Commit 4cab6da
catalog: key mz_cluster_replica_frontiers_ind on (object_id, replica_id)
The `(object_id, replica_id)` join in `mz_hydration_statuses` was building
its own full-relation arrangement of `mz_cluster_replica_frontiers` because
the index was keyed on `object_id` only. That arrangement grows very large
under replica churn (observed ~47 GB on a production mz_catalog_server).
Widening the index key lets the join reuse the index; the remaining
`object_id`-only consumers all filter to `write_frontier IS NULL`, so their
arrangements stay small.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>1 parent 3b10a79 commit 4cab6da
3 files changed
Lines changed: 14 additions & 6 deletions
File tree
- src/catalog/src/builtin
- test
- sqllogictest
- testdrive
| Original file line number | Diff line number | Diff line change | |
|---|---|---|---|
| |||
2465 | 2465 | | |
2466 | 2466 | | |
2467 | 2467 | | |
2468 | | - | |
2469 | | - | |
| 2468 | + | |
| 2469 | + | |
| 2470 | + | |
| 2471 | + | |
| 2472 | + | |
| 2473 | + | |
| 2474 | + | |
| 2475 | + | |
| 2476 | + | |
2470 | 2477 | | |
2471 | 2478 | | |
2472 | 2479 | | |
2473 | | - | |
| 2480 | + | |
2474 | 2481 | | |
2475 | | - | |
| 2482 | + | |
| 2483 | + | |
2476 | 2484 | | |
2477 | 2485 | | |
2478 | 2486 | | |
| |||
| Original file line number | Diff line number | Diff line change | |
|---|---|---|---|
| |||
39 | 39 | | |
40 | 40 | | |
41 | 41 | | |
42 | | - | |
| 42 | + | |
43 | 43 | | |
44 | 44 | | |
45 | 45 | | |
| |||
| Original file line number | Diff line number | Diff line change | |
|---|---|---|---|
| |||
296 | 296 | | |
297 | 297 | | |
298 | 298 | | |
299 | | - | |
| 299 | + | |
300 | 300 | | |
301 | 301 | | |
302 | 302 | | |
| |||
0 commit comments