forked frompostgres/postgres
- Notifications
You must be signed in to change notification settings - Fork6
Commit0b56be8
committed
Found the problem with my operator-family changes: by fetching from
pg_opclass during LookupOpclassInfo(), I'd turned pg_opclass_oid_indexinto a critical system index. However the problem could only manifestduring a backend's first attempt to load opclass data, and then onlyif it had successfully loaded pg_internal.init and subsequently receiveda relcache flush; which made it impossible to reproduce in sequentialtests and darn hard even in parallel tests. Memo to self: whenexercising cache flush scenarios, must disable LookupOpclassInfo'sinternal cache too.1 parent5725b9d commit0b56be8
1 file changed
+5
-4
lines changedLines changed: 5 additions & 4 deletions
Original file line number | Diff line number | Diff line change | |
---|---|---|---|
| |||
8 | 8 |
| |
9 | 9 |
| |
10 | 10 |
| |
11 |
| - | |
| 11 | + | |
12 | 12 |
| |
13 | 13 |
| |
14 | 14 |
| |
| |||
2402 | 2402 |
| |
2403 | 2403 |
| |
2404 | 2404 |
| |
2405 |
| - | |
2406 |
| - | |
| 2405 | + | |
| 2406 | + | |
2407 | 2407 |
| |
2408 | 2408 |
| |
2409 | 2409 |
| |
| |||
2439 | 2439 |
| |
2440 | 2440 |
| |
2441 | 2441 |
| |
| 2442 | + | |
2442 | 2443 |
| |
2443 | 2444 |
| |
2444 | 2445 |
| |
2445 | 2446 |
| |
2446 | 2447 |
| |
2447 | 2448 |
| |
2448 |
| - | |
| 2449 | + | |
2449 | 2450 |
| |
2450 | 2451 |
| |
2451 | 2452 |
| |
|
0 commit comments
Comments
(0)