Nightly
PublicTests that are too slow or non-deterministic for the regular Test pipeline
Scheduled build
Failed in 6h 47m
Miri test (full)SQLsmith explain succeeded with known error logs, main history: 
- Known issue sqlsmith: High environmentd memory usage during simple queries (even explain) (#7046) in journalctl-merge.log:
Dec 02 00:41:10 hetzner-aarch64-4cpu-8gb-04d15d37 kernel: Memory cgroup out of memory: Killed process 14336 (environmentd) total-vm:12006448kB, anon-rss:3636704kB, file-rss:93536kB, shmem-rss:0kB, UID:999 pgtables:17372kB oom_score_adj:0
Test details & reproducer
Use SQLsmith to generate random queries (AST/code based) and run them against Materialize: https://github.com/MaterializeInc/sqlsmith The queries can be complex, but we can't verify correctness or performance.bin/mzcompose --find sqlsmith run default --max-joins=5 --explain-only --runtime=1500
Output consistency (feature-flags for DFR) failed, main history: 
- Unknown error in workflow-default:
Docker compose failed: docker compose -f/dev/fd/3 --project-directory /var/lib/buildkite-agent/builds/hetzner-aarch64-4cpu-8gb-72934f88/materialize/nightly/test/feature-flag-consistency up --detach --wait --quiet-pull mz_this mz_other
- Known issue thread 'tokio:work-108' panicked at src/transform/src/equivalence_propagation.rs:76:13: ColumnKnowledge performed work after EquivalencePropagation (#8797) in services.log:
feature-flag-consistency-mz_this-1 | thread 'tokio:work-1' panicked at src/transform/src/equivalence_propagation.rs:76:13: ColumnKnowledge performed work after EquivalencePropagation
Test details & reproducer
Verify that Materialize has the same results with different sets of feature flags.bin/mzcompose --find feature-flag-consistency run default --seed=019384c6-a899-4450-a64e-b7b3cac301a7 --max-runtime-in-sec=1200 --evaluation-strategy=dataflow_rendering
Total Job Run Time: 4d 2h