prezzy_dent
2018-09-01 12:18
https://provision.readthedocs.io/en/tip/doc/configuring.html#rs-configuring-default states for Ubunty it's RocketSkates (https://provision.readthedocs.io/en/tip/doc/configuring.html#default-template-identity); in console (not SSH) it doesn't work for the "root" user. I followed Quick Install (from tip) step by step, and installed Ubuntu 18.04 as per Install Your First Machine - https://provision.readthedocs.io/en/tip/doc/quickstart.html#install-your-first-machine. I will try the approach with SSH key next, but neither RocketSkates nor rocketskates work for "root" in console. Edit: 1) With CentOS 7.5.xxx Minimal ISO, default password works fine Edit: 2) I am using ssh-access with Ubuntu 18 instead of console as a workaround

shane
2018-09-01 14:53
@prezzy_dent - for Ubuntu - the preseed specifies that the username is `rocketskates`, not "root". In the `net-seed.tmpl`, around lines 83 to 86 - the username/info/password are specified. Line 85 sets the default username, if you do not override it with a Param. Root appears to not be set - so you need to login as `rocketskates`, then `sudo` to root. I just tested that path on the Console of a machine.

shane
2018-09-01 14:54
(and `RocketSkates` is the correct password in our default preseed for the `rocketskates` user)

prezzy_dent
2018-09-01 14:58
Thank you. Interestingly in Centos there's "root". I am still learning (reading the manual right now) my way around the workflows, stages, etc. Should be able to check by myself but need more practice

shane
2018-09-01 15:01
The differences between CentOS and Ubuntu are specific to how those Distros handle their default root/username policies. DRP simply mimics those patterns.

prezzy_dent
2018-09-02 04:58
`runner-service` - I searched the slack, poked around the GUI - can't figure out how to actually use it to install a non-default rpm or deb package? I clone this service, then it asks me to add a template and content. All I want is `apt-get install vim -y` - do I need to create or reuse templates for this? If a video where this is demo'd exists I wouldn't mind to watch it. Is `Wait` or `Do not wait` appropriate Runner Wait choice for a clone of runner service?

shane
2018-09-02 05:02
Check the net-seed.tmpl - I think there is a Param you can set to add additional pkgs. Remember you can only add pkgs that are in the repo mirrors, otherwise the install will hang and not complete successfully....

shane
2018-09-02 05:04
Param is `extra-packages`

vladimir.karnushin
2018-09-02 05:24
Hey, guys! How do I access to nested fields of the some `Param` object? For instance: ``` {{if .ParamExists "network-configuration" -}} d-i netcfg/choose_interface select {{.Param "network-configuration/interface"}} d-i netcfg/disable_autoconfig boolean true d-i netcfg/get_ipaddress string {{.Param "network-configuration/ipaddress"}} d-i netcfg/get_netmask string {{.Param "network-configuration/netmask"}} d-i netcfg/get_gateway string {{.Param "network-configuration/gateway"}} d-i netcfg/get_nameservers string {{.Param "network-configuration/nameservers"}} d-i netcfg/confirm_static boolean true {{else -}} {{end}} ``` When I try to do this, I'm getting a error: ``` ??? 02 07:34:42 proxy1 dr-provision[631]: [0:8]Static FS: Dynamic file error for /machines/d5962282-8187-4d68-8b98-28fcb85c1284/seed: template: :438:4: executing "select-kickseed.tmpl" at <.CallTemplate>: error calling CallTemplate: template: qantor-preseed.tmpl:20:37: executing "qantor-preseed.tmpl" at <.Param>: error calling Param: No such machine parameter network-configuration/interface ```

shane
2018-09-02 16:58
@vladimir.karnushin - I assume that `network-configuration` is a param structure that you created? The "`/`" notation does not by itself reference "nested elements" in the Param. We use it for separation of Content Pack params - so we often notate things like "krib/foo" as "belongs to the 'krib' content, and is Param "foo". However - the Params full/correct name is still "krib/foo". The slash does not denote anything "special" about the param - it's just part of the name of the param.

shane
2018-09-02 17:00
If you want to use a more complex single YAML/JSON object with an array structure - then you'll need to look at a few other examples. I'd recommend checking the KRIB content pack for some examples - here the "`krib/cluster-masters`" param as an example (in YAML): https://github.com/digitalrebar/provision-content/blob/master/krib/params/krib-cluster-masters.yaml

shane
2018-09-02 17:03
You can see it in use in the template with the `range` golang template call iterating over it, at: https://github.com/digitalrebar/provision-content/blob/master/krib/templates/krib-kubeadm.cfg.tmpl#L14

shane
2018-09-02 17:07
An operating example of this structure in use looks like: ```- Address: 10.75.67.9 Name: machines-ewr1-01.unspecified.domain.local Uuid: b963457f-a2c8-453c-9223-3620c95f7af2 - Address: 10.75.63.11 Name: machines-ewr1-02.unspecified.domain.local Uuid: b962q57f-a2c8-453c-9297-3620c9447af2```

bagricola
2018-09-03 13:02
Getting a panic / 500 ?No decoder for content-type application/x-gzip? when trying to use drpcli users token [id]? but i swear it worked earlier when the user I was using was the default (not provided to drpcli via env var)

zehicle
2018-09-03 13:51
That can mean your password is not set @bagricola

zehicle
2018-09-03 13:52
If that fixes it, let's make sure it's in the faq

bagricola
2018-09-03 13:53
yeah, worked that out just now actually

bagricola
2018-09-03 13:53
i think my add user command didn?t work properly

bagricola
2018-09-03 13:53
there?s a user, but it has no roles ? but it does have a secret set

bagricola
2018-09-03 13:56
ahh crap i know why, passing it a username AND a json blob which i think just creates a blank user and ignores the json

bagricola
2018-09-03 13:56
:smile:

bagricola
2018-09-03 14:07
urgh

bagricola
2018-09-03 14:07
yeah it?s just a problem with how ansible escapes stuff in command line

bagricola
2018-09-03 16:40
also? i remember there being a kexec-ok param which i now can?t find which I thought was in community content? where does it come from? :slightly_smiling_face:

shane
2018-09-03 17:17
@bagricola you need tip DRP version and top content

bagricola
2018-09-04 10:35
yeah got it working, thought i was using tip DRP but I wasn?t, cheers

bagricola
2018-09-04 12:31
So I?m spinning up DRP using Ansible? so far it does stuff like uploads boot isos, sets up user accounts, configures subnets, sets boot env defaults etc. Basically right now it?ll get a machine to the point of being able to PXE boot a host to discovery, but no further. What I?d like to do is have a full DRP setup (workflows, stages, tasks, params etc) all imported and configured, when DRP is built. To that end I?m building out a content pack that contains everything for our specific setup, but one thing I?ve realised while looking at packing it is that it relies on some tasks and stages (lldp + others iirc) which come from e.g. `task-library` or `dev-library`. Is it possible to import those content packs automatically? Would it be better to just copy those stages into our own content pack, or do you guys already have a workflow around this type of thing?

