StarRocks / starrocks

The world's fastest open query engine for sub-second analytics both on and off the data lakehouse. With the flexibility to support nearly any scenario, StarRocks provides best-in-class performance for multi-dimensional analytics, real-time analytics, and ad-hoc queries. A Linux Foundation project.
https://starrocks.io
Apache License 2.0
9.04k stars 1.82k forks source link

[42000][1064] get_applied_rowsets failed, tablet updates is in error state: actual row size changed after compaction #30320

Closed ddecaro11 closed 1 month ago

ddecaro11 commented 1 year ago

Hello,

I have created a table and I am loading it with some parquet files coming from azure delta lake storage. The load job completes successfully and the number of rows scanned is correct (around 300M). When I try to query the table I receive the following error:

[42000][1064] get_applied_rowsets failed, tablet updates is in error state: tablet:10104 actual row size changed after compaction 67133362 -> 58040099tablet:10104 #version:3 [1 2.1@2 2.1] #pending:0 backend:127.0.0.1

Steps to reproduce the behavior (Required)

  1. create table signals ( IngestionId STRING NOT NULL, DeviceId INT, VehicleId INT, MessageId INT, SignalId INT, BusId INT, ApplicationTypeId INT, SignalTypeId INT, RegisteredOn BIGINT, RegisteredTimestamp DATETIME, ConfigurationId INT, ConfigurationName STRING, ConfigurationVersion INT, ProductLineId INT, PlatformId INT, ProgramId INT, SegmentId INT, IngestedOn BIGINT, MessageName STRING, MessageAddress INT, PGN INT, SourceAddress INT, DestinationAddress INT, SignalName STRING, BusName STRING, SignalValue DOUBLE, SignalUnit STRING, SignalValueString STRING, LoggerId INT, LogInstanceId INT ) ENGINE=OLAP PRIMARY KEY(IngestionId, DeviceId, VehicleId, MessageId, SignalId, BusId, ApplicationTypeId, SignalTypeId, RegisteredOn ) DISTRIBUTED BY HASH(IngestionId) PROPERTIES("replication_num" = "1");

  2. LOAD LABEL sr.label_test ( DATA INFILE("abfss://xxx@xxxx.dfs.core.windows.net/Signal/DeviceId=46/VehicleId=/RegisteredPartition=/*") INTO TABLE signals FORMAT AS "parquet" (IngestionId, MessageId, SignalId, BusId, ApplicationTypeId, SignalTypeId, RegisteredTimestamp, ConfigurationId, ConfigurationName, ConfigurationVersion, ProductLineId, PlatformId, ProgramId, SegmentId, RegisteredOn, IngestedOn, MessageName, MessageAddress, PGN, SourceAddress, DestinationAddress, SignalName, BusName, SignalValue, SignalUnit, SignalValueString, LoggerId, LogInstanceId ) COLUMNS FROM PATH AS (DeviceId, VehicleId) ) WITH BROKER ( "azure.adls2.storage_account" = "xxxxx", "azure.adls2.shared_key" = "xxxxx" )

  3. select count(1) from signals;

Expected behavior (Required)

Return 313506982

Real behavior (Required)

Error: [42000][1064] get_applied_rowsets failed, tablet updates is in error state: tablet:10104 actual row size changed after compaction 67133362 -> 58040099tablet:10104 #version:3 [1 2.1@2 2.1] #pending:0 backend:127.0.0.1

StarRocks version (Required)

I am using the all in 1 docker image

Thank you!

ddecaro11 commented 1 year ago

The same happens with version 2.5.11 b75b46d55d

decster commented 1 year ago

please provide the BE logs?

sevev commented 1 year ago

can you provide the BE logs?

decster commented 1 year ago

usually, BE logs are in be/log/be.INFO.log.* , if they are to large, you can just grepp&send relevant logs

grep 10104 be.INFO.log.* > 10104.log
ddecaro11 commented 1 year ago

10102.log

This tablet number is different but is the same error.

Thank you very much

github-actions[bot] commented 8 months ago

We have marked this issue as stale because it has been inactive for 6 months. If this issue is still relevant, removing the stale label or adding a comment will keep it active. Otherwise, we'll close it in 10 days to keep the issue queue tidy. Thank you for your contribution to StarRocks!

ddecaro11 commented 8 months ago

Any updates on this?

github-actions[bot] commented 2 months ago

We have marked this issue as stale because it has been inactive for 6 months. If this issue is still relevant, removing the stale label or adding a comment will keep it active. Otherwise, we'll close it in 10 days to keep the issue queue tidy. Thank you for your contribution to StarRocks!