functest | ||
ironic_discoverd | ||
.gitignore | ||
.gitreview | ||
CONTRIBUTING.rst | ||
example.conf | ||
ironic-discoverd.8 | ||
LICENSE | ||
MANIFEST.in | ||
plugin-requirements.txt | ||
README.rst | ||
requirements.txt | ||
setup.py | ||
test-requirements.txt | ||
tox.ini |
Hardware introspection for OpenStack Ironic
This is an auxiliary service for discovering hardware properties for a node managed by OpenStack Ironic. Hardware introspection or hardware properties discovery is a process of getting hardware parameters required for scheduling from a bare metal node, given it's power management credentials (e.g. IPMI address, user name and password).
A special discovery ramdisk is required to collect the information on a node. The default one can be built using diskimage-builder and ironic-discoverd-ramdisk element (see Configuration below).
Support for ironic-discoverd is present in Tuskar UI --OpenStack Horizon plugin for TripleO.
ironic-discoverd requires OpenStack Juno (2014.2) release or newer.
Please use launchpad to report bugs and ask questions. Use PyPI for downloads and accessing the released version of this README. Refer to CONTRIBUTING.rst for instructions on how to contribute.
Workflow
Usual hardware introspection flow is as follows:
- Operator installs undercloud with ironic-discoverd (e.g. using instack-undercloud).
- Operator enrolls nodes into Ironic either manually or by uploading CSV file to Tuskar UI. Power management credentials should be provided to Ironic at this step.
- Operator sends nodes on introspection either manually using ironic-discoverd HTTP API or again via Tuskar UI.
- On receiving node UUID ironic-discoverd:
- validates node power credentials, current power and provisioning states,
- allows firewall access to PXE boot service for the nodes,
- issues reboot command for the nodes, so that they boot the discovery ramdisk.
- The discovery ramdisk collects the required information and posts it back to ironic-discoverd.
- On receiving data from the discovery ramdisk,
ironic-discoverd:
- validates received data,
- finds the node in Ironic database using it's BMC address (MAC address in case of SSH driver),
- fills missing node properties with received data and creates missing ports.
- Separate HTTP API can be used to query introspection results for a given node.
Starting DHCP server and configuring PXE boot environment is not part of this package and should be done separately.
Installation
ironic-discoverd is available as an RPM from Fedora 22 repositories or from Juno RDO for Fedora 20, 21 and EPEL 7. It will be installed and preconfigured if you used instack-undercloud to build your undercloud. Otherwise after enabling required repositories install it using:
yum install openstack-ironic-discoverd
Alternatively (e.g. if you need the latest version), you can install package from PyPI (you may want to use virtualenv to isolate your environment):
pip install ironic-discoverd
The third way for RPM-based distros is to use ironic-discoverd copr which contains unstable git snapshots of ironic-discoverd.
Configuration
Copy example.conf
to some permanent place
(/etc/ironic-discoverd/discoverd.conf
is what is used in
the RPM). Fill in at least configuration values with names starting with
os_. They configure how ironic-discoverd
authenticates with Keystone.
Also set database option to where you want ironic-discoverd SQLite database to be placed.
See comments inside example.conf for the other possible configuration options.
Note
Configuration file contains a password and thus should be owned by
root
and should have access rights like
0600
.
As for PXE boot environment, you'll need:
TFTP server running and accessible (see below for using dnsmasq). Ensure
pxelinux.0
is present in the TFTP root.Build and put into your TFTP directory kernel and ramdisk from the diskimage-builder ironic-discoverd-ramdisk element:
ramdisk-image-create -o discovery fedora ironic-discoverd-ramdisk
You need diskimage-builder 0.1.38 or newer to do it.
You need PXE boot server (e.g. dnsmasq) running on the same machine as ironic-discoverd. Don't do any firewall configuration: ironic-discoverd will handle it for you. In ironic-discoverd configuration file set
dnsmasq_interface
to the interface your PXE boot server listens on. Here is an example dnsmasq.conf:port=0 interface={INTERFACE} bind-interfaces dhcp-range={DHCP IP RANGE, e.g. 192.168.0.50,192.168.0.150} enable-tftp tftp-root={TFTP ROOT, e.g. /tftpboot} dhcp-boot=pxelinux.0
Configure your
$TFTPROOT/pxelinux.cfg/default
with something like:default discover label discover kernel discovery.kernel append initrd=discovery.initramfs discoverd_callback_url=http://{IP}:5050/v1/continue ipappend 3
Replace
{IP}
with IP of the machine (do not use loopback interface, it will be accessed by ramdisk on a booting machine).Note
There are some prebuilt images which use obsolete
ironic_callback_url
instead ofdiscoverd_callback_url
. Modifypxelinux.cfg/default
accordingly if you have one of these.
