CI flake: replay_event_table_drop_after_snapshot `failed to acquire lock`

[clockwork-labs-bot]

Detected from a failed GitHub Actions job.

Job: Test Suite
Run/job: https://github.com/clockworklabs/SpacetimeDB/actions/runs/27430320192/job/81079467683
Head SHA: 8199dd9e9d80c9e2965b1b3ac6abca6646f795c3

Test:

• db::update::test::replay_event_table_drop_after_snapshot

Error:

Error: DatastoreError: Failed to acquire lock on "/tmp/stdb_testK0fGhe/snapshots/00000000000000000002.snapshot_dir": failed to create lockfile "/tmp/stdb_testK0fGhe/snapshots/00000000000000000002.lock": File exists (os error 17)

Useful nearby log context:

[INFO  spacetimedb_datastore::locking_tx_datastore::datastore] Capturing snapshot of database ... at TX offset 2
[INFO  spacetimedb_snapshot] [...] SNAPSHOT 00000000000000000002: Hardlinked 0 objects and wrote 9 objects
[INFO  spacetimedb_datastore::locking_tx_datastore::datastore] Capturing snapshot of database ... at TX offset 2

The failed run shows two snapshot captures at TX offset 2 immediately before the lock collision.

Command:

cargo test --all --exclude spacetimedb-smoketests --exclude spacetimedb-sdk --exclude spacetimedb -- --test-threads=2 --skip unreal

Rerun hint from the log:

cargo test -p spacetimedb-core --lib

[bfops]

looks like this is a non-flake and actual PR issue in #5289