shane
2018-09-04 13:12
```#!/bin/bash # # customize me RACKN_ACCOUNT_ID=${RACKN_ACCOUNT_ID:-""} ? [[ -z "$RACKN_ACCOUNT_ID" ]] && read -p "RackN Account ID: " RACKN_ACCOUNT_ID ? RACKN_AUTH="?username=${RACKN_ACCOUNT_ID}" CURL="curl -sfSL" BASE_URL="https://qww9e4paf1.execute-api.us-west-2.amazonaws.com/main/catalog" CONTENTS="drp-community-content task-library terraform krib" ? for CONTENT in $CONTENTS do echo "Saving '$CONTENT' content pack as './${CONTENT}.json' ... " $CURL \ ${BASE_URL}/content/${CONTENT}${RACKN_AUTH} \ -o ${CONTENT}.json # example inject in to DRP endpoint # drpcli contents create ${CONTENT}.json done```

shane
2018-09-04 13:13
@bagricola ^^^ Sample shell script snippet to download content pack

shane
2018-09-04 13:14
You need a valid portal user ID (account ID) for the Auth piece

bagricola
2018-09-04 13:17
@shane aaaaaah awesome, I assumed the AWS url used by the UI was transient and didn?t want to write anything around it but if that?s the way, then off I go :smile: Thanks muchly

greg
2018-09-04 14:27
@bagricola - you willing to share your ansible playbook?

bagricola
2018-09-04 14:29
@greg it?s on internal VCS right now as not finished but yeah, should be doable

shane
2018-09-04 14:30
@bagricola - when you're ready to show your stuff ( :slightly_smiling_face: ) I'll be happy to work with you to get it published

bagricola
2018-09-04 14:31
Cool :+1:

greg
2018-09-04 14:31
Thanks!

bagricola
2018-09-04 14:33
use case my side is bootstrapping new ovirt-based virt clusters so our ansible stuff (will contain) cumulus roles and such

bagricola
2018-09-04 14:33
has anyone run the DRP runner on cumulus linux before? :smile:

bagricola
2018-09-04 16:08
Sorry for all the questions :slightly_smiling_face: - do tokens generated with `drpcli users token <id>` expire? Having a weird issue with `drpcli contents create` (from ansible task below), where when using `RS_TOKEN` in the environment the first run of `drpcli ...` works and then all the following using the same token fail with `AUTH: Validation failed`. If I comment `RS_TOKEN` and use `RS_KEY` instead it works fine. Could activating one of those content files invalidate the token for example?

greg
2018-09-04 16:10
The default timeout unless specified is an hour.

greg
2018-09-04 16:11
you can add `ttl <seconds>` to the cli to increase the duration.

bagricola
2018-09-04 16:12
hrm - this is all over in a matter of seconds , activating content files is basically the next step after getting the token

bagricola
2018-09-04 16:13
think i?ll give it a try later manually and see what happens, just to rule out ansible weirdness

greg
2018-09-04 16:14
You may have to make sure you don?t get double double quotes

greg
2018-09-04 16:14
Does json_query return a double quoted json string?

bagricola
2018-09-04 16:16
nope, returns as just a normal python string so if I change the command to `env` i get `RS_TOKEN=<token>` with no double quotes

bagricola
2018-09-04 16:28
baffling? near as I can tell, the environment ansible is sending is correct

greg
2018-09-04 16:28
it may be a bug in our token auth side

bagricola
2018-09-04 16:29
however if i split the commands up and run them manually (two separate tasks) it doesn?t happen

bagricola
2018-09-04 16:29
wonder if it?s a timing thing?

greg
2018-09-04 16:29
clock sync issue (probably not - all localhost).

bagricola
2018-09-04 16:29
oh between drpcli and dr-provision

greg
2018-09-04 16:30
in general shouldn?t matter. All tokens are calculated relative to the dr-provision

bagricola
2018-09-04 16:32
i added a 5s delay in and it does the same thing, so I guess not timing

bagricola
2018-09-04 16:34
ah well, home time - problem for later :slightly_smiling_face:

zvanduyn
2018-09-04 19:31
Where can I find more information on RackN pricing?

zehicle
2018-09-04 19:35
generally, we setup a call to discuss

zehicle
2018-09-04 19:35
@zvanduyn ^^

zehicle
2018-09-04 19:35
I'll send you an email

