DSE 6.8.x is compatible with Apache Cassandra™ 3.11 and adds additional production-certified changes, if any. Components that are indicated with an asterisk (*) (if any) are known to be updated since the prior patch version.
18 May 2021
- Apache Solr™ 6.0.1.4.2840
- Apache Spark™ 2.4.0.18
- Apache TinkerPop™ 3.4.5-20200107-6cec00d8
- Apache Tomcat® 8.5.65
- DSE Java Driver 1.10.0-dse+20210424
- Netty 4.1.25.7.dse
- Spark JobServer 0.8.0.50
- AOSS returns additional parameter in
statusendpoint: "connection_hostname". The new parameter is a FQDN of the node hosting AOSS, it may be used for connections (instead of connection_address) if needed. (DSP-21811)
- Fixed an issue with DSE daemon being unable to stop after the default timeout expired. This issue only affected systems that use package install and init.d, such as centos. (DSP-21804)
- Fixes a problem where the Gremlin
phrase()predicate may not match Solr results for equivalent search. (DSP-21724)
- A new JVM option is added:
dse.search.fc.warmup:AUTO,ALWAYS&NEVER. (DSP-21813)
26 April 2021
- Apache Solr™ 6.0.1.4.2840*
- Apache Spark™ 2.4.0.18
- Apache TinkerPop™ 3.4.5-20200107-6cec00d8
- Apache Tomcat® 8.5.65*
- DSE Java Driver 1.10.0-dse+20200217
- Netty 4.1.25.7.dse
- Spark JobServer 0.8.0.50
- Fix an error in cqlsh encoding unicode in multi-line statements (DB-4855)
- Make cqlsh prefer newer TLS versions. (DB-4966)
- Address a problem where new or rebooted nodes may not be able to gossip with peers. (DSP-21753)
- Upgrade apache commons-compress to address CVE-2019-12402 (DSP-21679)
- Update tomcat version 8.5.61 to 8.5.65 (DSP-21798)
- Fixed a bug where under heavy load solr query worker threads would use 100% CPU due to contention on thread local map (DSP-21746)
9 April 2021
- Apache Solr™ 6.0.1.4.2814
- Apache Spark™ 2.4.0.18
- Apache TinkerPop™ 3.4.5-20200107-6cec00d8
- Apache Tomcat® 8.5.61*
- DSE Java Driver 1.10.0-dse+20200217
- Netty 4.1.25.7.dse
- Spark JobServer 0.8.0.50*
- Fixes an issue where a login attempt with missing credentials logged a misleading warning message with stack trace instead of an error message about the missing username or password. (DB-4806)
- Fixes a bug when in rare cases a terminated repair session would leak on-heap memory (DB-4833)
- Print a timestamp when nodetool exits due to an error (DB-4826)
- Data export from cqlsh is now less noisy in the logs (DSP-21494)
- Fixes intermittent ERROR: java.util.ConcurrentModificationException at org.apache.cassandra.transport.CBUtil.writeStringList (DSP-21336)
- Fix for DESCRIBE TYPES in cqlsh (DSP-21667)
- Add asynchronous update to KMIP key cache to fix blocking of commit log (DSP-20582)
- Fixes CVE-2020-1945 affecting Apache Ant (DSP-21716)
- Fixes SRCCLR-SID-22742: Insecure Input Validation Vulnerability in the Apache Commons Codec library (DSP-21747)
- Update Tomcat version 8.0.53 to 8.5.61 (fixes CVE-2002-0493 CVE-2009-3548 CVE-2013-2185 CVE-2016-1240 CVE-2016-5018 CVE-2016-5388 CVE-2016-6796 CVE-2016-6797 CVE-2016-8745 CVE-2016-9774 CVE-2016-9775 CVE-2020-8022) (DSP-21394)
- Fixes a severe issue where flushing an empty MemtableIndex causes the index to not be queryable (DB-4934)
- Fixes a bug where FilterCache warmup triggered by node health change can block GossipStage-1 thread for several seconds (DSP-21674)
