pyinfra turns Python code into shell commands and runs them on your servers. Execute ad-hoc commands and write declarative operations. Target SSH servers, local machine and Docker containers. Fast and scales from one server to thousands.
server.service operationfiles.* factsTMPDIR when asking for sudo password (@jaysoffian)python.* operationsAdded:
selinux.boolean, selinux.file_context, selinux.file_context_mapping & selinux.port operations (@morrison12)selinux.SEBoolean, selinux.FileContextMapping, selinux.SEPorts, selinux.SEPort facts (@morrison12)snap.package operation & snap.SnapPackage, snap.SnapPackages facts (@pabloxio)files.block operation implementation (@morrison12)Fixed:
systemd.SystemdStatus fact (@mariusmuja)rpm.RpmPackage fact (@mariusmuja)host.loop methodserver.Date fact, should resolve any outstanding parse errorsOther changes:
macos-latest GitHub runner (@morrison12)Delayed getting this out, lots of little improvements.
Added:
server.user_authorized_keys operation_continue_on_error argumentdir_mode argument to files.sync operation (@filips123)mode=True in files.put operationheaders and insecure arguments to files.download operationFixed:
lxd.container operation (@zachwaite)files.put operationOther changes:
git.bare_repo operationpathlib objectsRelatively small quick release with two additions and a bunch of fixes.
Added:
create_home argument to server.user operationstdin
Fixed:
assume_exists=True in files.put operationInternal:
The main feature of 2.2 is the switch to using a DAG to generate operation order. This mostly replaces line-number ordering (still used to tie-break) and means hacks such as state.preserve_loop_order are no longer required!
The second highlight feature is the inclusion of type stub files for operations that include all of the global arguments. Thank you to @StefanBRas for implementing this.
Other changes:
server.user operation (@yunzheng)replace as a whole line in files.line operationmysql.privileges invalid argument requesting MysqlUserGrants factFirst 2.x point release! Major feature: nested operations (at last!).
Based on the changes to operations in 2.x nested operations make it possible to generate & execute operations on the fly at execution time, rather than using the low-level connector API. This unlocks all kinds of complex deploys that were previously impossible or complex to implement. Let's look at an example:
from pyinfra import logger
from pyinfra.operations import python, server
def callback():
result = server.shell(commands=["echo output"])
logger.info(f"Got result: {result.stdout}")
python.call(
name="Execute callback function",
function=callback,
)
Other new stuff:
host.reload_fact(...) - bypasses the fact cache to force reloading of fact datadeb.DebArch factssh_paramiko_connect_kwargs host data used in the @ssh connectorBugfixes:
files.put hashing local file that doesn't exist when assume_exists=True
zypper over apt when both present in server.packages operationInternal changes:
black and isort across the codebase, now part of CIconfig.SUDO/etc handling for operation argument defaultssystemd facts using systemctl show
files.template operationaccept-new SSH config for StrictHostKeyChecking
config.REQUIRE_PYINFRA_VERSION & config.REQUIRE_PACKAGES handlingpyinfra automates/provisions/manages/deploys infrastructure. It can be used for ad-hoc command execution, service deployment, configuration management and more. See the readme for more information.
The first 2.x release! Like v0 -> v1 this release mostly removes legacy APIs and methods which show warnings in v1. Major changes:
Breaking: Python 2.7 (finally!), 3.5 support dropped, Python 3.6 is now the minimum required version.
Breaking: the "deploy directory" concept has been removed - everything now executes from the current working directory which removes the ambiguous magic v1 used to pick a deploy directory. A new --chdir CLI flag has been added to set the working directory before pyinfra executes.
This may affect scripts or CI workflows currently setup, for example:
# Old v1, deploy directory becomes deploys/elasticsearch/
pyinfra deploys/elasticsearch/inventories/production.py deploys/elasticsearch/deploy.py
# New v2, explicit chdir required
pyinfra --chdir deploys/elasticsearch/ inventories/production.py deploy.py
Parallel operation generation & facts rewrite - this is a huge improvement to how pyinfra generates commands to run on target hosts. This is now run in parallel across all hosts. Facts are now collected by individual host rather than across all hosts which may yield significant speedups in certain situations.
This change also brings support for all of the execution global arguments to facts, and hugely simplifies the facts implementation. Global arguments will now be read from host data in exactly the same way they are for operations, which was often a confusing gotcha in v1. This also means that the arguments can have different values for each host and this will not cause issues.
Other breaking changes (warnings shown in v1 for most):
AttributeError when accessed via host.data.X
branch argument to None in git.repo operationpresent argument removed from mysql.privileges operationconfig objecthost.fact.fact_name access has been removedinit.* operations have been removed--facts and --operations CLI flags--debug-data CLI flagWindows prefix on all Windows factsname argument to path in windows_files.* operationspyinfra.modules module/importconfig.MIN_PYINFRA_VERSION
branch and create_branch arguments in git.worktree operationtouch_periodic argument in apt.update operation (never used)pyinfra.api.connectors module moved to pyinfra.connectors
Deprecated (showing warnings, to be removed in v3):
state and host arguments no longer need to be passed into operation or deploy functionspostgresql_* arguments renamed to psql_* in postgresql.* operations & facts