zvanduyn
2018-09-04 19:37
Sounds good. It sounds like we are going to timebox an effort to implement DRB at a customer site, just not sure I can articulate what RackN would additionally provide + cost.

zehicle
2018-09-04 19:39
That's why a call is helpful to review cost, components and needs.

zvanduyn
2018-09-04 19:39
Totally. I'll reply to this thread you've started. Thanks.

zehicle
2018-09-04 19:58
This is your quarterly request to help promote the project! Any external promotion helps us spread the word about Digital Rebar Provision. Even github stars are helpful: https://github.com/digitalrebar/provision

allen
2018-09-04 20:29
has joined #community201809

ben.le
2018-09-04 20:33
Hi All, how to reset a user password? if he/she?s forgot

ben.le
2018-09-04 20:36
thanks all, i figured it out

greg
2018-09-04 20:36
nvm -?

greg
2018-09-04 20:36
:slightly_smiling_face:

ben.le
2018-09-04 20:36
thanks @greg

ben.le
2018-09-04 21:06
@greg why the ISO images of Debian-8 and Debian-9 are too small? i ran into issue when install both of them ?can?t find the kernel?. How to fix it?

shane
2018-09-05 00:03
@ben.le - just deployed Debian9 without problems - but you may need to upgrade DRP and DRP community content to latest versions.... Debian pulled version of 8 and 9 "off of the shelf" - so there may be problems with some of the older versions

shane
2018-09-05 00:04
we use the "mini" ISO to do Debian deployments against, the ISO is small by design, and all packages come from the mirrors

bagricola
2018-09-05 11:01
Trying to track down that auth issue from yesterday - it looks like the user `Secret` changes between the first `drpcli contents create` and the subsequent ones

bagricola
2018-09-05 11:01
the token being sent by drpcli is identical, but the User?s `Secret` field has changed so I assume it invalidates the token

bagricola
2018-09-05 11:02
`grantorSecret` and `userSecret` both change, as does the return from `drpcli users list`

bagricola
2018-09-05 11:22
so on a hunch, I commented out the line which changes the `user.Secret` when `user.ChangePassword` is run? and validation passes.

bagricola
2018-09-05 11:24
I then put the line which regenerates the `Secret` *before* the line that saves the user on password change, and it seems to work like that as well? I reckon the Secret is regenerated but not saved into the user store, so when I generate the token it?s somehow done using the ephemeral user token and doesn?t match when retrieved from store. Or something like that?

bagricola
2018-09-05 12:07
Or maybe the modified User object is being saved into the store at some point in the future with the new `Secret` after the token has been generated, invalidating it?

bagricola
2018-09-05 12:49

greg
2018-09-05 13:16
Thanks, @bagricola

prezzy_dent
2018-09-05 16:52
This might sound weird, but is there a way to name hosts' MAC addresses or IPs and just stream disk content and reboot (without any smarts or tricks)? I'm asking because I figure at this stage that would probably work fine (if it could be done) for my repeated provisioning task where everything is always identical (so I could provision manually once and save each server's disk image). Otherwise with custom ISOs that we use in these systems (different non-standard distro and app combos packaged as ISOs, each is custom made, used for different types of hosts) it's going to take me months to figure out how to explode them and so on...

greg
2018-09-05 17:44
@prezzy_dent - That is basically a two part thing. first part - can you capture a system. Second part, can you image an system from a capture.

greg
2018-09-05 17:45
The image-deploy plugin can be used to do the second. @shane has been working on parts to do the first.

greg
2018-09-05 17:46
Between @vlowther and @shane, they have content packs (internal to RackN right now) that build sledgehammer and capture it for use as a bootenv.

greg
2018-09-05 17:46
These are RackN licensed items. For full disclosure.

bagricola
2018-09-06 10:54
is the DRP catalog serverless? Just ran into a random 502 attempting to download drp-community-content and then all the others worked fine, felt like a spin up delay or something :slightly_smiling_face:

zehicle
2018-09-06 11:24
Yes

bagricola
2018-09-06 12:30
cool, i?ll just make Ansible retry a couple of times then :slightly_smiling_face:

bagricola
2018-09-06 13:32
sigh? just realised a day later that I opened an issue for DRP on the drp-content repo :confused:

shane
2018-09-06 13:55
no worries - we track both :slightly_smiling_face:

bagricola
2018-09-06 14:06
added it as a PR to the correct repo anyway, just in case my fix is the correct one :expressionless:

shane
2018-09-06 14:13
thanks @bagricola - we haven't had a chance to review yet, hopefully @greg will get a chance to look at it today

bagricola
2018-09-06 14:22
:+1: - on an unrelated note - am I right that it?s not possible to change the ?global? profile name assigned to all machines right now? Looks like the default profile name is hardcoded in `DataTracker`. Reason I ask is that ideally I?d like to include my own `global` profile in our own content pack, but as `global` is autocreated I get `keysCannotBeOverridden: global is already in layer 0`. I can obviously modify the global profile myself but that misses out all the niceties of being able to have everything in the content pack, tracked in git and appropriately CI?d :slightly_smiling_face:

shane
2018-09-06 14:23
yes - `global` is hard coded, and intended to always be available on all machines

shane
2018-09-06 14:25
it has the lowest order of precedence - so any Params specified in other profiles override it, and subsequently, any Params directly applied to the Machine Object override those

shane
2018-09-06 14:25
can you just call your global something like `global-squiz` or similarly ... ?

zehicle
2018-09-06 14:26
you can add a stage that would always add that provide to any machine during discovery

bagricola
2018-09-06 14:26
^ thats what I was missing, thanks :smile:

zehicle
2018-09-06 14:26
the RackN classify content package performs this function (proxy for @greg )

shane
2018-09-06 14:26
we also have some Classification content that can be used to verify a machines configuration, and based on some values - apply specific profiles and workflows to those machines

zehicle
2018-09-06 14:27
jinx

bagricola
2018-09-06 14:27
hah, thanks guys

bagricola
2018-09-06 14:27
okay i?ll just create a separate one and do it that way

bagricola
2018-09-06 14:32
the whole content pack thing is super useful btw, being able to version control all of it helps a lot?

shane
2018-09-06 14:34
yep - a lesson we learned many years ago ... :slightly_smiling_face:

cpizzi
2018-09-06 14:34
That's what attracted our company to DR to begin with -- the ability to deploy infrastructure with git webhooks.

bagricola
2018-09-06 14:38
yeah? thats nice

bagricola
2018-09-06 14:39
haven?t even really thought what more we can do with this yet more than what we were using foreman for

cpizzi
2018-09-06 16:49
Is it typical for a Machine to have pending tasks after a successful OS installation workflow?

shane
2018-09-06 16:55
@cpizzi - what does your Workflow look like, and what are the pending tasks? The current kickstart/preseed solution allows for adding Stages/Tasks to be executed post-install, but prior to reboot. One of the (often) common tasks is to run the `runner-service` stage to add the Agent as a service in the final OS piece ... some others include injecting SSH keys for remote access, etc...

cpizzi
2018-09-06 16:56
```{ "Validated": true, "Available": true, "Errors": [], "ReadOnly": false, "Meta": { "color": "black", "icon": "shuffle", "title": "OracleLinux" }, "Name": "oracle-7", "Description": "Oracle Enterprise Linux", "Documentation": "", "Stages": [ "oracle-7-install", "complete-nowait" ] }```

cpizzi
2018-09-06 16:56
`runner-service` rings a bell though, thanks.

cpizzi
2018-09-06 17:00
Is runner-service included in community content? I don't see it on my local, isolated installation.

shane
2018-09-06 17:00
it's in `task-library`

cpizzi
2018-09-06 17:10
Do I need a license for task-library? I'm having trouble installing it from the RackN portal

cpizzi
2018-09-06 17:22
Never mind I got it

cpizzi
2018-09-06 17:32
@shane that's just what I needed, thanks!

cpizzi
2018-09-06 17:33
Got the green checkmark of awesomeness :slightly_smiling_face:

shane
2018-09-06 17:33
cool

cpizzi
2018-09-06 17:55
What is a _node_ in terms of licensing paid content?

shane
2018-09-06 17:56
@cpizzi - answering on direct message

bagricola
2018-09-06 20:19
man kexec reboots are awesome :smile:

greg
2018-09-06 20:32
lol :slightly_smiling_face:

bagricola
2018-09-07 15:31
Can load data from netbox during discovery now :smile:

greg
2018-09-07 15:43
what do you mean by that? You have a task that queries netbox for network information to then deploy? @bagricola We looked at plugin integration with netbox to update inventory information and eventually pull info back, but it was unclear what the general pattern of operation with netbox is.

bagricola
2018-09-07 16:02
Yeah I?m not really sure if there is a well defined pattern of operation with Netbox itself - internally we use it as our ?source of truth? for certain things (IP allocations, rack locations, network connections etc) - a lot of it right now is ?put device into netbox and then match that config manually in the various locations?

bagricola
2018-09-07 16:03
So I wrote a stage which uses curl and JQ to pull information from the Netbox API during discovery - it?s configurable but currently using the gohai SerialNumber value to look up the machine by asset tag, then return its? device information and interfaces

bagricola
2018-09-07 16:04
would be easy to write data back as well, so could update the netbox device with provisioning status for example

greg
2018-09-07 16:04
Okay - I have a netbox plugin that will push machines and sync machines based upon events.

bagricola
2018-09-07 16:04
ahh cool

greg
2018-09-07 16:04
yeah, but ? I slow rolled it because there are mapping problems.

bagricola
2018-09-07 16:10
My end goal is to get provisioning (particularly re-provisioning) to be completely automated? I have the classification stuff working but the one sticking point is IP allocation, as we only use DHCP on the boot network and all other vlans are statically assigned. Right now the machine gets classified and configured but as it has no IP assigned there?s manual process to set it on the machine in DRP. As it already exists in Netbox, I should just be able to pull that information in, generate the network config and then trigger the install workflow automatically

bagricola
2018-09-07 16:10
and at that point we can start treating servers as immutable, which we don?t really do right now (and that sucks)

greg
2018-09-07 16:13
Yeah that makes sense. We have some stuff coming that an integration with netbox would be interesting.

bagricola
2018-09-07 17:16
Cool? was interested by the NetWrangler stuff from the last ?meet? or whatever it is you guys do? at the moment I?m just creating interface files using a template but it?s not exactly ?nice?, and one of my biggest annoyances is the format differences between the kickstart network config and the actual running network config post install..

shane
2018-09-07 17:34
Yep - NetWrangler is a nice piece to solve that problem and tie in nicely with the workflow

bagricola
2018-09-07 17:42
fwiw the ansible play / role I wrote is at a nice point to move out of our internal repo so I think I?ll do that on Monday in case it?s useful for anyone else, or at least to get some feedback on it :slightly_smiling_face:

bagricola
2018-09-07 17:42
but now is beer time!

shane
2018-09-07 17:50
cool - catch up with you on Monday ... have a cold one for me :slightly_smiling_face:

zehicle
2018-09-07 20:59
@stanchan also had done some playbook work I think

vlowther
2018-09-07 21:15
is installing centos 7 onto an arm64 vm. Very slowly.

vlowther
2018-09-07 21:16
But all through dr-provision.

zehicle
2018-09-10 00:00
Just added ISTIO MESH for Kubernetes (KRIB) using the advanced Helm chart configuration features just pulled into tip. https://youtu.be/qWSB3TTVUC8

zehicle
2018-09-10 00:00
This capability is available for anything with a Helm chart. We tried to enable a lot of features in the new helm chart runner - we'd love to hear if we got all of the ones you need

ido.samuelson
2018-09-10 06:43
Anyone know how to solve the "Bad archive mirror" when trying to install debian?

loswillios
2018-09-10 14:34
has joined #community201809

matt.welch
2018-09-10 16:35
has joined #community201809

shane
2018-09-10 16:46
@ido.samuelson sorry - not familiar with the Debian side - are you specifying a mirror site, or just using the open mirrors ?

shane
2018-09-10 16:46
@loswillios $welcome

2018-09-10 16:46
Digital Rebar welcome information is here > http://rebar.digital/community/welcome.html

shane
2018-09-10 16:46
@matt.welch welcome to you too :slightly_smiling_face:

matt.welch
2018-09-10 16:50
Thank you Shane

stanchan
2018-09-10 16:53
@zehicle @bagricola I had created a role a while back, but need to update it to the latest release. https://github.com/stanchan/ansible-role-dr-provision

greg
2018-09-10 16:57
nice - I had forgot about that.

stanchan
2018-09-10 17:01
Tried to make it flexible for any OS type, but not tested on ARM yet. Will try to setup a RPi cluster to test.

stanchan
2018-09-10 17:01
Unfortunately, been a little busy with other work stuff.

ido.samuelson
2018-09-10 17:27
@shane I think its internet connection as I got 4 nics on this server that i'm trying to automate

greg
2018-09-10 17:30
np

ido.samuelson
2018-09-10 17:31
is there any document or tutorial on how to setup an ISO install from scratch?

shane
2018-09-10 17:33
the quickstart document does that for the Ubuntu/CentOS BootEnvs - for Debian, you'd need to create a new BootEnv specification using the Full ISO and not the minimal ISO we specify - then use the same `drpcli bootenvs uploadiso ...` command structure

ido.samuelson
2018-09-10 17:33
I figured that much...guess i'm missing something :confused:

ido.samuelson
2018-09-10 17:34
as when i upload the iso...it still marked as not available

shane
2018-09-10 17:34
I assume you cloned an existing BootEnv and changed it ?

ido.samuelson
2018-09-10 17:35
yes, cloned both the debian one and ubuntu

shane
2018-09-10 17:35
you need to make sure you specify the correct sha256sum for the new ISO - or don't specify it if you don't want to do hash checks of the ISO (not recommended)

shane
2018-09-10 17:36
if you bore in on the BootEnv, you should see the error message why it's unavailable

ido.samuelson
2018-09-10 17:36
yes, I did

ido.samuelson
2018-09-10 17:37
yes it was 4 errors..i'll share it when i get to it later today

matt.welch
2018-09-10 17:44
At what phase would machine templates get created in `/var/lib/dr-provision/tftpboot/machines` for TFTP?

greg
2018-09-10 17:44
never

greg
2018-09-10 17:44
well - cheap answer.

greg
2018-09-10 17:44
They are never rendered to the file system.

greg
2018-09-10 17:45
You can check them by: `curl http://<ip>:8091/machines/?.`

matt.welch
2018-09-10 17:45
Does it render upon tftp request?

greg
2018-09-10 17:45
Yes

matt.welch
2018-09-10 17:45
aye, 10-4

greg
2018-09-10 17:45
That way it can be the latest values without having to track filesystem objects.

matt.welch
2018-09-10 17:45
Just picking things up and the first use case is probably the more difficult for ESX

greg
2018-09-10 17:46
Check the esxi content pack if you haven?t already. `os-other`

matt.welch
2018-09-10 17:46
I was using that but get an error it?s unable to find the boot.cfg file

matt.welch
2018-09-10 17:47
dumb me created the folder and file manually to see if that was ?it? and it got passed but felt wrong

greg
2018-09-10 17:48
not dumb

greg
2018-09-10 17:51
You may want to search the history for esxi. I thnk there are some notes in the channel.