Here is discoverd.conf you may end up with:
[discoverd]
debug = false
os_auth_url = http://127.0.0.1:5000/v2.0
os_username = admin
os_password = password
os_tenant_name = admin
dnsmasq_interface = br-ctlplane
Note
Set debug = true
if you want to see complete logs.
Running
If you installed ironic-discoverd from the RPM, you already have a systemd unit, so you can:
systemctl enable openstack-ironic-discoverd
systemctl start openstack-ironic-discoverd
Otherwise run as root
:
ironic-discoverd --config-file /etc/ironic-discoverd/discoverd.conf
Note
Running as root
is not required if
ironic-discoverd does not manage the firewall (i.e.
manage_firewall
is set to false
in the
configuration file).
A good starting point for writing your own systemd unit should be one used in Fedora.
Usage
ironic-discoverd has a simple client library bundled within it. It provides functions:
ironic_discoverd.client.introspect
for starting introspectionironic_discoverd.client.get_status
for querying introspection status
both accepting:
uuid
-
node UUID
base_url
-
optional ironic-discoverd service URL (defaults to
127.0.0.1:5050
) auth_token
-
optional Keystone token.
For testing purposes you can also use it from CLI:
python -m ironic_discoverd.client --auth-token TOKEN introspect UUID
python -m ironic_discoverd.client --auth-token TOKEN get_status UUID
Note
This CLI interface is not stable and may be changed without prior notice. Proper supported CLI is expected later.
HTTP API
By default ironic-discoverd listens on
0.0.0.0:5050
, port can be changed in configuration.
Protocol is JSON over HTTP.
The HTTP API consist of these endpoints:
POST /v1/introspection/<UUID>
initiate hardware discovery for node<UUID>
. All power management configuration for this node needs to be done prior to calling the endpoint.Requires X-Auth-Token header with Keystone token for authentication.
Optional parameters:
new_ipmi_password
if set, ironic-discoverd will try to set IPMI password on the machine to this value. Power credentials validation will be skipped and manual power on will be required. See Setting IPMI credentials for details.new_ipmi_username
provides new IPMI user name in addition to password set bynew_ipmi_password
. Defaults to currentipmi_username
in nodedriver_info
field.
Response:
- 202 - accepted discovery request
- 400 - bad request
- 401, 403 - missing or invalid authentication
- 404 - node cannot be found
Client library function:
ironic_discoverd.client.introspect
.GET /v1/introspection/<UUID>
get hardware discovery status.Requires X-Auth-Token header with Keystone token for authentication.
Response:
- 200 - OK
- 400 - bad request
- 401, 403 - missing or invalid authentication
- 404 - node cannot be found
Response body: JSON dictionary with keys:
finished
(boolean) whether discovery is finishederror
error string ornull
Client library function:
ironic_discoverd.client.get_status
.POST /v1/continue
internal endpoint for the discovery ramdisk to post back discovered data. Should not be used for anything other than implementing the ramdisk. Request body: JSON dictionary with at least these keys:cpus
number of CPUcpu_arch
architecture of the CPUmemory_mb
RAM in MiBlocal_gb
hard drive size in GiBinterfaces
dictionary filled with data from all NIC's, keys being interface names, values being dictionaries with keys:mac
MAC addressip
IP address
boot_interface
optional MAC address of the NIC that the machine PXE booted from either in standard format11:22:33:44:55:66
or in PXELinuxBOOTIF
format01-11-22-33-44-55-66
.block_devices
optional block devices information forroot_device_hint
plugin, dictionary with keys:serials
list of serial numbers of block devices.
Note
This list highly depends on enabled plugins, provided above are expected keys for the default set of plugins. See Plugins for details.
Response:
- 200 - OK
- 400 - bad request
- 403 - node is not on introspection
- 404 - node cannot be found or multiple nodes found
Response body: JSON dictionary. If Setting IPMI Credentials is requested, body will contain the following keys:
ipmi_setup_credentials
booleanTrue
ipmi_username
new IPMI user nameipmi_password
new IPMI password
Setting IPMI Credentials
If you have physical access to your nodes, you can use ironic-discoverd to set IPMI credentials for them without knowing the original ones. The workflow is as follows:
- Ensure nodes will PXE boot on the right network by default.
- Set
enable_setting_ipmi_credentials = true
in the ironic-discoverd configuration file. - Enroll nodes in Ironic with setting their
ipmi_address
only. This step allows ironic-discoverd to distinguish nodes. - Set maintenance mode on nodes. That's an important step, otherwise Ironic might interfer with introspection process.
- Start introspection with providing additional parameters:
new_ipmi_password
IPMI password to set,new_ipmi_username
IPMI user name to set, defaults to one in node driver_info.
- Manually power on the nodes and wait.
- After introspection is finished (watch nodes power state or use ironic-discoverd status API) you can turn maintenance mode off.
Note that due to various limitations on password value in different BMC, ironic-discoverd will only accept passwords with length between 1 and 20 consisting only of letters and numbers.
