Skip to content
This repository has been archived by the owner on Jan 19, 2022. It is now read-only.

Commit

Permalink
Merge b0067da into 2b43e8c
Browse files Browse the repository at this point in the history
  • Loading branch information
lundjordan committed Jun 29, 2015
2 parents 2b43e8c + b0067da commit 2925b67
Show file tree
Hide file tree
Showing 11 changed files with 576 additions and 0 deletions.
1 change: 1 addition & 0 deletions relengapi/app.py
Original file line number Diff line number Diff line change
Expand Up @@ -83,6 +83,7 @@ def _load_bp(n):
'slaveloan',
'tokenauth',
'tooltool',
'archiver',
]]


Expand Down
118 changes: 118 additions & 0 deletions relengapi/blueprints/archiver/__init__.py
Original file line number Diff line number Diff line change
@@ -0,0 +1,118 @@
# This Source Code Form is subject to the terms of the Mozilla Public
# License, v. 2.0. If a copy of the MPL was not distributed with this
# file, You can obtain one at http://mozilla.org/MPL/2.0/.

import logging

from random import randint

from flask import Blueprint
from flask import current_app
from flask import redirect
from flask import url_for
from relengapi.blueprints.archiver.tasks import create_and_upload_archive
from relengapi.blueprints.archiver.types import MozharnessArchiveTask
from relengapi.lib import api

bp = Blueprint('archiver', __name__)
log = logging.getLogger(__name__)

GET_EXPIRES_IN = 300


@bp.route('/status/<task_id>')
@api.apimethod(MozharnessArchiveTask, unicode)
def task_status(task_id):
"""
Check and return the current state of the create_and_upload_archive celery task with task id
of <task_id>.
If the task is unknown, state will be PENDING. Once the task starts it will be updated to
STARTED and finally, if it completes, it will be either SUCCESS (no exceptions), or FAILURE.
See update_state() within create_and_upload_archive and
http://celery.readthedocs.org/en/latest/reference/celery.states.html for more details.
If state is SUCCESS, it is safe to check response['s3_urls'] for the archives submitted to s3
"""
task = create_and_upload_archive.AsyncResult(task_id)
task_info = task.info or {}
response = {
'state': task.state,
}
if task.state != 'FAILURE':
response['status'] = task_info.get('status', 'no status available at this point.')
response['src_url'] = task_info.get('src_url', '')
response['s3_urls'] = task_info.get('s3_urls', {})
else:
# something went wrong
response['status'] = str(task.info) # this is the exception raised
response['src_url'] = ''
response['s3_urls'] = {}

return MozharnessArchiveTask(**response)


@bp.route('/hgmo/<path:repo>/<rev>')
@api.apimethod(None, unicode, unicode, unicode, unicode, unicode, status_code=302)
def get_hgmo_archive(repo, rev, subdir=None, suffix='tar.gz', preferred_region=None):
"""
An archiver for hg.mozilla.org related requests. Uses relengapi.blueprints.archiver.get_archive
:param repo: the repo location off of hg.mozilla.org/
:param rev: the rev associated with the repo
:param subdir: optional subdir path to only archive a portion of the repo
:param suffix: the archive extension type. defaulted to tar.gz
:param preferred_region: the preferred s3 region to use
"""
src_url = current_app.config['ARCHIVER_HGMO_URL_TEMPLATE'].format(
repo=repo, rev=rev, suffix=suffix, subdir=subdir or ''
)
# though slightly odd to append the archive suffix extension with a subdir, this:
# 1) allows us to have archives based on different subdir locations from the same repo and rev
# 2) is aligned with the hg.mozilla.org format
key = '{repo}-{rev}.{suffix}'.format(repo=repo, rev=rev, suffix=suffix)
if subdir:
key += '/{}'.format(subdir)
return get_archive(src_url, key, preferred_region)


