Commit faaa7a5
fix e2e: handle NULL flush_lsn + correct ed25519 cert size threshold
Two bugs in the new tests caught by the previous CI run:
1. slot_blocks_wal_pruning_when_consumer_stops: the SIGSTOP'd sink
transiently disappears from pg_stat_replication, so flush_lsn could
come back NULL. We were deserialising as bool and panicking with
"error retrieving column 0". Switch to Option<bool> + treat NULL as
not-yet-caught-up.
2. tls_integration_postgres_with_beyond_pg_cert: the sanity threshold
was 1024 bytes, but ed25519 PEM certs are ~470 bytes (small key,
small cert). Drop the threshold to 300 bytes — still well above any
empty/stub but below real ed25519 cert content.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>1 parent e2dfe3f commit faaa7a5
1 file changed
Lines changed: 17 additions & 10 deletions
| Original file line number | Diff line number | Diff line change | |
|---|---|---|---|
| |||
6504 | 6504 | | |
6505 | 6505 | | |
6506 | 6506 | | |
6507 | | - | |
| 6507 | + | |
| 6508 | + | |
6508 | 6509 | | |
6509 | 6510 | | |
6510 | | - | |
| 6511 | + | |
6511 | 6512 | | |
6512 | 6513 | | |
6513 | 6514 | | |
| |||
7343 | 7344 | | |
7344 | 7345 | | |
7345 | 7346 | | |
7346 | | - | |
| 7347 | + | |
| 7348 | + | |
| 7349 | + | |
7347 | 7350 | | |
7348 | 7351 | | |
7349 | 7352 | | |
7350 | 7353 | | |
7351 | 7354 | | |
7352 | 7355 | | |
7353 | 7356 | | |
7354 | | - | |
7355 | | - | |
7356 | | - | |
| 7357 | + | |
| 7358 | + | |
| 7359 | + | |
| 7360 | + | |
| 7361 | + | |
7357 | 7362 | | |
7358 | 7363 | | |
7359 | 7364 | | |
| |||
7442 | 7447 | | |
7443 | 7448 | | |
7444 | 7449 | | |
7445 | | - | |
| 7450 | + | |
7446 | 7451 | | |
7447 | 7452 | | |
7448 | 7453 | | |
7449 | 7454 | | |
7450 | 7455 | | |
7451 | 7456 | | |
7452 | 7457 | | |
7453 | | - | |
7454 | | - | |
7455 | | - | |
| 7458 | + | |
| 7459 | + | |
| 7460 | + | |
| 7461 | + | |
| 7462 | + | |
7456 | 7463 | | |
7457 | 7464 | | |
7458 | 7465 | | |
| |||
0 commit comments