[
https://issues.apache.org/jira/browse/IMPALA-13009?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17844099#comment-17844099
]
ASF subversion and git services commented on IMPALA-13009:
----------------------------------------------------------
Commit ee21427d26620b40d38c706b4944d2831f84f6f5 in impala's branch
refs/heads/master from stiga-huang
[ https://gitbox.apache.org/repos/asf?p=impala.git;h=ee21427d2 ]
IMPALA-13009: Fix catalogd not sending deletion updates for some dropped
partitions
*Background*
Since IMPALA-3127, catalogd sends incremental partition updates based on
the last sent table snapshot ('maxSentPartitionId_' to be specific).
Dropped partitions since the last catalog update are tracked in
'droppedPartitions_' of HdfsTable. When catalogd collects the next
catalog update, they will be collected. HdfsTable then clears the set.
See details in CatalogServiceCatalog#addHdfsPartitionsToCatalogDelta().
If an HdfsTable is invalidated, it's replaced with an IncompleteTable
which doesn't track any partitions. The HdfsTable object is then added
to the deleteLog so catalogd can send deletion updates for all its
partitions. The same if the HdfsTable is dropped. However, the
previously dropped partitions are not collected in this case, which
results in a leak in the catalog topic if the partition name is not
reused anymore. Note that in the catalog topic, the key of a partition
update consists of the table name and the partition name. So if the
partition is added back to the table, the topic key will be reused then
resolves the leak.
The leak will be observed when a coordinator restarts. In the initial
catalog update sent from statestore, coordinator will find some
partition updates that are not referenced by the HdfsTable (assuming the
table is used again after the INVALIDATE). Then a Precondition check
fails and the table is not added to the coordinator.
*Overview of the patch*
This patch fixes the leak by also collecting the dropped partitions when
adding the HdfsTable to the deleteLog. A new field, dropped_partitions,
is added in THdfsTable to collect them. It's only used when catalogd
collects catalog updates.
Removes the Precondition check in coordinator and just reports the stale
partitions since IMPALA-12831 could also introduce them.
Also adds a log line in CatalogOpExecutor.alterTableDropPartition() to
show the dropped partition names for better diagnostics.
Tests
- Added e2e tests
Change-Id: I12a68158dca18ee48c9564ea16b7484c9f5b5d21
Reviewed-on: http://gerrit.cloudera.org:8080/21326
Reviewed-by: Impala Public Jenkins <impala-public-jenk...@cloudera.com>
Tested-by: Impala Public Jenkins <impala-public-jenk...@cloudera.com>
> Potential leak of partition deletions in the catalog topic
> ----------------------------------------------------------
>
> Key: IMPALA-13009
> URL: https://issues.apache.org/jira/browse/IMPALA-13009
> Project: IMPALA
> Issue Type: Bug
> Components: Catalog
> Affects Versions: Impala 4.0.0, Impala 4.1.0, Impala 4.2.0, Impala 4.1.1,
> Impala 4.1.2, Impala 4.3.0
> Reporter: Quanlong Huang
> Assignee: Quanlong Huang
> Priority: Critical
>
> Catalogd might not send partition deletions to the catalog topic in the
> following scenario:
> * Some partitions of a table are dropped.
> * The HdfsTable object is removed sequentially before catalogd collects the
> dropped partitions.
> In such case, catalogd loses track of the dropped partitions so their updates
> keep existing in the catalog topic, until the partition names are reused
> again.
> Note that the HdfsTable object can be removed by commands like DropTable or
> INVALIDATE.
> The leaked partitions will be detected when a coordinator restarts. An
> IllegalStateException complaining stale partitions will be reported, causing
> the table not being added to the catalog cache of coordinator.
> {noformat}
> E0417 16:41:22.317298 20746 ImpaladCatalog.java:264] Error adding catalog
> object: Received stale partition in a statestore update:
> THdfsPartition(partitionKeyExprs:[TExpr(nodes:[TExprNode(node_type:INT_LITERAL,
> type:TColumnType(types:[TTypeNode(type:SCALAR,
> scalar_type:TScalarType(type:INT))]), num_children:0, is_constant:true,
> int_literal:TIntLiteral(value:106), is_codegen_disabled:false)])],
> location:THdfsPartitionLocation(prefix_index:0, suffix:p=106), id:138,
> file_desc:[THdfsFileDesc(file_desc_data:18 00 00 00 00 00 00 00 00 00 0E 00
> 1C 00 18 00 10 00 00 00 08 00 04 00 0E 00 00 00 18 00 00 00 8B 0E 2D EB 8E 01
> 00 00 04 00 00 00 00 00 00 00 0C 00 00 00 01 00 00 00 4C 00 00 00 36 00 00 00
> 34 34 34 37 62 35 66 34 62 30 65 64 66 64 65 31 2D 32 33 33 61 64 62 38 35 30
> 30 30 30 30 30 30 30 5F 36 36 34 31 30 39 33 37 33 5F 64 61 74 61 2E 30 2E 74
> 78 74 00 00 0C 00 14 00 00 00 0C 00...)], access_level:READ_WRITE,
> stats:TTableStats(num_rows:-1), is_marked_cached:false,
> hms_parameters:{transient_lastDdlTime=1713342582, totalSize=4,
> numFilesErasureCoded=0, numFiles=1}, num_blocks:1, total_file_size_bytes:4,
> has_incremental_stats:false, write_id:0, db_name:default, tbl_name:my_part,
> partition_name:p=106,
> hdfs_storage_descriptor:THdfsStorageDescriptor(lineDelim:10, fieldDelim:1,
> collectionDelim:1, mapKeyDelim:1, escapeChar:0, quoteChar:1, fileFormat:TEXT,
> blockSize:0))
> Java exception follows:
> java.lang.IllegalStateException: Received stale partition in a statestore
> update:
> THdfsPartition(partitionKeyExprs:[TExpr(nodes:[TExprNode(node_type:INT_LITERAL,
> type:TColumnType(types:[TTypeNode(type:SCALAR,
> scalar_type:TScalarType(type:INT))]), num_children:0, is_constant:true,
> int_literal:TIntLiteral(value:106), is_codegen_disabled:false)])],
> location:THdfsPartitionLocation(prefix_index:0, suffix:p=106), id:138,
> file_desc:[THdfsFileDesc(file_desc_data:18 00 00 00 00 00 00 00 00 00 0E 00
> 1C 00 18 00 10 00 00 00 08 00 04 00 0E 00 00 00 18 00 00 00 8B 0E 2D EB 8E 01
> 00 00 04 00 00 00 00 00 00 00 0C 00 00 00 01 00 00 00 4C 00 00 00 36 00 00 00
> 34 34 34 37 62 35 66 34 62 30 65 64 66 64 65 31 2D 32 33 33 61 64 62 38 35 30
> 30 30 30 30 30 30 30 5F 36 36 34 31 30 39 33 37 33 5F 64 61 74 61 2E 30 2E 74
> 78 74 00 00 0C 00 14 00 00 00 0C 00...)], access_level:READ_WRITE,
> stats:TTableStats(num_rows:-1), is_marked_cached:false,
> hms_parameters:{transient_lastDdlTime=1713342582, totalSize=4,
> numFilesErasureCoded=0, numFiles=1}, num_blocks:1, total_file_size_bytes:4,
> has_incremental_stats:false, write_id:0, db_name:default, tbl_name:my_part,
> partition_name:p=106,
> hdfs_storage_descriptor:THdfsStorageDescriptor(lineDelim:10, fieldDelim:1,
> collectionDelim:1, mapKeyDelim:1, escapeChar:0, quoteChar:1, fileFormat:TEXT,
> blockSize:0))
> at
> com.google.common.base.Preconditions.checkState(Preconditions.java:512)
> at
> org.apache.impala.catalog.ImpaladCatalog.addTable(ImpaladCatalog.java:523)
> at
> org.apache.impala.catalog.ImpaladCatalog.addCatalogObject(ImpaladCatalog.java:334)
> at
> org.apache.impala.catalog.ImpaladCatalog.updateCatalog(ImpaladCatalog.java:262)
> at
> org.apache.impala.service.FeCatalogManager$CatalogdImpl.updateCatalogCache(FeCatalogManager.java:120)
> at
> org.apache.impala.service.Frontend.updateCatalogCache(Frontend.java:565)
> at
> org.apache.impala.service.JniFrontend.updateCatalogCache(JniFrontend.java:196)
> {noformat}
> *Workaround*
> Add back the stale partitions to the table (if the table is also dropped, add
> back the table first) using the same partition names. Then drop them again.
> This gives catalogd chances to send deletions for them.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-all-unsubscr...@impala.apache.org
For additional commands, e-mail: issues-all-h...@impala.apache.org