def get_archive(src_url, key, preferred_region):
"""
A generic getter for retrieving an s3 location of an archive where the archive is based off a
src_url.
sub-dir: hg.mozilla.org supports archives of sub directories within a repository. This
flexibility allows for creating archives of only a portion of what would normally be an entire
repo archive.
logic flow:
If their is already a key within s3, a re-direct link is given for the
s3 location. If the key does not exist, download the archive from src url, upload it to s3
for each region supported and return all uploaded s3 url locations.
When the key does not exist, the remaining work will be assigned to a celery background task
with a url location returned immediately for obtaining task state updates.
"""
buckets = current_app.config['ARCHIVER_S3_BUCKETS']
random_region = buckets.keys()[randint(0, len(buckets.keys()) - 1)]
# use preferred region if available otherwise choose a valid one at random
region = preferred_region if preferred_region and preferred_region in buckets else random_region
bucket = buckets[region]
s3 = current_app.aws.connect_to('s3', region)

# first, see if the key exists
if not s3.get_bucket(bucket).get_key(key):
task_id = key.replace('/', '_') # keep things simple and avoid slashes in task url
if create_and_upload_archive.AsyncResult(task_id).state != 'STARTED':
# task is currently not in progress so start one.
create_and_upload_archive.apply_async(args=[src_url, key], task_id=task_id)
return {}, 202, {'Location': url_for('archiver.task_status', task_id=task_id)}

log.info("generating GET URL to {}, expires in {}s".format(key, GET_EXPIRES_IN))
# return 302 pointing to s3 url with archive
signed_url = s3.generate_url(
method='GET', expires_in=GET_EXPIRES_IN,
bucket=bucket, key=key
)
return redirect(signed_url)
81 changes: 81 additions & 0 deletions relengapi/blueprints/archiver/tasks.py
Original file line number Diff line number Diff line change
@@ -0,0 +1,81 @@
# This Source Code Form is subject to the terms of the Mozilla Public
# License, v. 2.0. If a copy of the MPL was not distributed with this
# file, You can obtain one at http://mozilla.org/MPL/2.0/.
import logging
import requests
import shutil
import tempfile

from boto.s3.key import Key
from celery.task import current
from flask import current_app
from random import randint

from relengapi.lib import celery

log = logging.getLogger(__name__)

GET_EXPIRES_IN = 300


def upload_url_archive_to_s3(key, url, buckets):
s3_urls = {}

# make the source request
resp = requests.get(url, stream=True)

# create a temporary file for it
tempf = tempfile.TemporaryFile()
# copy the data, block-by-block, into that file
resp.raw.decode_content = True
shutil.copyfileobj(resp.raw, tempf)

# write it out to S3
for region in buckets:
s3 = current_app.aws.connect_to('s3', region)
k = Key(s3.get_bucket(buckets[region]))
k.key = key
k.set_metadata('Content-Type', resp.headers['Content-Type'])
# give it the same attachment filename
k.set_metadata('Content-Disposition', resp.headers['Content-Disposition'])
k.set_contents_from_file(tempf, rewind=True) # rewind points tempf back to start for us
s3_urls[region] = s3.generate_url(expires_in=GET_EXPIRES_IN, method='GET',
bucket=buckets[region], key=key)

resp.close()

return s3_urls


@celery.task(bind=True, track_started=True, max_retries=3)
def create_and_upload_archive(self, src_url, key):
"""
A celery task that downloads an archive if it exists from a src location and attempts to upload
the archive to a supported bucket in each supported region.
Throughout this process, update the state of the task and finally return the location of the
s3 urls if successful.
"""
status = "Task completed! Check 's3_urls' for upload locations."
s3_urls = {}
buckets = current_app.config['ARCHIVER_S3_BUCKETS']

resp = requests.head(src_url)
if resp.status_code == 200:
try:
s3_urls = upload_url_archive_to_s3(key, src_url, buckets)
except Exception as exc:
# set a jitter enabled delay
# where an aggressive delay would result in: 7s, 49s, and 343s
# and a gentle delay would result in: 4s, 16s, and 64s
delay = randint(4, 7) ** (current.request.retries + 1) # retries == 0 on first attempt
current.retry(exc=exc, countdown=delay)
else:
status = "Url not found. Does it exist? url: '{}', response: '{}' ".format(src_url,
resp.status_code)
log.warning(status)
return {
'status': status,
'src_url': src_url,
's3_urls': s3_urls,
}
95 changes: 95 additions & 0 deletions relengapi/blueprints/archiver/test_archiver.py
Original file line number Diff line number Diff line change
@@ -0,0 +1,95 @@
# This Source Code Form is subject to the terms of the Mozilla Public
# License, v. 2.0. If a copy of the MPL was not distributed with this
# file, You can obtain one at http://mozilla.org/MPL/2.0/.
import json
import mock
import moto

