rustyrazorblade commented on issue #110:
URL:
https://github.com/apache/cassandra-easy-stress/issues/110#issuecomment-6348057009
## What the 100 cluster tests covered
This comment adds coverage tables to the results above. Every populate
option was tested with at least one value, but not every combination. All 100
tests passed in the final results.
### Option values tested
| Option | Values tested (number of tests) | Test IDs |
|---|---|---|
| `--populate` | `all` (38), `10000` (34), `5000` (10), workload option
ignores it (9), `200000` (2), `-1` (1), `10k` (1), `20000` (1), `25000` (1),
`1000` (1), `5` (1), `0` (1) | `-1`: 3; `10k`: 4; `0`: 24; above partitions,
wraps: 7 |
| `--populate.concurrency` | default 100 (80), `1` (6), `2` (1), `10` (1),
`100` (1), `500` (7), `1000` (1), `2000` (3) | sweep: 8–12 |
| `--populate.generator` | default `sequence` (93), `random` (4), `normal`
(2), explicit `sequence` (1) | 13–15, 34–35, 50, 55 |
| `--populate.batch` | default 1 (56), `1` (2), `2` (1), `4` (1), `5` (11),
`10` (19), `20` (3), `25` (1), `50` (2), `100` (2), `200` (1), `1000` (1) |
28–31, 39–40, 42–45 |
| `--populate.deletes` | default 0 (90), `0.1` (7), `0.2` (2), `1.0` (1) |
18, 20, 36, 49, 56, 64, 70, 74, 77, 80 |
| `--populate.deleteratio` (alias) | `0.5` (1) | 19 |
| `--threads` | 1 (62), 2 (10), 4 (25), 8 (3) | 8 threads: 6, 27, 41 |
| Main-phase options | `--rate 50` (1), `--deleterate 0.5` (1), `--maxwlat
20` (1), `--cl ONE` (1), `--cl ALL` (2), `-r 0` / `-r 0.5` (3) | 21–26, 81, 83 |
| Node failure | node down before populate (2), node stopped during populate
(3) | 82–86 |
### Features tested for each workload
| Workload | Tests | Batch > 1 | Deletes | Generators | Concurrency values |
`all` | Threads | Node failure |
|---|---:|---:|---:|---|---|---:|---|---|
| KeyValue | 34 | 2 | 3 | sequence, random, normal | 1, 2, 10, 100, 500,
1000, 2000 | 27 | 1, 2, 4, 8 | before, during |
| BasicTimeSeries | 19 | 18 | 1 | random, normal | 1, 500, 2000 | 1 | 1, 2,
4, 8 | during |
| CountersWide | 14 | 13 | 1 | random | 1, 2000 | 1 | 1, 2, 4 | during |
| Locking | 10 | 1 | 1 | random (overrides the workload) | 1 | – | 1, 4 | – |
| LWT | 4 | 1 | 1 | – | – | 4 | 1, 4 | – |
| Maps | 3 | 1 | – | – | 500 | 3 | 1, 4 | – |
| Sets (#105) | 3 | 1 | 1 | – | – | 3 | 1 | – |
| UdtTimeSeries | 3 | 2 | – | – | – | – | 1, 4 | – |
| RandomPartitionAccess | 3 | 1 | 1 | – | – | – | 1, 4 | – |
| AllowFiltering | 3 | 1 | 1 | – | – | – | 1, 4 | – |
| SAI | 3 | 1 | 1 | – | – | – | 1, 4 | – |
| CreateDrop (#112) | 1 | – | – | – | – | – | 1 | – |
### Not covered
- **Delete alias and an explicit 0.** `--populate.deleteratio` was tested
once, on `KeyValue`. An explicit `--populate.deletes 0` was not tested; 0 was
tested only as the default.
- **Deletes with batches.** Only 2 tests combine them, both with clustering
rows (`BasicTimeSeries`, `CountersWide`). No test combines a delete with a
batch on a workload that has one row per partition.
- **Generators.** `normal` ran only on `KeyValue` and `BasicTimeSeries`.
No generator override ran on `LWT`, `Maps`, `UdtTimeSeries`, or the other small
workloads.
- **Concurrency.** The full sweep ran only on `KeyValue`. High concurrency
with large batches ran only on `BasicTimeSeries` and `CountersWide`.
- **Human-readable suffixes.** Only `10k` was tested; `m` and `b` were not.
- **MCP path.** `-1` was tested on the command line only, not as JSON
through MCP.
- **Node failures.** They ran only on `KeyValue`, `BasicTimeSeries`, and
`CountersWide`, and only with one node at a time.
- **Workloads not tested:**
- `MaterializedViews` needs materialized views turned on.
- `TxnCounter` needs Accord.
- `DSESearch` needs DSE.
- `RangeScan` does not write.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
