[
https://issues.apache.org/jira/browse/IGNITE-20821?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17785400#comment-17785400
]
Ivan Bessonov commented on IGNITE-20821:
----------------------------------------
The problem is that we couldn't handle the {{ClosedChannelException}} in
{{{}AbstractFilePageStoreIo{}}}, because underlying IO implementation didn't
throw proper exceptions. I fixed the bug in AsyncFileIo to solve the issue.
> Sporadic exception on KeyValueView#get
> --------------------------------------
>
> Key: IGNITE-20821
> URL: https://issues.apache.org/jira/browse/IGNITE-20821
> Project: Ignite
> Issue Type: Bug
> Components: cache, clients
> Reporter: Ivan Artiukhov
> Assignee: Ivan Bessonov
> Priority: Major
> Labels: ignite-3, ignite3_performance
> Fix For: 3.0.0-beta2
>
> Attachments: run269-logs.zip
>
> Time Spent: 40m
> Remaining Estimate: 0h
>
> AI3, rev. 8ebd9e842a2aaec5e1ae7e3b5e1cb36a040dc565
> Benchmark:
> [https://github.com/gridgain/YCSB/blob/ycsb-2023.7/ignite3/src/main/java/site/ycsb/db/ignite3/IgniteClient.java]
>
> Cluster setup: 1 server node, aipersist, 25 partitions, raft.fsync=false
> h1. Steps
> Run the mentioned key-value benchmark to preload data (10 million unique
> keys):
>
> {noformat}
> Command line: -db site.ycsb.db.ignite3.IgniteClient -load -P
> /opt/pubagent/poc/config/ycsb/workloads/workloadc -threads 1 -p
> recordcount=250000 -p warmupops=50000 -p dataintegrity=true -p
> measurementtype=timeseries -p status.interval=1 -p hosts=192.168.1.63 -p
> recordcount=10000000 -p operationcount=10000000 -s
> {noformat}
> Observe that all 10M keys have been loaded:
>
>
> {noformat}
> [INSERT], Return=OK, 10000000
> {noformat}
> Run the 100% get operations over preloaded data:
>
> {noformat}
> Command line: -db site.ycsb.db.ignite3.IgniteClient -t -P
> /opt/pubagent/poc/config/ycsb/workloads/workloadc -threads 1 -p
> operationcount=250000 -p recordcount=250000 -p warmupops=50000 -p
> dataintegrity=true -p measurementtype=timeseries -p status.interval=1 -p
> hosts=192.168.1.63 -p recordcount=10000000 -p operationcount=10000000 -s
> {noformat}
> h1. Expected result
> All 10M keys were read without errors
> h1. Actual result
> Reading 4341 keys out of 10M keys failed with exception like the following
> one:
> {noformat}
> Caused by: org.apache.ignite.internal.storage.StorageException: IGN-STORAGE-1
> TraceId:069639f9-b2b9-4690-b066-3ca0bf2e9e06 Row version lookup failed:
> [rowId=RowId [partitionId=2, uuid=c18be8d8-e29d-4be9-9238-e85a8a18a3b9],
> tableId=5, partitionId=2]
> at
> org.apache.ignite.internal.storage.pagememory.mv.AbstractPageMemoryMvPartitionStorage.findVersionChain(AbstractPageMemoryMvPartitionStorage.java:879)
> at
> org.apache.ignite.internal.storage.pagememory.mv.AbstractPageMemoryMvPartitionStorage.lambda$read$8(AbstractPageMemoryMvPartitionStorage.java:332)
> at
> org.apache.ignite.internal.storage.pagememory.mv.AbstractPageMemoryMvPartitionStorage.busy(AbstractPageMemoryMvPartitionStorage.java:749)
> at
> org.apache.ignite.internal.storage.pagememory.mv.AbstractPageMemoryMvPartitionStorage.read(AbstractPageMemoryMvPartitionStorage.java:324)
> at
> org.apache.ignite.internal.table.distributed.replicator.PartitionReplicaListener.resolveRowByPkForReadOnly(PartitionReplicaListener.java:1843)
> ... 43 more
> Caused by: org.apache.ignite.internal.pagememory.tree.CorruptedTreeException:
> IGN-CMN-65535 TraceId:069639f9-b2b9-4690-b066-3ca0bf2e9e06 B+Tree is
> corrupted [groupId=5, pageIds=[562958543400296], groupName=5, msg=Runtime
> failure on lookup row:
> org.apache.ignite.internal.storage.pagememory.mv.VersionChainKey@1b8df27b]
> at
> org.apache.ignite.internal.pagememory.tree.BplusTree.corruptedTreeException(BplusTree.java:6724)
> at
> org.apache.ignite.internal.pagememory.tree.BplusTree.findOne(BplusTree.java:1580)
> at
> org.apache.ignite.internal.storage.pagememory.mv.AbstractPageMemoryMvPartitionStorage.findVersionChain(AbstractPageMemoryMvPartitionStorage.java:870)
> ... 47 more
> Caused by: java.lang.AssertionError
> at
> org.apache.ignite.internal.pagememory.tree.BplusTree$GetPageHandler.run(BplusTree.java:6464)
> at
> org.apache.ignite.internal.pagememory.tree.BplusTree$Search.run(BplusTree.java:369)
> at
> org.apache.ignite.internal.pagememory.tree.BplusTree$GetPageHandler.run(BplusTree.java:6451)
> at
> org.apache.ignite.internal.pagememory.util.PageHandler.readPage(PageHandler.java:157)
> at
> org.apache.ignite.internal.pagememory.datastructure.DataStructure.read(DataStructure.java:417)
> at
> org.apache.ignite.internal.pagememory.tree.BplusTree.read(BplusTree.java:6693)
> at
> org.apache.ignite.internal.pagememory.tree.BplusTree.findDown(BplusTree.java:1654)
> at
> org.apache.ignite.internal.pagememory.tree.BplusTree.findDown(BplusTree.java:1663)
> at
> org.apache.ignite.internal.pagememory.tree.BplusTree.findDown(BplusTree.java:1663)
> at
> org.apache.ignite.internal.pagememory.tree.BplusTree.doFind(BplusTree.java:1632)
> at
> org.apache.ignite.internal.pagememory.tree.BplusTree.findOne(BplusTree.java:1572)
> ... 48 more
> {noformat}
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