from nose.tools import eq_
from relengapi.blueprints.archiver.test_util import EXPECTED_TASK_STATUS_FAILED_RESPONSE
from relengapi.blueprints.archiver.test_util import EXPECTED_TASK_STATUS_SUCCESSFUL_RESPONSE
from relengapi.blueprints.archiver.test_util import create_s3_items
from relengapi.blueprints.archiver.test_util import fake_200_response
from relengapi.blueprints.archiver.test_util import fake_failed_task_status
from relengapi.blueprints.archiver.test_util import fake_successful_task_status
from relengapi.blueprints.archiver.test_util import setup_buckets

from relengapi.lib.testing.context import TestContext


cfg = {
'RELENGAPI_CELERY_LOG_LEVEL': 'DEBUG',

'AWS': {
'access_key_id': 'aa',
'secret_access_key': 'ss',
},

'ARCHIVER_S3_BUCKETS': {
'us-east-1': 'archiver-bucket-1',
'us-west-2': 'archiver-bucket-2'
},
'ARCHIVER_HGMO_URL_TEMPLATE': "https://hg.mozilla.org/{repo}/archive/{rev}.{suffix}/{subdir}",

'CELERY_BROKER_URL': 'memory://',
'CELERY_BACKEND': 'cache',
"CELERY_CACHE_BACKEND": 'memory',
'CELERY_ALWAYS_EAGER': True,
}

test_context = TestContext(config=cfg)


@moto.mock_s3
@test_context
def test_accepted_response_when_missing_s3_key(app, client):
setup_buckets(app, cfg)
with mock.patch("relengapi.blueprints.archiver.tasks.requests.get") as get, \
mock.patch("relengapi.blueprints.archiver.tasks.requests.head") as head:
# don't actually hit hg.m.o, we just care about starting a subprocess and
# returning a 202 accepted
get.return_value = fake_200_response()
head.return_value = fake_200_response()
resp = client.get('/archiver/hgmo/mozilla-central/9213957d166d?'
'subdir=testing/mozharness&preferred_region=us-west-2')
eq_(resp.status_code, 202, resp.status)


@moto.mock_s3
@test_context
def test_redirect_response_when_found_s3_key(app, client):
setup_buckets(app, cfg)
rev, repo, subdir, suffix = '203e1025a826', 'mozilla-central', 'testing/mozharness', 'tar.gz'
key = '{repo}-{rev}.{suffix}'.format(repo=repo, rev=rev, suffix=suffix)
if subdir:
key += '/{}'.format(subdir)
create_s3_items(app, cfg, key=key)

resp = client.get(
'/archiver/hgmo/{repo}/{rev}?subdir={subdir}&suffix={suffix}'.format(
rev=rev, repo=repo, subdir=subdir, suffix=suffix
)
)
eq_(resp.status_code, 302, resp.status)


@moto.mock_s3
@test_context
def test_task_status_when_failed(app, client):
expected_response = EXPECTED_TASK_STATUS_FAILED_RESPONSE
with mock.patch("relengapi.blueprints.archiver.create_and_upload_archive") as caua:
caua.AsyncResult.side_effect = fake_failed_task_status
response = client.get('/archiver/status/{task_id}'.format(task_id=123))
eq_(cmp(json.loads(response.data)['result'], expected_response), 0,
"a failed task status check does not equal expected status.")


