Segfault when shutting down database which got encrypted by another process [CORE6501] #6731
Labels
affect-version: 3.0.0
affect-version: 3.0.1
affect-version: 3.0.2
affect-version: 3.0.3
affect-version: 3.0.4
affect-version: 3.0.5
affect-version: 3.0.6
affect-version: 3.0.7
affect-version: 4.0 Alpha 1
affect-version: 4.0 Beta 1
affect-version: 4.0 Beta 2
affect-version: 4.0 Initial
affect-version: 4.0 RC 1
component: engine
fix-version: 3.0.8
fix-version: 4.0.0
priority: major
qa: covered by another tests
type: bug
Submitted by: @AlexPeshkoff
When database is shutting down header page should be fetched and next written to database in order to update transaction numbers. If AST, causing change of encryption state, arrived during DB shutdown right before TRA_update_counters() database encryption manager is already missing environment required to load encryption plugin (like attachment) and segfaults happens when plugin is tried to be loaded.
Update of DB header does not require presence of database crypt plugin - it's anyway not encrypted page. Therefore fix is ti check environment before attempt to load plugin.
Commits: 64fe26f 18019a6
====== Test Details ======
Crash occured several times on Linux only, during execution of test 'functional.util.gbak_zip'.
Separate test not required.
The text was updated successfully, but these errors were encountered: