mirror of
https://github.com/valitydev/salt.git
synced 2024-11-07 08:58:59 +00:00
330 lines
9.8 KiB
ReStructuredText
330 lines
9.8 KiB
ReStructuredText
================================
|
|
Miscellaneous Salt Cloud Options
|
|
================================
|
|
|
|
This page describes various miscellaneous options available in Salt Cloud
|
|
|
|
Deploy Script Arguments
|
|
=======================
|
|
Custom deploy scripts are unlikely to need custom arguments to be passed to
|
|
them, but salt-bootstrap has been extended quite a bit, and this may be
|
|
necessary. script_args can be specified in either the profile or the map file,
|
|
to pass arguments to the deploy script:
|
|
|
|
.. code-block:: yaml
|
|
|
|
ec2-amazon:
|
|
provider: ec2
|
|
image: ami-1624987f
|
|
size: t1.micro
|
|
ssh_username: ec2-user
|
|
script: bootstrap-salt
|
|
script_args: -c /tmp/
|
|
|
|
This has also been tested to work with pipes, if needed:
|
|
|
|
.. code-block:: yaml
|
|
|
|
script_args: | head
|
|
|
|
|
|
Selecting the File Transport
|
|
============================
|
|
By default, Salt Cloud uses SFTP to transfer files to Linux hosts. However, if
|
|
SFTP is not available, or specific SCP functionality is needed, Salt Cloud can
|
|
be configured to use SCP instead.
|
|
|
|
.. code-block:: yaml
|
|
|
|
file_transport: sftp
|
|
file_transport: scp
|
|
|
|
|
|
Sync After Install
|
|
==================
|
|
Salt allows users to create custom modules, grains, and states which can be
|
|
synchronised to minions to extend Salt with further functionality.
|
|
|
|
This option will inform Salt Cloud to synchronise your custom modules, grains,
|
|
states or all these to the minion just after it has been created. For this to
|
|
happen, the following line needs to be added to the main cloud
|
|
configuration file:
|
|
|
|
.. code-block:: yaml
|
|
|
|
sync_after_install: all
|
|
|
|
The available options for this setting are:
|
|
|
|
.. code-block:: yaml
|
|
|
|
modules
|
|
grains
|
|
states
|
|
all
|
|
|
|
|
|
Setting up New Salt Masters
|
|
===========================
|
|
It has become increasingly common for users to set up multi-hierarchal
|
|
infrastructures using Salt Cloud. This sometimes involves setting up an
|
|
instance to be a master in addition to a minion. With that in mind, you can
|
|
now lay down master configuration on a machine by specifying master options
|
|
in the profile or map file.
|
|
|
|
.. code-block:: yaml
|
|
|
|
make_master: True
|
|
|
|
This will cause Salt Cloud to generate master keys for the instance, and tell
|
|
salt-bootstrap to install the salt-master package, in addition to the
|
|
salt-minion package.
|
|
|
|
The default master configuration is usually appropriate for most users, and
|
|
will not be changed unless specific master configuration has been added to the
|
|
profile or map:
|
|
|
|
.. code-block:: yaml
|
|
|
|
master:
|
|
user: root
|
|
interface: 0.0.0.0
|
|
|
|
|
|
Delete SSH Keys
|
|
===============
|
|
When Salt Cloud deploys an instance, the SSH pub key for the instance is added
|
|
to the known_hosts file for the user that ran the salt-cloud command. When an
|
|
instance is deployed, a cloud provider generally recycles the IP address for
|
|
the instance. When Salt Cloud attempts to deploy an instance using a recycled
|
|
IP address that has previously been accessed from the same machine, the old key
|
|
in the known_hosts file will cause a conflict.
