Found commits after time :20230220161017756, please rollback greater commits first

[koochiswathiTR]

• Have you gone through our FAQs?

• Join the mailing list to engage in conversations and get faster support at [email protected].

• If you have triaged this as a bug, then file an issue directly.

Hi,
Our streaming job is failing with Found commits after time :20230220161017756, please rollback greater commits first,

We tried to rollback commits with the command commit rollback --commit commit_num but we are facing the below exception
Caused by: java.lang.IllegalArgumentException: Cannot use marker based rollback strategy on completed instant:[20230221130513323__deltacommit__COMPLETED]
at org.apache.hudi.common.util.ValidationUtils.checkArgument(ValidationUtils.java:40)
at org.apache.hudi.table.action.rollback.BaseRollbackActionExecutor.(BaseRollbackActionExecutor.java:93)
at org.apache.hudi.table.action.rollback.BaseRollbackActionExecutor.(BaseRollbackActionExecutor.java:73)
at org.apache.hudi.table.action.rollback.MergeOnReadRollbackActionExecutor.(MergeOnReadRollbackActionExecutor.java:48)
at org.apache.hudi.table.HoodieSparkMergeOnReadTable.rollback(HoodieSparkMergeOnReadTable.java:170)
at org.apache.hudi.client.BaseHoodieWriteClient.rollback(BaseHoodieWriteClient.java:766)
... 15 more

Our Hudi configs are

DataSourceWriteOptions.TABLE_TYPE.key() -> DataSourceWriteOptions.MOR_TABLE_TYPE_OPT_VAL,
DataSourceWriteOptions.RECORDKEY_FIELD.key() -> "guid",
DataSourceWriteOptions.PARTITIONPATH_FIELD.key() -> "collectionName",
DataSourceWriteOptions.PRECOMBINE_FIELD.key() -> "operationTime",
HoodieCompactionConfig.INLINE_COMPACT_TRIGGER_STRATEGY.key() -> CompactionTriggerStrategy.TIME_ELAPSED.name,
HoodieCompactionConfig.INLINE_COMPACT_TIME_DELTA_SECONDS.key() -> String.valueOf(60 * 60),
HoodieCompactionConfig.CLEANER_POLICY.key() -> HoodieCleaningPolicy.KEEP_LATEST_COMMITS.name(),
HoodieCompactionConfig.CLEANER_COMMITS_RETAINED.key() -> "624", 
HoodieCompactionConfig.MIN_COMMITS_TO_KEEP.key() -> "625",  
HoodieCompactionConfig.MAX_COMMITS_TO_KEEP.key() -> "648", 
HoodieCompactionConfig.ASYNC_CLEAN.key() -> "false", 
HoodieCompactionConfig.INLINE_COMPACT.key() -> "true",
HoodieMetricsConfig.TURN_METRICS_ON.key() -> "true",
HoodieMetricsConfig.METRICS_REPORTER_TYPE_VALUE.key() -> MetricsReporterType.DATADOG.name(),
HoodieMetricsDatadogConfig.API_SITE_VALUE.key() -> "US",
HoodieMetricsDatadogConfig.METRIC_PREFIX_VALUE.key() -> "tacticalnovusingest.hudi",
HoodieMetricsDatadogConfig.API_KEY_SUPPLIER.key() -> "com.tr.indigo.tacticalnovusingest.utils.DatadogKeySupplier",
HoodieMetadataConfig.ENABLE.key() -> "false",
HoodieWriteConfig.ROLLBACK_USING_MARKERS_ENABLE.key() -> "false",

Please help what caused this issue
Help us how to resolve this

Environment Description AWS

• Hudi version : 0.11.1

• Spark version : 3.1.2

• Hive version : NA

• Hadoop version :3

• Storage (HDFS/S3/GCS..) :S3

• Running on Docker? (yes/no) :no

Additional context

Add any other context about the problem here.

Stacktrace

Add the stacktrace of the error.

[danny0405]

Did you enable the lazy cleaning for multi writers?

[koochiswathiTR]

@danny0405 We are not using multi writer, Its only single writer.
What is lazy cleaning can you brief on this?

[danny0405]

What version of Hudi did you use, seems an uknown bug.

[nsivabalan]

can you post the contents of ".hoodie" w/ last mod time intact (ls -ltr).
Also, when you triggered rollback via cli, whats the entire command you passed.

I see we have an option --rollbackUsingMarkers. did you set it or no ?

[nsivabalan]

we also made some fix on rolling back a completed instant #6313. can you try 0.12.1 may be.

[chestnutqiang]

Did you enable the lazy cleaning for multi writers?

If there are multiple parallelism writes for different partitions of the same table, does HUDI have any plans to optimize this? For example, a Spark SQL is split into multiple applications to write.

[danny0405]

Guess not, becase until committing, Hudi has no idear whether the two(or more) commits have conflicts, so the rollback plan should execute from the latest instant to the oldest, to ensure the validity of the data set.