@moto.mock_s3
@test_context
def test_task_status_when_success(app, client):
expected_response = EXPECTED_TASK_STATUS_SUCCESSFUL_RESPONSE
with mock.patch("relengapi.blueprints.archiver.create_and_upload_archive") as caua:
caua.AsyncResult.return_value = fake_successful_task_status(expected_response)
response = client.get('/archiver/status/{task_id}'.format(task_id=123))
eq_(cmp(json.loads(response.data)['result'], expected_response), 0,
"A successful task status check does not equal expected status.")
76 changes: 76 additions & 0 deletions relengapi/blueprints/archiver/test_tasks.py
Original file line number Diff line number Diff line change
@@ -0,0 +1,76 @@
# This Source Code Form is subject to the terms of the Mozilla Public
# License, v. 2.0. If a copy of the MPL was not distributed with this
# file, You can obtain one at http://mozilla.org/MPL/2.0/.
import mock
import moto

from relengapi.blueprints.archiver.tasks import create_and_upload_archive
from relengapi.blueprints.archiver.test_util import fake_200_response
from relengapi.blueprints.archiver.test_util import fake_404_response
from relengapi.blueprints.archiver.test_util import setup_buckets
from relengapi.lib.testing.context import TestContext


cfg = {
'RELENGAPI_CELERY_LOG_LEVEL': 'DEBUG',

'AWS': {
'access_key_id': 'aa',
'secret_access_key': 'ss',
},

'ARCHIVER_S3_BUCKETS': {
'us-east-1': 'archiver-bucket-1',
'us-west-2': 'archiver-bucket-2'
},
'ARCHIVER_HGMO_URL_TEMPLATE': "https://hg.mozilla.org/{repo}/archive/{rev}.{suffix}/{subdir}",

'CELERY_BROKER_URL': 'memory://',
'CELERY_BACKEND': 'cache',
"CELERY_CACHE_BACKEND": 'memory',
'CELERY_ALWAYS_EAGER': True,
}

test_context = TestContext(config=cfg)


@moto.mock_s3
@test_context
def test_invalid_hg_url(app):
setup_buckets(app, cfg)
rev, repo, suffix = 'fakeRev', 'mozilla-central', 'tar.gz'
key = '{repo}-{rev}.{suffix}'.format(repo=repo, rev=rev, suffix=suffix)
src_url = cfg['ARCHIVER_HGMO_URL_TEMPLATE'].format(repo=repo, rev=rev, suffix=suffix,
subdir='testing/mozharness')
with app.app_context():
with mock.patch("relengapi.blueprints.archiver.tasks.requests.head") as head:
head.return_value = fake_404_response()
task = create_and_upload_archive.apply_async(args=[src_url, key],
task_id=key.replace('/', '_'))
assert "Url not found." in task.info.get('status', {}), "invalid hg url was not caught!"


@moto.mock_s3
@test_context
def test_successful_upload_archive_response(app):
setup_buckets(app, cfg)
rev, repo, subdir, suffix = '203e1025a826', 'mozilla-central', 'testing/mozharness', 'tar.gz'
key = '{repo}-{rev}.{suffix}'.format(repo=repo, rev=rev, suffix=suffix)
if subdir:
key += '/{}'.format(subdir)
src_url = cfg['ARCHIVER_HGMO_URL_TEMPLATE'].format(repo=repo, rev=rev, suffix=suffix,
subdir='testing/mozharness')
with app.app_context():
with mock.patch("relengapi.blueprints.archiver.tasks.requests.get") as get, \
mock.patch("relengapi.blueprints.archiver.tasks.requests.head") as head:
get.return_value = fake_200_response()
head.return_value = fake_200_response()
task = create_and_upload_archive.apply_async(args=[src_url, key],
task_id=key.replace('/', '_'))
expected_regions = [region for region in cfg['ARCHIVER_S3_BUCKETS']]
all_regions_have_s3_urls = [
task.info.get("s3_urls", {}).get(region) for region in expected_regions
]
assert all(all_regions_have_s3_urls), "s3 urls not uploaded for each region!"
assert task.info.get('src_url') == src_url, "src url doesn't match upload response!"
assert task.state == "SUCCESS", "completed task's state isn't SUCCESS!"

0 comments on commit 2925b67

Please sign in to comment.