|
|
|
|
In order to mitigate this issue, Salt Cloud can be configured to remove old
|
|
keys from the known_hosts file when destroying the node. In order to do this,
|
|
the following line needs to be added to the main cloud configuration file:
|
|
|
|
.. code-block:: yaml
|
|
|
|
delete_sshkeys: True
|
|
|
|
|
|
Keeping /tmp/ Files
|
|
===================
|
|
When Salt Cloud deploys an instance, it uploads temporary files to /tmp/ for
|
|
salt-bootstrap to put in place. After the script has run, they are deleted. To
|
|
keep these files around (mostly for debugging purposes), the --keep-tmp option
|
|
can be added:
|
|
|
|
.. code-block:: bash
|
|
|
|
salt-cloud -p myprofile mymachine --keep-tmp
|
|
|
|
For those wondering why /tmp/ was used instead of /root/, this had to be done
|
|
for images which require the use of sudo, and therefore do not allow remote
|
|
root logins, even for file transfers (which makes /root/ unavailable).
|
|
|
|
|
|
Hide Output From Minion Install
|
|
===============================
|
|
By default Salt Cloud will stream the output from the minion deploy script
|
|
directly to STDOUT. Although this can been very useful, in certain cases you
|
|
may wish to switch this off. The following config option is there to enable or
|
|
disable this output:
|
|
|
|
.. code-block:: yaml
|
|
|
|
display_ssh_output: False
|
|
|
|
|
|
Connection Timeout
|
|
==================
|
|
|
|
There are several stages when deploying Salt where Salt Cloud needs to wait for
|
|
something to happen. The VM getting it's IP address, the VM's SSH port is
|
|
available, etc.
|
|
|
|
If you find that the Salt Cloud defaults are not enough and your deployment
|
|
fails because Salt Cloud did not wait log enough, there are some settings you
|
|
can tweak.
|
|
|
|
.. admonition:: Note
|
|
|
|
All values should be provided in seconds
|
|
|
|
|
|
You can tweak these settings globally, per cloud provider, or event per profile
|
|
definition.
|
|
|
|
|
|
wait_for_ip_timeout
|
|
~~~~~~~~~~~~~~~~~~~
|
|
|
|
The amount of time Salt Cloud should wait for a VM to start and get an IP back
|
|
from the cloud provider. Default: 5 minutes.
|
|
|
|
|
|
wait_for_ip_interval
|
|
~~~~~~~~~~~~~~~~~~~~
|
|
|
|
The amount of time Salt Cloud should sleep while querying for the VM's IP.
|
|
Default: 5 seconds.
|
|
|
|
|
|
ssh_connect_timeout
|
|
~~~~~~~~~~~~~~~~~~~
|
|
|
|
The amount of time Salt Cloud should wait for a successful SSH connection to
|
|
the VM. Default: 5 minutes.
|
|
|
|
|
|
wait_for_passwd_timeout
|
|
~~~~~~~~~~~~~~~~~~~~~~~
|
|
|
|
The amount of time until an ssh connection can be established via password or
|
|
ssh key. Default 15 seconds.
|
|
|
|
|
|
wait_for_passwd_maxtries
|
|
~~~~~~~~~~~~~~~~~~~~~~~~
|
|
|
|
The number of attempts to connect to the VM until we abandon.
|
|
Default 15 attempts
|
|
|
|
|
|
wait_for_fun_timeout
|
|
~~~~~~~~~~~~~~~~~~~~
|
|
|
|
Some cloud drivers check for an available IP or a successful SSH connection
|
|
using a function, namely, SoftLayer, and SoftLayer-HW. So, the amount of time
|
|
Salt Cloud should retry such functions before failing. Default: 5 minutes.
|
|
|
|
|
|
wait_for_spot_timeout
|
|
~~~~~~~~~~~~~~~~~~~~~
|
|
|
|
The amount of time Salt Cloud should wait before an EC2 Spot instance is
|
|
available. This setting is only available for the EC2 cloud driver.
|
|
|
|
|
|
Salt Cloud Cache
|
|
================
|
|
|
|
Salt Cloud can maintain a cache of node data, for supported providers. The
|
|
following options manage this functionality.
|
|
|
|
|
|
update_cachedir
|
|
~~~~~~~~~~~~~~~
|
|
|
|
On supported cloud providers, whether or not to maintain a cache of nodes
|
|
returned from a --full-query. The data will be stored in ``msgpack`` format
|
|
under ``<SALT_CACHEDIR>/cloud/active/<DRIVER>/<PROVIDER>/<NODE_NAME>.p``. This
|
|
setting can be True or False.