- Fixes CVE-2014-0114, CVE-2014-0114 (DSP-21668)
11 March 2021
- Apache Solr™ 6.0.1.4.2814
- Apache Spark™ 2.4.0.18*
- Apache TinkerPop™ 3.4.5-20200107-6cec00d8
- Apache Tomcat® 8.0.53
- DSE Java Driver 1.10.0-dse+20200217
- Netty 4.1.25.7.dse
- Spark JobServer 0.8.0.49
- Works around a bug in JDK 1.8u282 (JDK-8260018) (DB-4884)
- Fixes a Null Pointer Excpetion in Gossip when upgrading from 5.1 to 6.8.6 (DB-4810)
- During package upgrade yum and apt managers overwrite unedited old jvm.options file. (DB-4705)
- Fixes a problem where FSReadError during streaming could causes DSE to shutdown (DB-4878)
- Port fix of CVE-2020-17516 onto DSE 6.8 (DB-4923)
- Dropped messages metrics calculation doesn't cause assertion errors when dropped messages contain remote batch mutation. (DB-3905)
- SSTablePartitions tool will no longer fail with "histogram overflowed" when its working for the server code (DB-2952)
- Fixes a problem where sstablescrub could not fix a corrupted file (DSP-21672)
- Addressed several Jackson databind vulnerabilities by upgrading jackson-databind to version 2.9.10.8 in DSE 5.1.21, 6.0.15 and 6.7.13 and version 2.10.5.1 in DSP 6.8.10. (DSP-21503) (DSP-21503)
- Fixes a problem where nodetool rebuild could fail intermittently with zerocopy streaming enabled (DSP-21564)
- Update Jetty to 9.4.34.v20201102 and update Spark Versions: DSE 5.1: 2.0.2.38; DSE 6.0: 2.2.3.16; DSE 6.7: 2.2.3.16; DSE 6.8: 2.4.0.17 (DSP-21506)
- SCC by default enables direct join optimization only when size_estimates for both tables are available. (DSP-21628)
- Fix: Spark Master fails to start if keystore (used by web UI) contains more than one certificate (DSP-21703)
- Both graph engines now accept either
byte[]orByteBufferfor blob-typed property values. (DSP-21643)
- Index segments are now merged into a single segment, after the index build. (DSP-19608)
- Fixes a problem where lucene threads were getting interrupted, causing problems with solr cores (DSP-21339)
- Search queries will no longer fail when querying clustering columns of certain types on which the order is reversed (DSP-21363)
- Spark Cassandra Connector supports Storage Attached Indexes (SAI). The connector pushes down predicates defined on columns with SAI indexes. (DSP-21655)
- DSE Spark supports connections to Astra clusters (DSP-21510)
7 January 2021
- Apache Solr™ 6.0.1.4.2814
- Apache Spark™ 2.4.0.16
- Apache TinkerPop™ 3.4.5-20200107-6cec00d8
- Apache Tomcat® 8.0.53
- DSE Java Driver 1.10.0-dse+20200217
- Netty 4.1.25.7.dse
- Spark JobServer 0.8.0.49
- Addresses a severe issue where streaming an older file format to a new node could crash sending and receiving nodes. (DB-4846)
- Add support for multiple authorization sources (LDAP + DSE Internal) (DSP-14233)
- Fixed direct join optimization for spark sql. (DSP-21498)
15 December 2020
6.8.7 and DSE 6.8.8, these releases have been retracted. We recommend against upgrading to these versions at this time. If you have already upgraded to these versions, please EITHER set zerocopy_streaming_enabled=false in the cassandra.yaml and perform a rolling restart AND/OR run upgradesstables on all nodes in your cluster before adding new nodes, running repair, or restoring from backups. This bug has been addressed in DSE 6.8.9. All features and fixes for 6.8.8 and 6.8.7 are present in 6.8.9.
- Apache Solr™ 6.0.1.4.2814*
- Apache Spark™ 2.4.0.16
- Apache TinkerPop™ 3.4.5-20200107-6cec00d8
- Apache Tomcat® 8.0.53
- DSE Java Driver 1.10.0-dse+20200217
- Netty 4.1.25.7.dse
- Spark JobServer 0.8.0.49
- During Backup Service startup, to avoid stalling during service initialization, check cluster readiness in response to endpoint "on alive" events. (DB-4818)
- Fixed an issue in DSE that prevented the Management API to work with DSE versions 6.8.5 - 6.8.7 (DSP-21607)
- The issue: When flushing text column indexes, the internal ordering of terms can be processed out of order causing an "*java.lang.AssertionError: Incremental trie requires sorted keys*" error. When this happens, all flushing of indexes involved in this transaction is aborted and the indexes are marked non-queryable. Recovering from this issue involves either rebuilding the indexes or restarting the nodes. (DSP-21580)
- Fixes a performance regression in SAI for versions 6.8.6 and 6.8.7 regarding *MultiRangeReadCommand*. (DSP-21601)
- A system property
dse.solr.fuzzy.max.expansionwas added which allows the user to define a custom number of fuzzy query expansions. The maximal possible value is 1024. When unset, the default number of max expansions is 50. (DSP-21605)
- Adjust available framework values for
--frameworkparameter. (DSP-21500)
23 November 2020
6.8.7 and DSE 6.8.8, these releases have been retracted. We recommend against upgrading to these versions at this time. If you have already upgraded to these versions, please EITHER set zerocopy_streaming_enabled=false in the cassandra.yaml and perform a rolling restart AND/OR run upgradesstables on all nodes in your cluster before adding new nodes, running repair, or restoring from backups. This bug has been addressed in DSE 6.8.9. All features and fixes for 6.8.8 and 6.8.7 are present in 6.8.9.
- Apache Solr™ 6.0.1.4.2794
- Apache Spark™ 2.4.0.16
- Apache TinkerPop™ 3.4.5-20200107-6cec00d8
- Apache Tomcat® 8.0.53
- DSE Java Driver 1.10.0-dse+20200217
- Netty 4.1.25.7.dse
- Spark JobServer 0.8.0.49
- Fixed a bug where the slow query log would fill with queries that do not meet the slow query threshold (DSP-21417)
- Fixed a bug where a single partition read might fail if the following conditions were true:
- several sstables had the same partition level deletion info
- some of the sstables had wide rows whereas others had not
- the sstables in question contained range tombstone markers (DSP-21346)
- Fixed a bug in
cassandra.repair.mutation_repair_rows_per_batchsetting that caused sending all repair mutations at once (DSP-21429)
- On the CQLSH
CREATE CUSTOM INDEX ... WITH OPTIONSstatement, SAI adds support for an ascii option. If set totrue, converts alphabetic, numeric, and symbolic characters that are not in the Basic Latin Unicode block (first 127 ASCII characters) to their ASCII equivalent, if one exists. For example, the filter changes à to a. The default isfalse. (DSP-21409) - Make the SAI read path synchronous. (DSP-21451)
- Fixed "java.lang.ArithmeticException: integer overflow" printing in
system.logwhen retrieving the SAI indexsegmentRowID(DSP-21522
- A meaningful error message is logged when two properties with the same name but different types are used in a single core graph. Classic graph was not affected. (DSP-21490)
- Optimized retrieval when
memberof_searchused the wrong attribute to retrieve groups of the user. (DSP-21537)
- Multi-datacenter backup and restore, new
CompositeStoretype of backup store. (DB-4489) - Adds the possibility to restore a backup marked as
INCOMPLETEby using the newFORCE RESTOREstatement.