Plugins
ironic-discoverd heavily relies on plugins for data
processing. Even the standard functionality is largely based on plugins.
Set processing_hooks
option in the configuration file to
change the set of plugins to be run on introspection data. Note that
order does matter in this option.
These are plugins that are enabled by default and should not be disabled, unless you understand what you're doing:
scheduler
-
validates and updates basic hardware scheduling properties: CPU number and architecture, memory and disk size.
validate_interfaces
-
validates network interfaces information.
Here are some plugins that can be additionally enabled:
ramdisk_error
-
reports error, if
error
field is set by the ramdisk. example
-
example plugin logging it's input and output.
root_device_hint
-
gathers block devices from ramdisk and exposes root device in multiple runs.
Refer to CONTRIBUTING.rst for information on how to write your own plugin.
Release Notes
1.1 Series
See 1.1.0 release tracking page for details.
Upgrade Notes
- This version no longer supports ancient ramdisks that sent
macs
instead ofinterfaces
. It also raises exception if no valid interfaces were found after processing.
Major Features
Default to only creating a port for the NIC that the ramdisk was PXE booted from, if such information is provided by ramdisk as
boot_interface
field. Adjustable byonly_pxe_booting_port
option.See better-boot-interface-detection blueprint for details.
Setting IPMI Credentials feature is considered stable now and is exposed in the client. It still needs to be enabled via configuration.
See setup-ipmi-credentials-take2 blueprint for what changed since 1.0.0 (tl;dr: everything).
Other Changes
Experimental plugin
edeploy
to use with eDeploy hardware detection and classification utilities.See eDeploy blueprint for details.
Plugin
root_device_hint
for in-band root device discovery.
Known Issues
1.0 Series
1.0 is the first feature-complete release series. It's also the first series to follow standard OpenStack processes from the beginning. All 0.2 series users are advised to upgrade.
See 1.0.0 release tracking page for details.
Upgrade notes
Action required:
- Fill in
database
option in the configuration file before upgrading. - Stop relying on ironic-discoverd setting maintenance mode itself.
- Stop relying on
discovery_timestamp
node extra field.
Action recommended:
- Switch your init scripts to use
ironic-discoverd --config-file <path>
instead of justironic-discoverd <path>
. - Stop relying on
on_discovery
andnewly_discovered
being set in nodeextra
field during and after introspection. Use new get status HTTP endpoint and client API instead. - Switch from
discover
tointrospect
HTTP endpoint and client API.
Major features
Introspection now times out by default, set
timeout
option to alter.New API
GET /v1/introspection/<uuid>
andclient.get_status
for getting discovery status.See get-status-api blueprint for details.
New API
POST /v1/introspection/<uuid>
andclient.introspect
is now used to initiate discovery,/v1/discover
is deprecated.See v1 API reform blueprint for details.
/v1/continue
is now sync:- Errors are properly returned to the caller
- This call now returns value as a JSON dict (currently empty)
Add support for plugins that hook into data processing pipeline. Refer to Plugins for information on bundled plugins and to CONTRIBUTING.rst for information on how to write your own.
See plugin-architecture blueprint for details.
Support for OpenStack Kilo release and new Ironic state machine -see Kilo state machine blueprint.
As a side effect, no longer depend on maintenance mode for introspection. Stop putting node in maintenance mode before introspection.
Cache nodes under introspection in a local SQLite database.
database
configuration option sets where to place this database. Improves performance by making less calls to Ironic API and makes possible to get results of introspection.
Other Changes
Firewall management can be disabled completely via
manage_firewall
option.Experimental support for updating IPMI credentials from within ramdisk.
Enable via configuration option
enable_setting_ipmi_credentials
. Beware that this feature lacks proper testing, is not supported officially yet and is subject to changes without keeping backward compatibility.See setup-ipmi-credentials blueprint for details.
Known Issues
- bug
1415040 it is required to set IP addresses instead of host names in
ipmi_address
/ilo_address
/drac_host
nodedriver_info
field for ironic-discoverd to work properly.
0.2 Series
0.2 series is designed to work with OpenStack Juno release. The major changes are:
API
- Authentication via Keystone for
/v1/discover
. - Expect
interfaces
instead ofmacs
in post-back from the ramdisk [version 0.2.1]. - If
interfaces
is present, only add ports for NIC's with IP address set [version 0.2.1]. /v1/discover
now does some sync sanity checks [version 0.2.2].- Nodes will be always put into maintenance mode before discovery [version 0.2.1].
Configuration
- Periodic firewall update is now configurable.
- On each start-up make several attempts to check that Ironic is available [version 0.2.2].
Misc
- Simple client in
ironic_discoverd.client
. - Preliminary supported for Python 3.3 (real support depends on Eventlet).
0.1 Series
First stable release series. Not supported any more.