Yupeng Fu

Right, if I start the worker first, then the PS will also show OOM error. Will `CUDA_VISIBLE_DEVICES ` disable the GPU devices for PS? btw, if this is required, then...

@tfboyd thanks for the information. Since pushing to the website can take a while, do you mind posting the instructions here once you have it? I took a look at...

> This has the same behavior as setting `numInstancesPerPartition` to 1, and `numPartitions` to the partitions for the table. Currently we don't support explicitly setting partitions for `RealtimeSegmentAssignment`, so we...

> I see whaat you mean. Then that's a reasonable suggestion. But I hope we can find a way to enforce this as it's not so easy for users to...

We also ran into this issue in some scenarios: - A topic is recreated and the offset needs to be reset - User needs to change the Kafka cluster of...

yes, I believe it can address the issue. Thanks for the improvement!

Thanks a lot for the detailed report and investigation! Regarding 3, I think it's possible if the Trino execution takes long and the segment fetch takes some time apart. There's...

One idea on debugging is to print virtual columns of `$docId` and `$segmentName` of the duplicated records. This will give some insights on the internal state, and along with some...

@madhurranjan, could you share the patch that you achieved this? Thanks.

Awesome. Thanks for sharing the solution, @madhurranjan!