You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@ignite.apache.org by "Ivan Artukhov (JIRA)" <ji...@apache.org> on 2019/02/07 10:12:00 UTC
[jira] [Created] (IGNITE-11242) Update counters mismatch after
loading data via DataStreamer into MVCC-enabled caches
Ivan Artukhov created IGNITE-11242:
--------------------------------------
Summary: Update counters mismatch after loading data via DataStreamer into MVCC-enabled caches
Key: IGNITE-11242
URL: https://issues.apache.org/jira/browse/IGNITE-11242
Project: Ignite
Issue Type: Bug
Components: mvcc, streaming
Affects Versions: 2.7
Reporter: Ivan Artukhov
*Steps:*
Start 1000 caches in 100 cache groups. Caches configuration:
{code:xml}
<property name="atomicityMode" value="TRANSACTIONAL_SNAPSHOT"/>
<property name="backups" value="2"/>
<property name="cacheMode" value="PARTITIONED"/>
<property name="writeSynchronizationMode" value="FULL_SYNC"/>
<property name="partitionLossPolicy" value="READ_WRITE_SAFE"/>
{code}
Preload 10000 keys into each cache using DataStreamer
Run *VerifyBackupPartitionsTaskV2* to check partition inconsistencies between hashes or update counters
*Expected:*
There are no differences between primary and backup partitions
*Actual:*
Two or three *backup* partitions in each cache group have update counters which differ from corresponding primary partitions. E.g.:
{code}
[2019-02-07 11:04:23,286][INFO ][Task-threadNum-0] Conflict partition: PartitionKeyV2 [grpId=-1093321804, grpName=cache_group_100, partId=261]
[2019-02-07 11:04:23,286][INFO ][Task-threadNum-0] Partition instances: [PartitionHashRecordV2 [isPrimary=true, consistentId=id-0, updateCntr=100, size=100, partHash=665729765], PartitionHashRecordV2 [isPrimary=false, consistentId=id-1, updateCntr=91, size=100, partHash=665729765], PartitionHashRecordV2 [isPrimary=false, consistentId=id-2, updateCntr=100, size=100, partHash=665729765]]
{code}
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)