Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

feat: refractored SQL-based alerting framework #10605

Merged
merged 17 commits into from
Sep 1, 2020

Conversation

JasonD28
Copy link
Contributor

@JasonD28 JasonD28 commented Aug 14, 2020

SUMMARY

Currently superset has all of its functionality centralized inside the Alerts table. This PR separates the different functionalities of alerting into separate components in order to increase the functionality of alerting and to allow this feature more flexibility for future upgrades.

Major Changes:

  • Created new table for SQLObservers which will handle the functionality of making SQL queries for the alert
  • Created new table for SQLObservations which will hold the information gathered from each SQLObserver query
  • Created new table for Validators which will handle looking at the values from SQLObservations to determine if an alert should be triggered

BEFORE/AFTER SCREENSHOTS OR ANIMATED GIF

Before:
Screen Shot 2020-08-27 at 12 13 29 AM
After:
Screen Shot 2020-08-27 at 12 16 15 AM
Screen Shot 2020-08-31 at 3 10 20 PM
Screen Shot 2020-08-31 at 3 11 07 PM
Screen Shot 2020-08-30 at 11 44 04 PM

TEST PLAN

  • unit test
  • local test
  • dropbox staging
  • dropbox prod

ADDITIONAL INFORMATION

  • Has associated issue:
  • Changes UI
  • Requires DB Migration.
  • Confirm DB Migration upgrade and downgrade tested.
  • Introduces new feature or API
  • Removes existing feature or API

@JasonD28 JasonD28 marked this pull request as draft August 14, 2020 20:50
superset/models/alerts.py Outdated Show resolved Hide resolved
superset/models/alerts.py Outdated Show resolved Hide resolved
superset/tasks/schedules.py Outdated Show resolved Hide resolved
superset/tasks/schedules.py Outdated Show resolved Hide resolved
@codecov-commenter
Copy link

codecov-commenter commented Aug 27, 2020

Codecov Report

Merging #10605 into master will increase coverage by 1.42%.
The diff coverage is 78.51%.

Impacted file tree graph

@@            Coverage Diff             @@
##           master   #10605      +/-   ##
==========================================
+ Coverage   58.92%   60.35%   +1.42%     
==========================================
  Files         756      363     -393     
  Lines       36072    23421   -12651     
  Branches     3301        0    -3301     
==========================================
- Hits        21256    14135    -7121     
+ Misses      14633     9286    -5347     
+ Partials      183        0     -183     
Flag Coverage Δ
#cypress ?
#python 60.35% <78.51%> (+0.03%) ⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files Coverage Δ
...ations/versions/01616b213391_refractor_alerting.py 0.00% <0.00%> (ø)
superset/app.py 80.00% <25.00%> (-0.96%) ⬇️
superset/views/alerts.py 73.68% <82.60%> (+7.01%) ⬆️
superset/tasks/schedules.py 76.13% <88.88%> (-3.40%) ⬇️
superset/tasks/alerts/validator.py 97.61% <97.61%> (ø)
superset/models/alerts.py 97.67% <98.03%> (ø)
superset/tasks/alerts/observer.py 100.00% <100.00%> (ø)
superset/models/schedules.py 86.79% <0.00%> (-7.55%) ⬇️
superset/db_engines/hive.py 82.14% <0.00%> (-3.58%) ⬇️
... and 401 more

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 234b6bb...7fe216f. Read the comment docs.

@JasonD28 JasonD28 changed the title fix: refractored SQL-based alerting framework [WIP] fix: refractored SQL-based alerting framework Aug 27, 2020
@JasonD28 JasonD28 marked this pull request as ready for review August 27, 2020 07:34
@bkyryliuk bkyryliuk changed the title fix: refractored SQL-based alerting framework feat: refractored SQL-based alerting framework Aug 27, 2020
Copy link
Member

@bkyryliuk bkyryliuk left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

really good progress here, some small comments.
let's start testing the solution on staging.

backref=backref("sql_observers", cascade="all, delete-orphan"),
)

