fix: Prevent cached bootstrap data from leaking between users w/ same first/last name #26023

jfrag1 · 2023-11-20T00:17:34Z

SUMMARY

The caching mechanism for bootstrap data was not working as intended in the case where two users share the same first/last name. The cache key was dependent on the User object passed. When a non-primitive object is used to generate a cache key, flask-caching calls repr() on it to determine the key: https://flask-caching.readthedocs.io/en/latest/index.html#memoization.

The User class comes from flask appbuilder and has the following __repr__:

def get_full_name(self):
    return "{0} {1}".format(self.first_name, self.last_name)

def __repr__(self):
    return self.get_full_name()

Therefore, when two users shared a first and last name, they would generate the same cache key and the cache could leak between the two users. This was noticed because it would result in affected users seeing menu results they shouldn't see, though it potentially had other subtle effects as well, particularly for anyone adding a user-dependent COMMON_BOOTSTRAP_OVERRIDES_FUNC.

This PR fixes the issue by passing the actual user id to the memoized function so that it's included in the cache key.

BEFORE/AFTER SCREENSHOTS OR ANIMATED GIF

TESTING INSTRUCTIONS

ADDITIONAL INFORMATION

Has associated issue:
Required feature flags:
Changes UI
Includes DB Migration (follow approval process in SIP-59)
- Migration is atomic, supports rollback & is backwards-compatible
- Confirm DB migration upgrade and downgrade tested
- Runtime estimates and downtime expectations provided
Introduces new feature or API
Removes existing feature or API

codecov · 2023-11-20T00:26:06Z

Codecov Report

All modified and coverable lines are covered by tests ✅

Comparison is base (2499a1c) 69.05% compared to head (6ee1eb8) 69.07%.
Report is 21 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master   #26023      +/-   ##
==========================================
+ Coverage   69.05%   69.07%   +0.02%     
==========================================
  Files        1938     1937       -1     
  Lines       75835    75835              
  Branches     8427     8427              
==========================================
+ Hits        52366    52385      +19     
+ Misses      21299    21280      -19     
  Partials     2170     2170

Flag	Coverage Δ
hive	`53.68% <100.00%> (+<0.01%)`	⬆️
mysql	`78.16% <100.00%> (-0.03%)`	⬇️
postgres	`78.26% <100.00%> (-0.03%)`	⬇️
presto	`53.64% <100.00%> (+<0.01%)`	⬆️
python	`82.95% <100.00%> (+0.05%)`	⬆️
sqlite	`76.91% <100.00%> (-0.03%)`	⬇️
unit	`55.79% <100.00%> (+0.09%)`	⬆️

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

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

john-bodley · 2023-11-20T17:50:53Z

superset/views/base.py

@@ -380,7 +380,9 @@ def menu_data(user: User) -> dict[str, Any]:


 @cache_manager.cache.memoize(timeout=60)
-def cached_common_bootstrap_data(user: User, locale: str) -> dict[str, Any]:
+def cached_common_bootstrap_data(  # pylint: disable=unused-argument
+    user: User, user_id: int | None, locale: str


Why do we need both User and user_id? Surely the former contains the ID which should then be used for the cache key given it's globally unique.

While yes, User "contains" the ID, it's __repr__ method that flask-caching uses to generate the cache key does not (see PR description).

The 2 alternatives to this fix as-is would be to:

change the __repr__ upstream on FAB to include the id. (cc @dpgaspar if this is a welcome change)

Subclass FAB's user class and add our own repr then update to use that user class everywhere, which seems like a pretty large change for this small fix

@jfrag1 sorry I missed that detail when initially glancing over the PR description. Is a third alternative to just pass in user_id instead of both user and user_id?

john-bodley · 2023-11-21T00:47:26Z

superset/views/base.py

@@ -380,7 +380,9 @@ def menu_data(user: User) -> dict[str, Any]:


 @cache_manager.cache.memoize(timeout=60)
-def cached_common_bootstrap_data(user: User, locale: str) -> dict[str, Any]:
+def cached_common_bootstrap_data(  # pylint: disable=unused-argument
+    user: User, user_id: int | None, locale: str


@jfrag1 sorry I missed that detail when initially glancing over the PR description. Is a third alternative to just pass in user_id instead of both user and user_id?

john-bodley · 2023-11-21T00:50:48Z

superset/views/base.py

@@ -424,8 +426,9 @@ def cached_common_bootstrap_data(user: User, locale: str) -> dict[str, Any]:


 def common_bootstrap_payload(user: User) -> dict[str, Any]:
+    user_id = user.id if hasattr(user, "id") else None


It seems like the FAB User class will always have the id attribute defined and thus simply passing in user.id on line 431 should be safe.

Grokking through the code it seems like this is always g.user, thus rather than passing around the global, it seems clearer to remove the user variable from said method and thus said logic simply becomes,

from superset.utils.core import get_user_id def common_bootstrap_payload() -> dict[str, Any]: return { **cached_common_bootstrap_data(get_user_id(), get_locale()), "flash_messages": get_flashed_messages(with_categories=True), }

Oh great point it is always g.user, I'll make that change.