- Addressed a bug where a "CommitLogReplayException" is caused by a bad header but correct CRC after restart (DB-3996)
- Fixed a bug where some part of the commit log might not be replayed after injecting a foreign sstable to a node or, on 6.8, after zero-copy streaming of an sstable (DB-4629)
- Fixed an issue where zero copy streaming could cause file descriptor leakage (DB-4594)
- SStableloader now uses
native_transport_port_sslovernative_transport_portwhen passed a config file with the property set (DB-4632)
- Fixed memory leak in Netty resulting in OOM. (DB-4664)
- Fixed a problem in the scheduling and counting of active materialized view updates that could cause too many to be executed concurrently, overwhelming the node. (DB-4782)
12 November 2020
- Apache Solr™ 6.0.1.4.2794
- Apache Spark™ 2.4.0.16
- Apache TinkerPop™ 3.4.5-20200107-6cec00d8
- Apache Tomcat® 8.0.53
- DSE Java Driver 1.10.0-dse+20200217
- Netty 4.1.25.7.dse
- Spark JobServer 0.8.0.49
- A node may be stuck in repair while joining the cluster if broadcast_address is set differently than local_address (DB-4786)
20 October 2020
- Apache Solr™ 6.0.1.4.2794
- Apache Spark™ 2.4.0.16
- Apache TinkerPop™ 3.4.5-20200107-6cec00d8
- Apache Tomcat® 8.0.53
- DSE Java Driver 1.10.0-dse+20200217
- Netty 4.1.25.7.dse
- Spark JobServer 0.8.0.49
- Server side backup and restore now supports Microsoft Azure cloud storage as a backup target. (DB-3894)
- snapshot
schema.cqlfiles will now containIF NOT EXISTSclause forCREATE TYPEstatements (DB-4685)
- Fixes a problem where races in notifying compaction strategies of added and removed sstables can cause compaction to try to use non-existing sstables and repeatedly fail to make progress. (DB-4711)
- Fixed node restart issue after dropping a PointType column. (DSP-21326)
- Fixed extreme local pauses on all nodes in the cluster on a node restart. (DB-4657)
- Improves performance of estimation of partition counts for subranges. (DB-3679)
- Fsync nodes metadata to prevent FSReadError issues on startup. (DB-4672)
- Fixes LDAP user permissions problem following LDAP server restart. (DSP-21284)
- Fixes LDAP user permissions problem following LDAP server restart. (DSP-21284)
- Escape single-quotes in certain graph-search query predicates. (DSP-21450)
- Fix: Spark Application contacting Nodes in Non Local DC (DSP-19961)
DataStax Enterprise (DSE) 6.8.4 includes all changes from previous DSE versions. See TinkerPop upgrade documentation for all changes.
17 September 2020
- Apache Solr™ 6.0.1.4.2794
- Apache Spark™ 2.4.0.16
- Apache TinkerPop™ 3.4.5-20200107-6cec00d8
- Apache Tomcat® 8.0.53
- DSE Java Driver 1.10.0-dse+20200217
- Netty 4.1.25.7.dse
- Spark JobServer 0.8.0.49
- Fixes compaction getting stuck on acquiring references for non-existing sstables. (DB-4290)
- Distributes Netty connections more uniformly across TPC cores (DB-4683)
- Distributes Netty connections more uniformly across TPC cores (DB-4683)
- Adds TTL and TimeWindowCompactionStrategy (TWCS) to
system_distributed.repair_historyandsystem_distributed.parent_repair_historytables. (DB-2009) - DNS Service Discovery is now a part of the DSE/LDAP integration. (DSP-11450)
- New system property to cap the maximum amount of memory used by bloom filters:
-Dcassandra.max_bf_memory_mb}. By default, this is unlimited. (DSP-21344)
- DNS Service Discovery is now a part of the DSE/LDAP integration. (DSP-11450)
- DSEFS waits for a schema agreement before starting and issuing the first CQL query. (DSP-20743)
- Storage-Attached Indexing (SAI) adds support for creating multiple SAI indexes on the same collection map column. See SAI collection map examples with keys, values, and entries. (DSP-21306)
DataStax Enterprise (DSE) 6.8.4 includes all changes from previous DSE versions. See TinkerPop upgrade documentation for all changes.
Release notes for previous DSE patch releases can be found here: https://docs.datastax.com/en/dse/6.8/dse-admin/datastax_enterprise/releaseNotes/RNdse.html#RNdse