[
https://issues.apache.org/jira/browse/CASSANDRA-21848?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18138137#comment-18138137
]
Dmitry Konstantinov commented on CASSANDRA-21848:
-------------------------------------------------
It seems that after enabling paxos v2 in cassandra_latest.yaml we have now
bootstrap runs repairPaxosForTopologyChange.
The test sets num_tokens: 256. That makes one PaxosCleanup per range per table:
779 ranges × 2 tables (standard1, counter1) = 1558 sessions, each on 3
replicas.
- Each PAXOS2_CLEANUP_COMPLETE_REQ does a blocking flush of
system.paxos_repair_history (PaxosCleanupComplete → cfs.onPaxosRepairComplete →
TablePaxosRepairHistory.add(..., flush=true)). So, we have a lot of tiny
sstable flushes, one at a time per node. For slow disk hosts it can become a
challenge.
The test is focused on about disk balance, not paxos, so we can use
-Dcassandra.skip_paxos_repair_on_topology_change=true as a workaround to reduce
disk pressure (similar to replace_address_test.py and bootstrap_test.py)
> Flaky test: TestDiskBalance.test_disk_balance_bootstrap
> -------------------------------------------------------
>
> Key: CASSANDRA-21848
> URL: https://issues.apache.org/jira/browse/CASSANDRA-21848
> Project: Apache Cassandra
> Issue Type: Bug
> Components: Test/dtest/python
> Reporter: Dmitry Konstantinov
> Priority: Normal
> Labels: green_ci
>
> https://ci-cassandra.apache.org/job/Cassandra-5.0/698/testReport/junit/dtest-large-latest.disk_balance_test/TestDiskBalance/Tests___dtest_large_latest_jdk11_3_6___test_disk_balance_bootstrap/
--
This message was sent by Atlassian Jira
(v10.3.26#10030026)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