# TODO: Abstract observations from the sqlamodls e.g.
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

s/sqlamodls/sqlalchemy models
add some context, e.g. observation table will be constantly growing and would need to either have some retention policy or be moved to the more scalable db. Same as alert log

default=textwrap.dedent(
"""
{
"example_threshold": 50
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

let's keep the default empty

)


def not_null_executor(
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

this should live outside the models, you can create a separate module for them
e.g. tasks/schedules/alerts/validator.py

"""Returns True if a SQLObserver's recent observation is not NULL"""

observation = observer.get_observations(1)[0]
if observation.value:
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

this one could be tricky
if 0 returns false :)

@@ -538,16 +543,15 @@ def schedule_alert_query( # pylint: disable=unused-argument
logger.info("Ignoring deactivated alert")
return

if schedule.sql_observers:
sql = schedule.sql_observers[0].sql
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

could there be multiple observers?
if not, let's change sql_observers to sql_observer

@@ -41,95 +50,204 @@
def setup_database():
with app.app_context():
slice_id = db.session.query(Slice).all()[0].id
database_id = utils.get_example_database().id
database = utils.get_example_database()
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

s/database/example_database

tests/alerts_tests.py Show resolved Hide resolved
alert_type="email",
slice_id=slice_id,
database_id=database_id,
Alert(**common_data, id=1, label="alert_1"),
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

let's not specify ids here and those are autogenerated, use label to retrieve alerts you care about

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I specified id's in order to control with SQLObserver is linked to which alert (alert_id column in SQLObserver), I'm not sure if there is another way to do this

# Test error with alert lacking observer
alert5 = dbsession.query(Alert).filter_by(id=5).one()
check_alert(alert5.id, alert5.label)
assert alert5.logs[-1].state == AlertState.ERROR
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

let's add successful alert tests as well

dbsession = setup_database
null_val1 = Validator(
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

let's decouple db and validators, validators are just the simple functions.
You can pass observation list and config to them instead.

@JasonD28
Copy link
Contributor Author

JasonD28 commented Sep 1, 2020

@willbarrett when you have a chance could you take a look?

Copy link
Member

@bkyryliuk bkyryliuk left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LG%nit

tests/alerts_tests.py Outdated Show resolved Hide resolved
@bkyryliuk bkyryliuk merged commit b59f6b1 into apache:master Sep 1, 2020
@JasonD28 JasonD28 deleted the jdavis_alert_val branch September 1, 2020 21:38
@mistercrunch
Copy link
Member

@bkyryliuk the observer/validator pattern is really confusing, I heard there's been discussions on reverting this. Personally I feel like we should 100% revert this. What's the plan!?

@bkyryliuk
Copy link
Member

bkyryliuk commented Oct 5, 2020

@bkyryliuk the observer/validator pattern is really confusing, I heard there's been discussions on reverting this. Personally I feel like we should 100% revert this. What's the plan!?

@mistercrunch we have synced with @willbarrett on this topic. I will merge the observer & validator back into alert object within next 2 weeks. Will keep Observation object to support validation across multiple observations.

@willbarrett
Copy link
Member

Thanks for the response @bkyryliuk - if you're unable to get to it this week please let me know and we'll take on the task on our side. We're under some time pressure to evolve this feature. Thanks!

auxten pushed a commit to auxten/incubator-superset that referenced this pull request Nov 20, 2020
* added new tables for alerting refractor

* reformatted inheritance structure

* added workflow for updated framework

* added suggested changes

* cleaned up changes

* added obervations to alert table to enable view

* added comments

* added requested changes

* fix tests

* added styling changes

* mypy

* added requested changes

* updated operator logic

* requested changes, 1 validator, styling changes

* refactored tests

* fix test alert workflow

* fixed create_alert in test

Co-authored-by: Jason Davis <@dropbox.com>
@mistercrunch mistercrunch added 🏷️ bot A label used by `supersetbot` to keep track of which PR where auto-tagged with release labels 🚢 0.38.0 labels Mar 12, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
🏷️ bot A label used by `supersetbot` to keep track of which PR where auto-tagged with release labels size/XXL 🚢 0.38.0
Projects
None yet
Development

Successfully merging this pull request may close these issues.

5 participants