matt.welch
2018-09-10 17:55
doing some traces it looks like it?s not doing the intercept. re: history - aye

greg
2018-09-10 17:55
is the machine?s bootenv set?

matt.welch
2018-09-10 17:56
Yea

greg
2018-09-10 17:56
hmm - try and curl that file.

matt.welch
2018-09-10 17:57
That?s what I was doing but w/ tftp client

matt.welch
2018-09-10 17:57
```tftp> get machines/56946aa3-46c8-48f5-8997-d3c3a2f0abe0/boot.cfg Error code 1: open /var/lib/dr-provision/tftpboot/machines/56946aa3-46c8-48f5-8997-d3c3a2f0abe0/boot.cfg: no such file or directory```

greg
2018-09-10 17:59
hmmm okay check the bootenv

greg
2018-09-10 18:01
check the logs for dr-p, see if a parameter is missing.

greg
2018-09-10 18:01
Also make sure the bootenv is available

matt.welch
2018-09-10 18:05
It may not be intended to work w/ UEFI chains. I had to clone the current template to remove the loop ```#!ipxe iseq ${platform} efi && goto efi || goto noefi :efi echo Chainloading into the ESXi install environment via ipxe does not work yet sleep 120 goto efi chain {{.Env.PathFor "http" "efi/boot/bootx64.efi"}} {{.BootParams}} :noefi set 210:string /{{.Env.PathFor "tftp" ""}}/ chain esxi.0```

dan734
2018-09-10 18:29
has joined #community201809

shane
2018-09-10 18:30
@dan734 $welcome

2018-09-10 18:30
Digital Rebar welcome information is here > http://rebar.digital/community/welcome.html

dan734
2018-09-10 18:31
thanks, reading

shane
2018-09-10 18:33
Hmm - looks like our link to our Youtube videos is no longer valid - will get that fixed :disappointed:

greg
2018-09-10 19:48
@matt.welch - can you write up a bug on it.

matt.welch
2018-09-10 19:50
on which piece, the ipxe loop error or the no boot.cfg found?

matt.welch
2018-09-10 19:50
I don?t know if I found a bug or not :disappointed:

greg
2018-09-10 19:52
the looping part if it is real.

greg
2018-09-10 19:53
The not rendering should show an error message in drp logs.

matt.welch
2018-09-10 19:53
aye

matt.welch
2018-09-10 19:57
``` #!ipxe iseq ${platform} efi && goto efi || goto noefi :efi echo Chainloading into the ESXi install environment via ipxe does not work yet sleep 120 goto efi chain {{.Env.PathFor "http" "efi/boot/bootx64.efi"}} {{.BootParams}} :noefi set 210:string /{{.Env.PathFor "tftp" ""}}/ chain esxi.0``` - i?ll submit for that

matt.welch
2018-09-10 19:57
I don?t see how it ever breaks out of the loop

greg
2018-09-10 20:04
I won?t. Efi isn?t exactly supported correctly.

greg
2018-09-10 20:05
Or installs really only know how to chain load to com32 based systems which doesn?t work with efi systems.

greg
2018-09-10 20:06
Or is might, but might not. That is why we have the esxi.0 image that we boot to. It has enough com support to run out of hte iso. Now, VMWare may have updates to work better with EFI and a different path, but we haven?t built or found it.

ben.le
2018-09-10 20:25
Can someone give me some examples how to configure DHCP built-in on DR-Provision server via drpcli command?

shane
2018-09-10 20:26
@ben.le - the quickstart guide has some info on it: https://provision.readthedocs.io/en/tip/doc/quickstart.html#rs-qs-subnet

shane
2018-09-10 20:27
you have to construct a JSON blob with the appropriate values - sometimes it's easier to create what you need from the UX - then use `drpcli` to "show" the created resource, and use that as a template for your automation

ben.le
2018-09-10 20:28
thanks @shane

loswillios
2018-09-11 05:42
thanks! looking forward to connect