Issues

• Prometheus memtier_connections and StatsD connections report configured clients x active threads, not open connections: they stay at full count while every connection is down (intended?)

  redis/memtier_benchmark#749 · Filter by author fcostaoliveira opened on Oct 16, 2026

• Ctrl+C after the --test-time deadline (e.g. during the 10 s drain grace) stamps the interrupt time as the end time: Total duration covers the stall and Ops/sec is deflated (intended?)

  redis/memtier_benchmark#747 · Filter by author fcostaoliveira opened on Oct 16, 2026

• Flaky tests seen in CI (prometheus latency cross-check, CLIENT KILL timing, injector compile timeout, ASK reconnect timeout)

  redis/memtier_benchmark#740 · Filter by author fcostaoliveira opened on Oct 16, 2026
  1

• --requests --retry-on-error with the default unlimited --max-retries never ends when every request keeps failing - intended?

  redis/memtier_benchmark#734 · Filter by author fcostaoliveira opened on Oct 16, 2026

• CI: run the RLTest suites over a unix socket (--unix)

  redis/memtier_benchmark#726 · Filter by author fcostaoliveira opened on Oct 16, 2026

• --run-count: after Ctrl+C, BEST/WORST RUN RESULTS can select the interrupted partial run over complete runs (intended?)

  redis/memtier_benchmark#711 · Filter by author fcostaoliveira opened on Oct 16, 2026
  1

• --reconnect-on-error with --requests and unlimited --max-reconnect-attempts: a connection whose setup keeps being rejected reconnects forever

  redis/memtier_benchmark#693 · Filter by author fcostaoliveira opened on Oct 16, 2026

• Cluster --read-preference secondary: an idle connection to a primary that dies is never noticed, so the slot map is never refreshed

  redis/memtier_benchmark#689 · Filter by author fcostaoliveira opened on Oct 16, 2026

• Cluster: requests for a slot with no owner in a valid CLUSTER SLOTS reply are never sent; the thread spins and the run ignores --test-time/--requests

  redis/memtier_benchmark#685 · Filter by author fcostaoliveira opened on Oct 15, 2026

• Cluster --read-preference secondary: read-only runs never refresh the slot map, so reads keep going to a replica promoted by CLUSTER FAILOVER (intended?)

  redis/memtier_benchmark#678 · Filter by author fcostaoliveira opened on Oct 15, 2026

• Cluster --reconnect-on-error: when the seed node (-s/-p) dies, the whole client stops even after its replica is promoted

  redis/memtier_benchmark#677 · Filter by author fcostaoliveira opened on Oct 15, 2026

• Cluster mode: keyless --command requests are sent on replica connections even with --read-preference primary (or no --read-preference)

  redis/memtier_benchmark#676 · Filter by author fcostaoliveira opened on Oct 15, 2026

• Standalone --transaction with --reconnect-interval: reconnect splits MULTI blocks (SET runs outside the transaction, EXEC without MULTI) and discarded QUEUED commands are counted

  redis/memtier_benchmark#673 · Filter by author fcostaoliveira opened on Oct 15, 2026
  1

• --run-count: AGGREGATED AVERAGE JSON "Latency"/"Average Latency" is an unweighted mean of per-run averages and disagrees with the text Avg. Latency

  redis/memtier_benchmark#671 · Filter by author fcostaoliveira opened on Oct 15, 2026

• Cluster --read-preference: a write sent to a replica gets -MOVED to the slot's own primary, and every such reply triggers a CLUSTER SLOTS refresh

  redis/memtier_benchmark#664 · Filter by author fcostaoliveira opened on Oct 15, 2026

• Cluster: a connection that drops while its CLUSTER SLOTS refresh is in flight keeps set_keep_value(true) after reconnecting (copies every reply value)

  redis/memtier_benchmark#653 · Filter by author fcostaoliveira opened on Oct 14, 2026

• Cluster mode without --retry-on-error: a -MOVED/-ASK reply is counted as a completed op (Count > commands the server ran) - intended?

  redis/memtier_benchmark#650 · Filter by author fcostaoliveira opened on Oct 14, 2026

• Cluster --transaction: 'pin connection received MOVED mid-rotation' warning is printed once per rotation during a slot migration (hundreds of lines)

  redis/memtier_benchmark#649 · Filter by author fcostaoliveira opened on Oct 14, 2026

• --transaction with --requests: run can end after a MULTI without its EXEC when --requests is not a multiple of the block length

  redis/memtier_benchmark#647 · Filter by author fcostaoliveira opened on Oct 14, 2026

• JSON output has no errors / retry_attempts / retried_ops counters for --retry-on-error runs

  redis/memtier_benchmark#646 · Filter by author fcostaoliveira opened on Oct 14, 2026

• test_json_output_integrity: latency tolerance lets a truncated average pass on low-count buckets and wrongly fails low-count sections

  redis/memtier_benchmark#645 · Filter by author fcostaoliveira opened on Oct 14, 2026
  2

• Vendored ax_code_coverage.m4 (serial 15) rejects lcov 2.x, fails code-coverage-capture on lcov 2.x, leaves .gcno on clean, and can't link with clang

  redis/memtier_benchmark#643 · Filter by author fcostaoliveira opened on Oct 14, 2026

• tests: harness crashes under RLTest --env existing-env and -s (dbDirPath / missing redis log, CONFIG GET reply sliced as a list)

  redis/memtier_benchmark#641 · Filter by author fcostaoliveira opened on Oct 14, 2026

• --monitor-input: a second --command "__monitor_line@__" takes over the stats slots of the first, which always report 0 ops

  redis/memtier_benchmark#639 · Filter by author fcostaoliveira opened on Oct 14, 2026

• Text output prints -nan Avg. Latency for a --command row with zero ops

  redis/memtier_benchmark#638 · Filter by author fcostaoliveira opened on Oct 14, 2026
  2