fix: make migration 17 fully idempotent for databases in intermediate states (#481)

* fix: make migration 17 idempotent and standardize column names

Migration 17 renamed columns from sdk_session_id to memory_session_id,
but the migration wasn't fully idempotent - it could fail on databases
in intermediate states. This caused errors like:
- "no such column: sdk_session_id" (when columns already renamed)
- "table observations has no column named memory_session_id" (when not renamed)

Changes:
- Rewrite renameSessionIdColumns() to check each table individually
- Use safeRenameColumn() helper that handles all edge cases gracefully
- Deprecate migration 19 (repair migration) since 17 is now idempotent
- Update maintenance scripts to use memory_session_id column name
- Update test files to use new column names

Fixes column mismatch bug in v8.2.6+

* Merge origin/main into column-mismatch

---------

Co-authored-by: Alex Newman <thedotmack@gmail.com>
This commit is contained in:
Lindsey Catlett
2025-12-29 23:30:04 -05:00
committed by GitHub
parent 61a23a14a9
commit d9e966d8f4
11 changed files with 198 additions and 240 deletions
+3 -3
View File
@@ -18,7 +18,7 @@ interface CorruptedObservation {
obs_created: number;
session_started: number;
session_completed: number | null;
sdk_session_id: string;
memory_session_id: string;
}
function formatTimestamp(epoch: number): string {
@@ -54,9 +54,9 @@ function main() {
o.created_at_epoch as obs_created,
s.started_at_epoch as session_started,
s.completed_at_epoch as session_completed,
s.sdk_session_id
s.memory_session_id
FROM observations o
JOIN sdk_sessions s ON o.sdk_session_id = s.sdk_session_id
JOIN sdk_sessions s ON o.memory_session_id = s.memory_session_id
WHERE o.created_at_epoch < s.started_at_epoch -- Observation older than session
OR (s.completed_at_epoch IS NOT NULL
AND o.created_at_epoch > (s.completed_at_epoch + 3600000)) -- More than 1hr after session