Modern Python API to Red Hat's Errata Tool.
python-errata-tool is a Python library that wraps the Errata Tool's REST API.
It uses requests_gssapi
to authenticate and parses JSON responses into Erratum
objects. You can
use it to create new advisories, or read and update existing advisories. The
ErratumConnector
class also provides lower-level access to all of the
Errata Tool's REST API.
Example:
from errata_tool import Erratum
e = Erratum(errata_id=1234)
print(e.errata_state)
# prints "NEW_FILES"
print(e.url())
# prints "https://errata.devel.redhat.com/advisory/1234"
Creating a new bugfix advisory:
e = Erratum(product='RHCEPH',
release='rhceph-2.1',
#errata_type='RHBA' # Default; may be omitted
synopsis='Red Hat Ceph Storage 2.1 bug fix update',
topic='An update for Red Hat Ceph 2.1 is now available.',
description='This update contains the following fixes ...',
solution='Before applying this update...',
qe_email='[email protected]',
qe_group='RHC (Ceph) QE',
owner_email='[email protected]',
manager_email='[email protected]',
)
e.commit()
print(e.url())
Creating a new enhancement (feature) advisory:
e = Erratum(product='RHCEPH',
release='rhceph-2.1',
errata_type='RHEA', # Set to RHEA for RHEA
synopsis='Red Hat Ceph Storage 2.1 enhancement update',
topic='An update for Red Hat Ceph 2.1 is now available.',
description='This update contains the following features ...',
solution='Before applying this update...',
qe_email='[email protected]',
qe_group='RHC (Ceph) QE',
owner_email='[email protected]',
manager_email='[email protected]',
)
e.commit()
print(e.url())
Creating a new security advisory. Note that RHSA (Security) advisories are given one of four impacts (Low, Moderate, Important, and Critical). See this link for more information: https://access.redhat.com/security/updates/classification
e = Erratum(product='RHCEPH',
release='rhceph-2.1',
errata_type='RHSA', # Set to RHSA for RHSA
security_impact='Moderate', # Required for RHSA
synopsis='Red Hat Ceph Storage 2.1 security update',
topic='An update for Red Hat Ceph 2.1 is now available.',
description='This update contains the following fixes ...',
solution='Before applying this update...',
qe_email='[email protected]',
qe_group='RHC (Ceph) QE',
owner_email='[email protected]',
manager_email='[email protected]',
)
e.commit()
print(e.url())
The errata-tool
CLI is a thin wrapper around the classes. You can use it to
query information from the Errata Tool or create new releases (releng):
errata-tool -h usage: errata-tool [-h] [--stage] [--dry-run] {advisory,product,release} ... positional arguments: {advisory,product,release} advisory Get or create an advisory product Get a product release Get or create a release (RCM) optional arguments: --stage use staging ET instance --dry-run show what would happen, but don't do it
Waiting and conditionally pushing an advisory:
As a release engineer one often checks the status of an advisory if it is ready
to be pushed. To avoid human polling of the state and to automate the advisory
push two options are provided under the errata-tool advisory push
option:
errata-tool advisory push --help usage: errata-tool advisory push [-h] [--target {stage,live}] [--wait-for-state {SHIPPED_LIVE,PUSH_READY}] [--push-when-ready] [--verbose] errata_id positional arguments: errata_id advisory id, "12345" optional arguments: -h, --help show this help message and exit --target {stage,live} stage (default) or live --wait-for-state {SHIPPED_LIVE,PUSH_READY} state : PUSH_READY or SHIPPED_LIVE --push-when-ready Push if the advisory enters state PUSH_READY --verbose print current state of the advisory
The --wait-for-state
option polls at regular interval for the advisory to
enter one of the two desired states - PUSH_READY or SHIPPED_LIVE
when the advisory reaches that state the polling stops and the script exits with
a successful exit code $? eq 0
.
Caveat: The script will wait forever until that state is reached or interrupted by the user. The option to have a cap on the wait time was dropped to keep the usage simple and for the lack of a compelling use case.
The --push-when-ready
option pushes the advisory if it is in PUSH_READY
state. The --push-when-ready
option can be used with --wait-for-state
option to repeatedly poll the advisory until it reaches PUSH_READY state before
pushing it.
Here are some usecases:
- usecase 1: Push advisory if it is in state PUSH_READY state
errata-tool --stage advisory push --target live --push-when-ready 12345
- usecase 2: Wait for advisory to enter PUSH_READY state and push the advisory
errata-tool --stage advisory push --target live --push-when-ready \
--wait-for-state PUSH_READY 12345
- usecase 3: Wait for advisory to enter SHIPPED_LIVE, push the advisory if
- it enters PUSH_READY state while waiting.
errata-tool --stage advisory push --target live --push-when-ready \
--wait-for-state SHIPPED_LIVE 12345
- usecase 4: Push independent advisories before pushing those which are
dependent on the independent advisories
# Ship advisory 12346 after shipping 12345
errata-tool --stage advisory push --target live --push-when-ready \
--wait-for-state SHIPPED_LIVE 12345 && \
errata-tool --stage advisory push --target live --push-when-ready \
--wait-for-state PUSH_READY 12346
Getting an erratum's name:
e = Erratum(errata_id=22986)
print(e.errata_name)
# prints "RH*A-YYYY:NNNNN", for example "RHBA-2018:12345"
Adding bugs:
e = Erratum(errata_id=22986)
e.addBugs([12345, 123678])
e.commit()
# You can read the current list of bugs with the "e.errata_bugs" property.
Removing bugs:
e = Erratum(errata_id=22986)
e.removeBugs([12345, 123678])
# You can simply call "commit()" without checking the return code, or check
# it and use refresh() to refresh our local instance data for the errata
# advisory.
need_refresh = e.commit()
if need_refresh:
print('refreshing')
e.refresh()
Checking whether an advisory is embargoed:
e = Erratum(errata_id=22986)
if e.embargoed:
# it's embargoed
else:
# it's not embargoed
Checking whether an advisory is text-only:
e = Erratum(errata_id=24075)
if e.text_only:
# it's text-only
# If it's an RHSA, you may want to get/set e.text_only_cpe here.
else:
# it's not text-only
Adding builds:
e = Erratum(errata_id=24075)
# The "release" kwarg is the Errata Tools's "product version" in
# composedb, for example "RHEL-7-CEPH-2".
e.addBuilds(['ceph-10.2.3-17.el7cp'], release='RHEL-7-CEPH-2')
Adding container builds:
e = Erratum(errata_id=34279)
# For non-RPM Brew builds, you must specify the file_types kwarg.
# For container builds, this is "tar".
e.addBuilds('rhceph-rhel7-container-3-9',
release='RHEL-7-CEPH-3',
file_types={'rhceph-rhel7-container-3-9': ['tar']})
Changing state:
e = Erratum(errata_id=24075)
e.setState('QE')
e.commit()
Changing docs reviewer:
e = Erratum(errata_id=24075)
e.changeDocsReviewer('[email protected]')
Adding someone to the CC list:
e = Erratum(errata_id=24075)
e.addCC('[email protected]')
Changing an advisory type:
e = Erratum(errata_id=33840)
e.update(errata_type='RHBA')
e.commit()
Reloading the all specific builds that lack product listings:
e = Erratum(errata_id=24075)
if e.missing_product_listings: # a (possibly-empty) list of build NVRs
result = e.reloadBuilds(no_rpm_listing_only=True)
# result is a dict for this job tracker
Determining if an advisory has RPMs or containers:
e = Erratum(errata_id=24075)
content_types = e.content_types
# result is a list, like ["rpm"], or ["docker"]
Get active RPMDiff results for an advisory:
e = Erratum(errata_id=24075)
bad = []
for result in e.externalTests(test_type='rpmdiff'):
if result['attributes']['status'] not in ('PASSED', 'WAIVED'):
# See result['attributes']['external_id'] for the integer to pass
# into RPMDiff's run API.
bad.append(result)
Set the CDN repos for a container advisory (only applies for advisories containing Docker images):
e = Erratum(errata_id=24075)
assert 'docker' in e.content_types
e.metadataCdnRepos(enable=['rhel-7-server-rhceph-3-mon-rpms__x86_64'])
Same thing, but for text-only advisories:
e = Erratum(errata_id=24075)
assert e.text_only
e.textOnlyRepos(enable=['rhel-7-server-rhceph-3-mon-rpms__x86_64'])
The errata_tool.product.Product
class can look up existing products.
Looking up a product:
from errata_tool.product import Product
p = Product('RHCEPH')
print(p.id) # 104
print(p.name) # "RHCEPH"
print(p.description) # "Red Hat Ceph Storage"
The errata_tool.release.Release
class can look up existing releases or
create new release entries.
Looking up a release:
from errata_tool.release import Release
r = Release(name='rhceph-2.4')
print(r.id) # 792
print(r.name) # "rhceph-2.4"
print(r.description) # "Red Hat Ceph Storage 2.4"
print(r.type) # "QuarterlyUpdate"
print(r.is_active) # True
print(r.enabled) # True
print(r.blocker_flags) # ['ceph-2.y', 'pm_ack', 'devel_ack', 'qa_ack']
print(r.edit_url) # https://errata.devel.redhat.com/release/edit/792
Finding all "NEW_FILES" advisories for a release:
from errata_tool.release import Release
rel = Release(name='rhceph-3.0')
advisories = rel.advisories()
new_files = [a for a in advisories if a['status'] == 'NEW_FILES']
print(new_files) # prints the list of advisories' data
Creating a new release (this requires the "releng" role in the Errata Tool):
from errata_tool.release import Release
r = Release.create(
name='rhceph-3.0',
product='RHCEPH',
product_versions=['RHEL-7-CEPH-3'],
type='QuarterlyUpdate',
program_manager='anharris',
blocker_flags='ceph-3.0',
default_brew_tag='ceph-3.0-rhel-7-candidate',
)
print('created new rhceph-3.0 release')
print('visit %s to edit further' % r.edit_url)
To use the staging Errata Tool environment without affecting production, set
the ErrataConnector._url
member variable to the staging URL.
from errata_tool import ErrataConnector, Erratum
ErrataConnector._url = 'https://errata.stage.engineering.redhat.com/'
# Now try something like creating an advisory, and it will not show up in
# prod, or bother people with emails, etc.
e = Erratum(product='RHCEPH',
release='rhceph-2.1',
synopsis='Red Hat Ceph Storage 2.1 bug fix update',
...
)
e.commit()
Maybe your application makes many API calls (lots of advisories, builds, etc), When processing large numbers of errata from higher-level tools, it's helpful to understand where the time is spent to see if multiple calls can be avoided.
Set ErrataConnector.debug = True
, and then your connector object will
record information about each call it makes. Each GET/PUT/POST is recorded,
along with totals / mean / min / max.
URL APIs are deduplicated based on their name, so two calls to different errata on the same API is recorded as a single API.
To extract the information and print it, one might use PrettyTable:
e = Erratum(errata_id=24075)
pt = PrettyTable()
for c in ErrataConnector.timings:
for u in ErrataConnector.timings[c]:
pt.add_row([c, u,
ErrataConnector.timings[c][u]['count'],
ErrataConnector.timings[c][u]['total'],
ErrataConnector.timings[c][u]['mean'],
ErrataConnector.timings[c][u]['min'],
ErrataConnector.timings[c][u]['max']])
print(pt.get_string())
This library verifies the ET server's HTTPS certificate by default. This is more of a python-requests thing, but if you receive an SSL verification error, it's probably because you don't have the Red Hat IT CA set up for your Python environment. Particularly if you're running this in a virtualenv, you'll want to set the following configuration variable:
REQUESTS_CA_BUNDLE=/etc/pki/ca-trust/source/anchors/RH-IT-Root-CA.crt
Where "RH-IT-Root-CA.crt" is the public cert that signed the ET server's HTTPS certificate.
When using RHEL 7's python-requests RPM, requests simply checks
/etc/pki/tls/certs/ca-bundle.crt
, so you'll need to add the IT CA cert to
that big bundle file.
If you've already added the Red Hat IT CA to your system-wide bundle, you can have your Python code always use this file:
if 'REQUESTS_CA_BUNDLE' not in os.environ:
os.environ['REQUESTS_CA_BUNDLE'] = '/etc/pki/tls/certs/ca-bundle.crt'
This will make requests behave the same inside or outside your virtualenv. In other words, with this code, your program will always validate the Red Hat IT CA.
Install fedpkg, then use the Makefile:
$ make srpm
You can then upload the SRPM to Copr. Or, to build RPMs on your local computer, using mock:
$ make rpm
Check out the CHANGELOG.