|
|
|
|
|
|
diff_cache_events
|
|
~~~~~~~~~~~~~~~~~
|
|
|
|
When the cloud cachedir is being managed, if differences are encountered
|
|
between the data that is returned live from the cloud provider and the data in
|
|
the cache, fire events which describe the changes. This setting can be True or
|
|
False.
|
|
|
|
Some of these events will contain data which describe a node. Because some of
|
|
the fields returned may contain sensitive data, the ``cache_event_strip_fields``
|
|
configuration option exists to strip those fields from the event return.
|
|
|
|
.. code-block:: yaml
|
|
|
|
cache_event_strip_fields:
|
|
- password
|
|
- priv_key
|
|
|
|
The following are events that can be fired based on this data.
|
|
|
|
|
|
salt/cloud/minionid/cache_node_new
|
|
**********************************
|
|
A new node was found on the cloud provider which was not listed in the cloud
|
|
cachedir. A dict describing the new node will be contained in the event.
|
|
|
|
|
|
salt/cloud/minionid/cache_node_missing
|
|
**************************************
|
|
A node that was previously listed in the cloud cachedir is no longer available
|
|
on the cloud provider.
|
|
|
|
|
|
salt/cloud/minionid/cache_node_diff
|
|
***********************************
|
|
One or more pieces of data in the cloud cachedir has changed on the cloud
|
|
provider. A dict containing both the old and the new data will be contained in
|
|
the event.
|
|
|
|
|
|
SSH Known Hosts
|
|
===============
|
|
|
|
Normally when bootstrapping a VM, salt-cloud will ignore the SSH host key. This
|
|
is because it does not know what the host key is before starting (because it
|
|
doesn't exist yet). If strict host key checking is turned on without the key
|
|
in the ``known_hosts`` file, then the host will never be available, and cannot
|
|
be bootstrapped.
|
|
|
|
If a provider is able to determine the host key before trying to bootstrap it,
|
|
that provider's driver can add it to the ``known_hosts`` file, and then turn on
|
|
strict host key checking. This can be set up in the main cloud configuration
|
|
file (normally ``/etc/salt/cloud``) or in the provider-specific configuration
|
|
file:
|
|
|
|
.. code-block:: yaml
|
|
|
|
known_hosts_file: /path/to/.ssh/known_hosts
|
|
|
|
If this is not set, it will default to ``/dev/null``, and strict host key
|
|
checking will be turned off.
|
|
|
|
It is highly recommended that this option is *not* set, unless the user has
|
|
verified that the provider supports this functionality, and that the image
|
|
being used is capable of providing the necessary information. At this time,
|
|
only the EC2 driver supports this functionality.
|
|
|
|
SSH Agent
|
|
=========
|
|
|
|
.. versionadded:: 2015.5.0
|
|
|
|
If the ssh key is not stored on the server salt-cloud is being run on, set
|
|
ssh_agent, and salt-cloud will use the forwarded ssh-agent to authenticate.
|
|
|
|
.. code-block:: yaml
|
|
|
|
ssh_agent: True
|
|
|
|
File Map Upload
|
|
===============
|
|
|
|
.. versionadded:: 2014.7.0
|
|
|
|
The ``file_map`` option allows an arbitrary group of files to be uploaded to the
|
|
target system before running the deploy script. This functionality requires a
|
|
provider uses salt.utils.cloud.bootstrap(), which is currently limited to the ec2,
|
|
gce, openstack and nova drivers.
|
|
|
|
The ``file_map`` can be configured globally in ``/etc/salt/cloud``, or in any cloud
|
|
provider or profile file. For example, to upload an extra package or a custom deploy
|
|
script, a cloud profile using ``file_map`` might look like:
|
|
|
|
.. code-block:: yaml
|
|
|
|
ubuntu14:
|
|
provider: ec2-config
|
|
image: ami-98aa1cf0
|
|
size: t1.micro
|
|
ssh_username: root
|
|
securitygroup: default
|
|
file_map:
|
|
/local/path/to/custom/script: /remote/path/to/use/custom/script
|
|
/local/path/to/package: /remote/path/to/store/package
|