pxe 213 Error: Unable to establish LAN session 214 Error: Unable to establish LAN session 215 Error: Unable to establish LAN session

greg
2017-04-26 00:59
Cool - now reboot them. You can use the ipmitool chassis power cycle

greg
2017-04-26 00:59
like above but different. :slightly_smiling_face:

2017-04-26 01:30
@galthaus rebooted them but no new nodes showing up in list

greg
2017-04-26 01:31
do you see DHCP leases for the nodes in the DHCP nav section.

2017-04-26 01:34
@galthaus yes, but only 4 nodes are binding

greg
2017-04-26 01:36
hmmm - so nothing in the nodes list. hmm

2017-04-26 01:37
@galthaus there are 4 nodes in the nodes list but there should be 16

greg
2017-04-26 01:37
ah - okay.

2017-04-26 01:37
or at least 12

2017-04-26 01:37
@galthaus - would this be easier to skype?

greg
2017-04-26 01:39
probably - can we try tomorrow morning?

2017-04-26 01:41
@galthaus yes, I will try to call you but I have meetings up the wazzu :-)

greg
2017-04-26 01:41
okay - sorry I'm fading is the problem and I need to finish some doc work I'm doing.

greg
2017-04-26 05:15
A new release of Digital Rebar Provision is out, v3.0.1 and stable has been updated. https://github.com/digitalrebar/provision/releases/tag/v3.0.1

2017-04-26 15:01
@galthaus Ready whenever you are to Skype.

greg
2017-04-26 15:01
there

2017-04-26 18:16
You guys ever test DR on Centos 6.9?

greg
2017-04-26 18:47
not specifically.

2017-04-26 21:08
can I get a slack invite? lae@lae.is

greg
2017-04-26 21:21
Done

lae
2017-04-26 21:41
has joined #community201704

2017-04-27 14:22
I've been looking at the DRP docs - trying to add custom ISO image - I'm not clear on the process, does anyone here have suggestions on this topic. I apologize if this is not the proper channel for posting DRP questions - Thanks in advance:)

greg
2017-04-27 14:39
I'm working on docs now. Check operations section about bootenvs. What are trying to do?

gopherstein
2017-04-27 14:57
has joined #community201704

2017-04-27 15:01
we have a custom CentOS-7 iso that we'd like to deploy to the targeted hosts - also, once this image has been installed on those hosts, I'd like to be able to reboot them without re-imaging them through the PXE boot - I know changing the boot order in the bios would accomplish this, but perhaps there is a better/easier way to do so - thanks

lae
2017-04-27 15:02
you'd set bootenv to local as part of %post in your centos kickstart file

greg
2017-04-27 15:28
Thanks, @lae ! Exactly. I'm in the process of documing and templatizing this so that you can include it in any custom kickstart. something like this:

