Rare segfault when preparing request working with a table with triggers [CORE5428] #5700
Labels
affect-version: 2.1.7
affect-version: 2.5.0
affect-version: 2.5.1
affect-version: 2.5.2 Update 1
affect-version: 2.5.2
affect-version: 2.5.3 Update 1
affect-version: 2.5.3
affect-version: 2.5.4
affect-version: 2.5.5
affect-version: 2.5.6
affect-version: 3.0.0
affect-version: 3.0.1
affect-version: 4.0 Initial
component: engine
fix-version: 2.5.7
fix-version: 3.0.2
fix-version: 4.0 Alpha 1
priority: major
qa: cannot be tested
type: bug
Submitted by: @AlexPeshkoff
This old and rarely happening bug happens when structure of a table with triggers is modified (DDL) when other attachment is preparing request working with that table. AST by itself does not cause any harm to data structures but rescan of table's metadata (which happens on any call to metadata cache for given table) drops old triggers description and creates new one. But in some cases pointers to old triggers are cached by request compiler and next use of them makes classic server process die.
Commits: 431b33b 0ca6c0d f88dded
====== Test Details ======
Significant (and artificial) slowdown in FB engine is required for reproducing this issue.
There are two tests for check FB stability while applying DDL under strong DML workload, but both of them did not produce any crash.
May be these tests will be added into fbt-repo but I will do it later (after further discuss with Alex).
The text was updated successfully, but these errors were encountered: