[
https://issues.apache.org/jira/browse/CASSANDRA-18638?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Brandon Williams updated CASSANDRA-18638:
-----------------------------------------
Resolution: Invalid
Status: Resolved (was: Triage Needed)
As you have found, NODE_LOCAL is not a supported consistency level and is not
intended for end users.
> ERROR: Invalid consistency level: NODE_LOCAL when using partition denylist in
> 4.1.2
> -----------------------------------------------------------------------------------
>
> Key: CASSANDRA-18638
> URL: https://issues.apache.org/jira/browse/CASSANDRA-18638
> Project: Cassandra
> Issue Type: Bug
> Reporter: Ke Han
> Priority: Normal
> Attachments: cassandra.yaml, system.log
>
>
> When I was testing the upgrade from Cassandra-3.11.15 to Cassandra-4.1.2
> (single node), for the debugging purpose, I enabled
> partition_denylist_enabled and set the partition consistency level to
> {_}NODE_LOCAL{_}.
> I encountered the following error message in the system log
> {code:java}
> ERROR [main] 2023-06-29 14:12:17,870 PartitionDenylist.java:156 - Failed to
> load partition denylist
> java.lang.UnsupportedOperationException: Invalid consistency level: NODE_LOCAL
> at
> org.apache.cassandra.db.ConsistencyLevel.blockFor(ConsistencyLevel.java:168)
> at
> org.apache.cassandra.locator.ReplicaPlans.contactForRead(ReplicaPlans.java:558)
> at
> org.apache.cassandra.locator.ReplicaPlans.forRangeRead(ReplicaPlans.java:611)
> at
> org.apache.cassandra.service.reads.range.ReplicaPlanIterator.computeNext(ReplicaPlanIterator.java:76)
> at
> org.apache.cassandra.service.reads.range.ReplicaPlanIterator.computeNext(ReplicaPlanIterator.java:42)
> at
> org.apache.cassandra.utils.AbstractIterator.hasNext(AbstractIterator.java:47)
> at java.util.Iterator.forEachRemaining(Iterator.java:115)
> at
> org.apache.cassandra.service.reads.range.RangeCommands.sufficientLiveNodesForSelectStar(RangeCommands.java:135)
> at
> org.apache.cassandra.schema.PartitionDenylist.checkDenylistNodeAvailability(PartitionDenylist.java:169)
> at
> org.apache.cassandra.schema.PartitionDenylist.initialLoad(PartitionDenylist.java:148)
> at
> org.apache.cassandra.service.StorageProxy.initialLoadPartitionDenylist(StorageProxy.java:2914)
> at
> org.apache.cassandra.service.StorageService.joinTokenRing(StorageService.java:1224)
> at
> org.apache.cassandra.service.StorageService.joinTokenRing(StorageService.java:1145)
> at
> org.apache.cassandra.service.StorageService.initServer(StorageService.java:936)
> at
> org.apache.cassandra.service.StorageService.initServer(StorageService.java:854)
> at
> org.apache.cassandra.service.CassandraDaemon.setup(CassandraDaemon.java:424)
> at
> org.apache.cassandra.service.CassandraDaemon.activate(CassandraDaemon.java:751)
> at
> org.apache.cassandra.service.CassandraDaemon.main(CassandraDaemon.java:875)
> INFO [main] 2023-06-29 14:12:17,870 PartitionDenylist.java:163 - Exception
> while loading partition denylist cache. Scheduled retry in 5 seconds. {code}
> h2. How to reproduce
> I encountered it after the upgrade process. But I think it can also be
> reproduced without the upgrade process.
> 1. Set up Cassandra-4.1.2 single node, add the follow configurations to
> cassandra.yaml
> {code:java}
> denylist_consistency_level: NODE_LOCAL
> partition_denylist_enabled: true{code}
> 2. Start up Cassandra, the error message will show in system.log file.
> h2. Root Cause
> After setting denylist_consistency_level to NODE LEVEL, when loading
> denylist, it invokes the _blockFor_ method
> {code:java}
> public int blockFor(AbstractReplicationStrategy replicationStrategy)
> {
> switch (this)
> {
> case ONE:
> case LOCAL_ONE:
> return 1;
> case ANY:
> return 1;
> case TWO:
> return 2;
> case THREE:
> return 3;
> case QUORUM:
> case SERIAL:
> return quorumFor(replicationStrategy);
> case ALL:
> return replicationStrategy.getReplicationFactor().allReplicas;
> case LOCAL_QUORUM:
> case LOCAL_SERIAL:
> return localQuorumForOurDc(replicationStrategy);
> case EACH_QUORUM:
> if (replicationStrategy instanceof NetworkTopologyStrategy)
> {
> NetworkTopologyStrategy strategy = (NetworkTopologyStrategy)
> replicationStrategy;
> int n = 0;
> for (String dc : strategy.getDatacenters())
> n += localQuorumFor(replicationStrategy, dc);
> return n;
> }
> else
> {
> return quorumFor(replicationStrategy);
> }
> default:
> throw new UnsupportedOperationException("Invalid consistency
> level: " + toString());
> }
> } {code}
> However, the _NODE_LOCAL_ enum is not included in those cases and thus
> results in the exception message.
> Should this enum be included, or does Cassandra forbid users from using
> partition denylist with the _NODE_LOCAL_ consistency level (even for
> debugging purposes)?
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