It seems like the FAB User class will always have the id attribute defined and thus simply passing in user.id on line 431 should be safe.

The type hint was a bit of a lie since g.user could also be AnonymousUserMixin. I'd update it but it's getting removed instead

@jfrag1 I think my comment is somewhat of a moot point given that it seems like we're going down the get_user_id() route.

john-bodley

LGTM

… first/last name (apache#26023) (cherry picked from commit 630734b)

sadpandajoe · 2023-11-28T17:36:33Z

🏷️ preset:2023.47

… first/last name (#26023)

… first/last name (apache#26023)

Pass user id to

ebc24ec

pull-request-size bot added the size/XS label Nov 20, 2023

Appease pylint

4c9ac95

john-bodley reviewed Nov 20, 2023

View reviewed changes

john-bodley reviewed Nov 21, 2023

View reviewed changes

Refactor

6ee1eb8

pull-request-size bot added size/S and removed size/XS labels Nov 21, 2023

jfrag1 requested a review from john-bodley November 21, 2023 22:38

john-bodley approved these changes Nov 21, 2023

View reviewed changes

jfrag1 closed this Nov 21, 2023

jfrag1 reopened this Nov 21, 2023

jfrag1 closed this Nov 21, 2023

jfrag1 reopened this Nov 21, 2023

jfrag1 closed this Nov 21, 2023

jfrag1 reopened this Nov 21, 2023

eschutho merged commit 630734b into apache:master Nov 21, 2023
119 of 123 checks passed

eschutho deleted the jack/fix-bootstrap-memoize-for-when-2-users-share-name branch November 21, 2023 23:39

michael-s-molina added the v3.0 Label added by the release manager to track PRs to be included in the 3.0 branch label Nov 22, 2023

sadpandajoe pushed a commit to preset-io/superset that referenced this pull request Nov 28, 2023

fix: Prevent cached bootstrap data from leaking between users w/ same…

7fe41ea

… first/last name (apache#26023) (cherry picked from commit 630734b)

michael-s-molina pushed a commit that referenced this pull request Dec 4, 2023

fix: Prevent cached bootstrap data from leaking between users w/ same…

ff5de25

… first/last name (#26023)

josedev-union pushed a commit to Ortege-xyz/studio that referenced this pull request Jan 22, 2024

fix: Prevent cached bootstrap data from leaking between users w/ same…

49da069

… first/last name (apache#26023)

cccs-rc pushed a commit to CybercentreCanada/superset that referenced this pull request Mar 6, 2024

fix: Prevent cached bootstrap data from leaking between users w/ same…

67f3c6b

… first/last name (apache#26023)

mistercrunch added 🍒 3.0.3 🍒 3.0.4 🏷️ bot A label used by `supersetbot` to keep track of which PR where auto-tagged with release labels 🚢 3.1.0 labels Mar 8, 2024

sfirke pushed a commit to sfirke/superset that referenced this pull request Mar 22, 2024

fix: Prevent cached bootstrap data from leaking between users w/ same…

93f6e36

… first/last name (apache#26023)

vinothkumar66 pushed a commit to vinothkumar66/superset that referenced this pull request Nov 11, 2024

fix: Prevent cached bootstrap data from leaking between users w/ same…

e8d5ec9

… first/last name (apache#26023)

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

fix: Prevent cached bootstrap data from leaking between users w/ same first/last name #26023

fix: Prevent cached bootstrap data from leaking between users w/ same first/last name #26023

jfrag1 commented Nov 20, 2023 •

edited

Loading

codecov bot commented Nov 20, 2023 •

edited

Loading

john-bodley Nov 20, 2023

jfrag1 Nov 20, 2023

john-bodley Nov 21, 2023

john-bodley Nov 21, 2023

john-bodley Nov 21, 2023

john-bodley Nov 21, 2023

jfrag1 Nov 21, 2023

john-bodley Nov 21, 2023

john-bodley left a comment

sadpandajoe commented Nov 28, 2023

		@@ -424,8 +426,9 @@ def cached_common_bootstrap_data(user: User, locale: str) -> dict[str, Any]:


		def common_bootstrap_payload(user: User) -> dict[str, Any]:
		user_id = user.id if hasattr(user, "id") else None

fix: Prevent cached bootstrap data from leaking between users w/ same first/last name #26023

fix: Prevent cached bootstrap data from leaking between users w/ same first/last name #26023

Conversation

jfrag1 commented Nov 20, 2023 • edited Loading

SUMMARY

BEFORE/AFTER SCREENSHOTS OR ANIMATED GIF

TESTING INSTRUCTIONS

ADDITIONAL INFORMATION

codecov bot commented Nov 20, 2023 • edited Loading

Codecov Report

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

john-bodley left a comment

Choose a reason for hiding this comment

sadpandajoe commented Nov 28, 2023

jfrag1 commented Nov 20, 2023 •

edited

Loading

codecov bot commented Nov 20, 2023 •

edited

Loading