[Bug]: querynode crash

[BUPTAnderson]

Is there an existing issue for this?

• I have searched the existing issues

Environment

- Milvus version:master
- Deployment mode(standalone or cluster): cluster
- SDK version(e.g. pymilvus v2.0.0rc2): pymilvus v2.0.1
- OS(Ubuntu or CentOS): CentOS
- CPU/Memory: 32Cores/192GB
- GPU: 
- Others:

Current Behavior

one querynode crash

Expected Behavior

querynode running healthily

Steps To Reproduce

No response

Milvus Log

querynode.log

Anything else?

No response

[BUPTAnderson]

Assert "row_count_opt_.value() == size" at /opt/milvus/internal/core/src/segcore/SegmentSealedImpl.cpp:177 => field (0) data has different row count (72639) than other column's row count (64285)
It looks like there is a problem with the data validation for segment。

[yanliang567]

looks like a known issue that we are just trying to fix.
@congqixia could you please take a look?
/assign @congqixia
/unassign

[congqixia]

@BUPTAnderson yes, this log show the row count does not match among segment fields. Could you please provide how did you generates there segments? Say, schema of the collection and the batch number or clients number used to insert?

[BUPTAnderson]

@BUPTAnderson yes, this log show the row count does not match among segment fields. Could you please provide how did you generates there segments? Say, schema of the collection and the batch number or clients number used to insert?

the collection schema is in #17851

[BUPTAnderson]

crashed querynode log is bigger than 25MB, i split it to two parts.
querynode-19-part1.log

[BUPTAnderson]

part-2 log:
querynode-19-part2.log

[xiaofan-luan]

@BUPTAnderson Thanks ! I will handle it.
I probably also need the log from QueryCoord and IndexCoord/Node, could you help me on this