automation: establish an automation framework

This commit is contained in:
Iain Learmonth 2022-05-08 17:20:04 +01:00
parent 1b53bf451c
commit 8abe5d60fa
31 changed files with 586 additions and 274 deletions

View file

@ -37,6 +37,8 @@ class ProxyAutomation(TerraformAutomation):
for group in groups:
subgroup = 0
for origin in group.origins:
if origin.destroyed is not None:
continue
while True:
if subgroups[group.id][subgroup] >= self.subgroup_max:
subgroup += 1
@ -87,7 +89,7 @@ class ProxyAutomation(TerraformAutomation):
self.deprecate_orphaned_proxies()
self.destroy_expired_proxies()
def tf_posthook(self):
def tf_posthook(self, *, prehook_result):
self.import_state(self.tf_show())
def tf_generate(self):

View file

@ -1,16 +1,11 @@
from azure.identity import ClientSecretCredential
from azure.mgmt.alertsmanagement import AlertsManagementClient
from app import app
from app.alarms import get_proxy_alarm
from app.extensions import db
from app.models.mirrors import Proxy
from app.models.alarms import AlarmState
from app.terraform.proxy import ProxyAutomation
class ProxyAzureCdnAutomation(ProxyAutomation):
short_name = "proxy_azure_cdn"
description = "Deploy proxies to Azure CDN"
provider = "azure_cdn"
subgroup_max = 25
parallelism = 1
@ -170,33 +165,3 @@ class ProxyAzureCdnAutomation(ProxyAutomation):
for proxy in proxies:
proxy.url = f"https://{proxy.slug}.azureedge.net"
db.session.commit()
def import_monitor_alerts():
credential = ClientSecretCredential(
tenant_id=app.config['AZURE_TENANT_ID'],
client_id=app.config['AZURE_CLIENT_ID'],
client_secret=app.config['AZURE_CLIENT_SECRET'])
client = AlertsManagementClient(
credential,
app.config['AZURE_SUBSCRIPTION_ID']
)
firing = [x.name[len("bandwidth-out-high-bc-"):]
for x in client.alerts.get_all()
if x.name.startswith("bandwidth-out-high-bc-") and x.properties.essentials.monitor_condition == "Fired"]
for proxy in Proxy.query.filter(
Proxy.provider == "azure_cdn",
Proxy.destroyed == None
):
alarm = get_proxy_alarm(proxy.id, "bandwidth-out-high")
if proxy.origin.group.group_name.lower() not in firing:
alarm.update_state(AlarmState.OK, "Azure monitor alert not firing")
else:
alarm.update_state(AlarmState.CRITICAL, "Azure monitor alert firing")
if __name__ == "__main__":
with app.app_context():
auto = ProxyAzureCdnAutomation()
auto.automate()
import_monitor_alerts()

View file

@ -1,17 +1,13 @@
import datetime
import boto3
from app import app
from app.alarms import get_proxy_alarm
from app.extensions import db
from app.models.alarms import Alarm, AlarmState
from app.models.mirrors import Proxy
from app.terraform.proxy import ProxyAutomation
class ProxyCloudfrontAutomation(ProxyAutomation):
short_name = "proxy_cloudfront"
description = "Deploy proxies to AWS CloudFront"
provider = "cloudfront"
template_parameters = [
@ -87,55 +83,3 @@ class ProxyCloudfrontAutomation(ProxyAutomation):
proxy.terraform_updated = datetime.datetime.utcnow()
break
db.session.commit()
def import_cloudwatch_alarms():
cloudwatch = boto3.client('cloudwatch',
aws_access_key_id=app.config['AWS_ACCESS_KEY'],
aws_secret_access_key=app.config['AWS_SECRET_KEY'],
region_name='us-east-2')
dist_paginator = cloudwatch.get_paginator('describe_alarms')
page_iterator = dist_paginator.paginate(AlarmNamePrefix="bandwidth-out-high-")
for page in page_iterator:
for cw_alarm in page['MetricAlarms']:
dist_id = cw_alarm["AlarmName"][len("bandwidth-out-high-"):]
proxy = Proxy.query.filter(Proxy.slug == dist_id).first()
if proxy is None:
print("Skipping unknown proxy " + dist_id)
continue
alarm = get_proxy_alarm(proxy.id, "bandwidth-out-high")
if cw_alarm['StateValue'] == "OK":
alarm.update_state(AlarmState.OK, "CloudWatch alarm OK")
elif cw_alarm['StateValue'] == "ALARM":
alarm.update_state(AlarmState.CRITICAL, "CloudWatch alarm ALARM")
else:
alarm.update_state(AlarmState.UNKNOWN, f"CloudWatch alarm {cw_alarm['StateValue']}")
alarm = Alarm.query.filter(
Alarm.alarm_type == "cloudfront-quota"
).first()
if alarm is None:
alarm = Alarm()
alarm.target = "service/cloudfront"
alarm.alarm_type = "cloudfront-quota"
alarm.state_changed = datetime.datetime.utcnow()
db.session.add(alarm)
alarm.last_updated = datetime.datetime.utcnow()
deployed_count = len(Proxy.query.filter(
Proxy.destroyed == None).all())
old_state = alarm.alarm_state
if deployed_count > 370:
alarm.alarm_state = AlarmState.CRITICAL
elif deployed_count > 320:
alarm.alarm_state = AlarmState.WARNING
else:
alarm.alarm_state = AlarmState.OK
if alarm.alarm_state != old_state:
alarm.state_changed = datetime.datetime.utcnow()
db.session.commit()
if __name__ == "__main__":
with app.app_context():
auto = ProxyCloudfrontAutomation()
auto.automate()
import_cloudwatch_alarms()