adriangb closed issue #16452: SortQueryFuzzer found a failing case on main
URL: https://github.com/apache/datafusion/issues/16452
--
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.
alamb commented on issue #16452:
URL: https://github.com/apache/datafusion/issues/16452#issuecomment-2994203073
We think we have fixed this in
- https://github.com/apache/datafusion/pull/16491
Let's keep this PR open until
1. We have some more time where the tests are passing s
adriangb commented on issue #16452:
URL: https://github.com/apache/datafusion/issues/16452#issuecomment-2994208242
> We revert the code changes in Temporarily fix bug in dynamic top-k
optimization #16465 that were unrelated. I'll make a PR to do that (update
Restore topk filtering tests #16
adriangb commented on issue #16452:
URL: https://github.com/apache/datafusion/issues/16452#issuecomment-2993586317
https://github.com/apache/datafusion/pull/16491 now has the fix!
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHu
alamb commented on issue #16452:
URL: https://github.com/apache/datafusion/issues/16452#issuecomment-2993537741
Sounds good -- thank you for the investigation @adriangb
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use
adriangb commented on issue #16452:
URL: https://github.com/apache/datafusion/issues/16452#issuecomment-2993562613
This is actually the same error that @blaginin reported in
https://github.com/apache/datafusion/issues/16452#issuecomment-2991948064:
https://github.com/apache/datafusion/actio
adriangb commented on issue #16452:
URL: https://github.com/apache/datafusion/issues/16452#issuecomment-2993553156
So yeah that's a trivially reproducible bug that is completely unrelated to
this mess but I guess just happens to be hit by the new code paths:
```rust
Literal::new(Sc
adriangb commented on issue #16452:
URL: https://github.com/apache/datafusion/issues/16452#issuecomment-2993339850
I can confirm it's related to #15770. It seems like there may be multiple
errors. This is one I found tonight:
```
thread 'tokio-runtime-worker' panicked at
/.cargo/r
adriangb commented on issue #16452:
URL: https://github.com/apache/datafusion/issues/16452#issuecomment-2993036342
I'm continuing to investigate. Unfortunately this is [very
flaky](https://github.com/apache/datafusion/pull/16485#issuecomment-2993026589)
and slow to test and I've run out of
alamb commented on issue #16452:
URL: https://github.com/apache/datafusion/issues/16452#issuecomment-2992153005
Right -- so in summary
- We tried to back out what we thought was the offending code:
https://github.com/apache/datafusion/pull/16465
- However, when we started running t
blaginin commented on issue #16452:
URL: https://github.com/apache/datafusion/issues/16452#issuecomment-2991948064
For transparency, new failure:
https://github.com/apache/datafusion/actions/runs/15780623790/job/44485106315
--
This is an automated message from the Apache Git Service.
To r
AdamGS commented on issue #16452:
URL: https://github.com/apache/datafusion/issues/16452#issuecomment-2989157877
done - https://github.com/apache/datafusion/pull/16465/files
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and
adriangb commented on issue #16452:
URL: https://github.com/apache/datafusion/issues/16452#issuecomment-2989145334
Let's merge that ASAP. I'm AFK for the next two hours? Could you prepare a
PR by chance?
--
This is an automated message from the Apache Git Service.
To respond to the messag
AdamGS commented on issue #16452:
URL: https://github.com/apache/datafusion/issues/16452#issuecomment-2989118706
that indeed makes the failure to go away
--
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
adriangb commented on issue #16452:
URL: https://github.com/apache/datafusion/issues/16452#issuecomment-2988934802
> datafusion.optimizer.enable_dynamic_filter_pushdown doesn't seem to make a
difference
@Dandandan I wonder if it's the filtering being done inside of the TopK?
@AdamG
adriangb commented on issue #16452:
URL: https://github.com/apache/datafusion/issues/16452#issuecomment-2988929483
I think this is a reproducer?
https://github.com/apache/datafusion/actions/runs/15764484967/job/44438090077?pr=16461
--
This is an automated message from the Apache Git Servi
AdamGS commented on issue #16452:
URL: https://github.com/apache/datafusion/issues/16452#issuecomment-2988819800
Some more findings:
1. `datafusion.optimizer.enable_dynamic_filter_pushdown` doesn't seem to
make a difference
2. Played around with the seeds, seems like the only one that'
adriangb commented on issue #16452:
URL: https://github.com/apache/datafusion/issues/16452#issuecomment-2988401513
Btw to disable the new topk changes you can set
`datafusion.optimizer.enable_dynamic_filter_pushdown = false`
--
This is an automated message from the Apache Git Service.
To
AdamGS commented on issue #16452:
URL: https://github.com/apache/datafusion/issues/16452#issuecomment-2988170885
I've built a repro that fails consistently, and the most interesting part of
it is that when using the default test runtime (which is a tokio current thread
variant), the test pa
alamb commented on issue #16452:
URL: https://github.com/apache/datafusion/issues/16452#issuecomment-2987732939
Perhaps this is related to the recent work from @adriangb with topk limit
pushdown (once we figure out how to reliably reproduce it, we can verify by
turning off the feature)
20 matches
Mail list logo
