ahmed-mez commented on PR #19562:
URL: https://github.com/apache/datafusion/pull/19562#issuecomment-3716188596
> The reason the benchmarks are failing I think is that that they have
basically slowed down to a crawl.
>
> ```
> SELECT "WatchID", "ClientIP", COUNT(*) AS c, SUM("IsRefresh"),
AVG("ResolutionWidth") FROM hits GROUP BY "WatchID", "ClientIP" ORDER BY c DESC
LIMIT 10;
>
> Query 32 iteration 0 took 147686.4 ms and returned 10 rows
> ```
>
> I killed it after this
>
> Perhaps you can run the benchmarks locally
>
> ```shell
> benchmarks/bench.sh data clickbench_partitioned
> benchmarks/bench.sh run clickbench_partitioned
> ```
Oh that's interesting! That benchmark query runs much more faster locally
(same branch and commit)
```
SELECT "WatchID", "ClientIP", COUNT(*) AS c, SUM("IsRefresh"),
AVG("ResolutionWidth") FROM hits GROUP BY "WatchID", "ClientIP" ORDER BY c DESC
LIMIT 10;
Query 32 iteration 0 took 11717.4 ms and returned 10 rows
Query 32 iteration 1 took 11660.8 ms and returned 10 rows
Query 32 iteration 2 took 11855.8 ms and returned 10 rows
Query 32 iteration 3 took 12496.2 ms and returned 10 rows
Query 32 iteration 4 took 12555.9 ms and returned 10 rows
Query 32 avg time: 12057.21 ms
```
I'll keep digging and try to compare to base locally.
Apart from that, is the failed benchmark CI job accessible somehow for
further debugging?
--
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]