greg
2017-04-27 15:29
``` {{ template "update-drp-local" . }} ```

2017-04-27 18:14
thanks, another question: is there a way to install ssh key during the deployment, or to set user/password - BTW, what is the default user/pass for the centos-7-3.1611 image after deployment?

greg
2017-04-27 19:25
Yes and yes.

greg
2017-04-27 19:25
root/RocketSkates



greg
2017-04-27 19:28
There are some more docs and examples coming describing just those issues.

2017-04-28 19:58
@galthaus Our reinstall (ll41) is "stuck" in the provisioner-os-install mode where it keeps repeating the output: 2017/04/28 19:44:15 Could not connect to Rebar: Head https://127.0.0.1:3000/api/v2/digest: dial tcp 127.0.0.1:3000: getsockopt: connection refused //tmp/scriptjig-Dkrnn4/provisioner-os-install/01-install-os.sh@51(): sleep 1 //tmp/scriptjig-Dkrnn4/provisioner-os-install/01-install-os.sh@50(): rebar nodes get d71485f0-5c26-4438-8b9c-9648c8b2b4b5 attrib provisioner-active-bootstate //tmp/scriptjig-Dkrnn4/provisioner-os-install/01-install-os.sh@50(): grep -q -- -install

greg
2017-04-28 20:01
rebar nodes update d71485f0-5c26-4438-8b9c-9648c8b2b4b5 '{ "bootenv": "local" }'

greg
2017-04-28 20:01
rebar nodes update d71485f0-5c26-4438-8b9c-9648c8b2b4b5 '{ "bootenv": "rhel-7.3-server-install" }'

greg
2017-04-28 20:02
See if that clears it.

2017-04-28 20:04
@galthaus Nope, gets stuck in same loop

greg
2017-04-28 20:05
hmm - bigger hammer

greg
2017-04-28 20:05
rebar nodes redeploy d71485f0-5c26-4438-8b9c-9648c8b2b4b5

2017-04-28 20:09
@galthaus Nope, is is still stuck in that loop. Got a biggerer hammer?

greg
2017-04-28 20:09
The redeploy should have rebooted the node. If it didn't then try and reboot the node.

greg
2017-04-28 20:10
rebar nodes power d71485f0-5c26-4438-8b9c-9648c8b2b4b5 reboot

2017-04-28 20:11
@galthaus That forced the reboot

greg
2017-04-28 20:13
what is the node's bootenv?

2017-04-28 20:17
sledgehammer

greg
2017-04-28 20:17
okay that is good. Let's see what happens. It should redo the whole process.

2017-04-28 20:19
I will let you know if it works. Going to put all the missing drives into the hardware to add another 24 nodes.

greg
2017-04-28 20:20
cool

zehicle
2017-04-28 20:30
note: edits do NOT make it into the gitter channel...

greg
2017-04-28 20:31
yes - I know. Teach them to join slack. :slightly_smiling_face:

2017-04-28 21:22
Probably a silly question, but I'm brand new to rebar and trying to setup a bare metal environment. I've run the "run-in-system" script with provisioner and dhcp containers and then I found out that I need to get the RAID and BIOS tools :-P So I did that but I'm not totally sure how to re-run just raid-tools-install role.

greg
2017-04-28 21:25
There is a retry button in the ux. The annealer view top right spiral looking icon.

greg
2017-04-28 21:25
Select that.

greg
2017-04-28 21:25
The errors will be at the top. On the right, there should be a retry all button.

2017-04-28 21:26
hmmm.. no errors listed.

2017-04-28 21:33
here is what I have done so far: - pulled digital rebar - ran sudo ./run-in-system.sh --deploy-admin=local --access=host --con-provisioner --con-dhcp --admin-ip=<ip/subnet> - downloaded and copied RAID and bios tools into ~/.cache/.../files/raid

2017-04-28 21:33
but when I login to the UX, I don't see the "Provisioner" tab.

2017-04-28 22:01
Hi guys, with the quickstart deploy, I try to add drpcli bootenvs install bootenvs/centos-7... and it fails to explode. Where can I look for errors? The error message in the logs just ends with /sbin/selinunxenabled

2017-04-28 22:01
error code 255

2017-04-28 22:02
the bootenv should have some error messages

2017-04-28 22:02
Nothing actionable, seemingly: Explode ISO: explode_iso.sh failed for centos-7.3.1611-install: exit status 255", "Command output:\nExplode iso centos-7.3.1611 /root/drp-data/tftpboot /root/drp-data/tftpboot/isos/CentOS-7-x86_64-Minimal-1611.iso /root/drp-data/tftpboot/centos-7.3.1611/install\nExtracting /root/drp-d ata/tftpboot/isos/CentOS-7-x86_64-Minimal-1611.iso for centos-7.3.1611\n/sbin/selinuxenabled\n

2017-04-28 22:02
the install combines several actions, you may need to try them one at a time. Also, I think there's a --debug flag for the CLI

2017-04-28 22:02
Not OK to run as root?

2017-04-28 22:03
DRP must run w/ root privs

2017-04-28 22:04
but the CLI does not requiire it

2017-04-28 22:04
disk space?

2017-04-28 22:04
Plenty free.

2017-04-28 22:06
also, drpcli bootenvs list doesn't show the ID of a bootenv, so you can't know the ID to use with drpcli bootenvs destroy <ID>

2017-04-28 22:06
Unless it's "name'

2017-04-28 22:07
I'm not much help on this score, soryr

2017-04-28 22:08
Ok, so name == ID

2017-04-28 22:09
bootenvs destroy and create again put it in available: true

2017-04-28 22:09
Timing issue, mayhaps.

2017-04-28 22:22
Now I'm getting malformed basic auth strings:

2017-04-28 22:22
dr-provision2017/04/28 22:21:15.241830 Malformed basic auth string: dG9vbHMvZGlzY292ZXJ5LWxvYWQuc2g= [GIN] 2017/04/28 - 18:21:15 | 401 | 55.948Ás | 192.168.1.1 | GET /api/v3/bootenvs dr-provision2017/04/28 22:21:20.085661 Bad auth header: Basic [GIN] 2017/04/28 - 18:21:20 | 401 | 42.698Ás | 192.168.1.1 | GET /api/v3/bootenvs

2017-04-28 22:22
Can't login anymore.

greg
2017-04-29 02:05
Huh?

greg
2017-04-29 02:06
@newgoliath - selinux could keep explode iso from doing its explode work in the local directory or /var/lib/tftpboot

greg
2017-04-29 02:06
I haven't tried on a system selinux enabled.

greg
2017-04-29 02:07
With regard to auth, the RS_KEY should be username:password or -U and -P should be used.

greg
2017-04-29 02:07
Make sure RS_TOKEN is not set.

greg
2017-04-29 02:08
@spencerwjensen - missing the provisioner tab means that the containers didn't come up correctly.

greg
2017-04-29 02:09
cd digitalrebar/deploy/compose

greg
2017-04-29 02:09
docker-compose ps

greg
2017-04-29 02:09
wait - start over - host should be HOST. You may get a in wonky mode that way.

greg
2017-04-29 02:10
Rerun the run-in-system command.

2017-04-29 05:39
Wow! Thanks @galthaus! I will give that a shot!

wdennis
2017-05-01 03:19
@greg - you going to be at DOD Austin this year?

greg
2017-05-01 03:23
Yes - I think the plan is for all of us to be around. Victor, Rob, and I.

wdennis
2017-05-01 03:24
Nice, would be great to meet you all?

greg
2017-05-01 03:24
Yes - put real faces with people.

wdennis
2017-05-01 03:25
I hear there?s no Uber/Lyft down there yet; what?s the local cab service(s)? (Don?t plan on renting a car?)

greg
2017-05-01 03:26
There are replacements, but I'm not sure what they are. Should be able to google.

wdennis
2017-05-01 03:26
OK

greg
2017-05-01 03:28
It is a little funky. There are cabs, but @zehicle may know.

wdennis
2017-05-01 03:57
Did some research - there?s some Austin ride-share services available ? downloaded apps for RideAustin and Fasten

wdennis
2017-05-01 03:57
I need to get out and eat BBQ / maybe see some music when I?m down there? :grinning:

greg
2017-05-01 03:59
Yes -