You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Cassandra CI: test_abac fixtures get HTTP 500 from the management account create and DeleteTable, intermittently #610
The cassandra-pytest job fails intermittently in tests/test_abac.py with HTTP 500 from two calls the test fixtures make: the management API's account create (POST /accounts, asserted as 201 in the account_id fixture at tests/test_abac.py:77) and DeleteTable in the test teardown. The same job passes on rerun, and the failing PRs change no Cassandra code path these calls use. The job log carries no server-side line for either 500, so the cause is not recorded.
69 ERROR, account create answered 500 for every class from TestResourceTagCondition onward; rerun: 8 ERROR in TestUserIdRecreate and TestTokenIssueTimeRoleSession, same 500
3 FAILED, DeleteTable answered InternalServerError in TestOrganizationKeysAbsent
In every run the rest of the suite passed (331 passed in the job's comprehensive step each time), and the failures sit in whichever test_abac classes ran while the server was in the failing state.
What would settle it
The job log tails only the pytest output. Capturing the server's log on failure (the extenddb serve process the job starts) would record the error the 500 wraps. Two candidates from the shape: the management account create and DeleteTable both write catalog rows through lightweight transactions on the single-node Cassandra container, and a timed-out LWT surfaces as 500 on both paths today.
Not this issue
#600 covers data-plane requests answering 500 once a DeleteTable has dropped the base data table; here the DeleteTable call itself answers 500, and the account create has no table involved.
Activity