Best effort attempt to get tableId from CFS, fallback to null which
still allows snapshot to be loaded
End result is that nodetool clearsnapshot and listsnapshot function for such
snapshots on restart
In addition to 5.0 patch we move starting SnapshotManager to after TCM is initiliazed.
The consequence of such a move is that cleanup of ephemeral snapshots happens after startup checks pass.
patch by Matt Byrd; reviewed by Stefan Miklosovic, Brandon Williams and Francisco Guerrero for CASSANDRA-21246
In 6.0, only rename test/data/serialization/5.1 to 6.0
In 7.0, same plus update edges in UpgradeTestBase
patch by Stefan Miklosovic; reviewed by Dmitry Konstantinov for CASSANDRA-21319
This is the continuation of CASSANDRA-21319 to make branching of cassandra-6.0 done.
The change in .build/run-ci will start to route jobs to cassandra-6.0 pipeline in pre-ci.
patch by Stefan Miklosovic; reviewed by Brandon Williams for CASSANDRA-21319
Add an extra optimization if it is enabled: invoke Thread.setName only if a phase changed
patch by Dmitry Konstantinov; reviewed by Francisco Guerrero, Jyothsna Konisa for CASSANDRA-21089
patch by C. Scott Andreas; reviewed by Caleb Rackliffe and Dmitry Konstantinov for CASSANDRA-21285
Co-authored-by: C. Scott Andreas <cscotta@users.noreply.github.com>
Co-authored-by: Caleb Rackliffe <calebrackliffe@gmail.com>
Also Add:
- txn_graph and txn_graph_all virtual tables for debugging transaction dependency graphs
- txn_cache virtual table to debug accord cache contents
Also Fix:
- NPE in AccordDebugKeyspace.TxnOpsTable
- AccordCoordinatorMetrics latency metrics should use Timer not Histogram
- OnDiskIndex.readLast
Also Improve:
- Catchup should run after progress log is started
- Cancelling AccordTask should early-terminate range scanning
- Improve RangeTxnScanner description
- Add ballot to txn tables
- Add NodeId to logging output
patch by Benedict; reviewed by Alex Petrov for CASSANDRA-21306
With negotiated authentication (CEP-50), nodes may be configured with multiple authenticators. Prior to this change,
a number of areas in the code assumed that there was a single configured authenticator and contained logic that
switched depending on the authenticator type. This logic won't work when multiple authenticators can be configured.
This change eliminates most calls to DataDescriptor.getAuthenticator(), by enabling individual authenticators to
declare the role attributes they support, requiring callers to specify the type of authenticator they're looking
for, and directly returning whether the node can enforce authn or not rather than inferring it by the presence of
an authenticator.
Testing done: Unit tests for auth and config packages; d-tests for auth-related functionality (e.g. ColumnMasks).
patch by Joel Shepherd; reviewed by Stefan Miklosovic, Andy Tolbert for CASSANDRA-20834
The same idea was applied previously in UnfilteredSerializer#serializeRowBody. A state is passed into a consumer function using the existing helper object.
patch by Dmitry Konstantinov; reviewed by Caleb Rackliffe, Francisco Guerrero for CASSANDRA-21289
Rust CQL did not have any commits in 9 years. It doesn't even support
protocol V4, and most of the features that a user could expect from the
driver - like load balancing or pagination. No reasonable app would use
it in production.
This commit replaces it with Scylla Rust Driver on the page with
recommended drivers. Scylla Rust Driver is actively developed and
feature-rich. Even though it is primarily developed for Scylla, we do
always make sure it is also compatible with Cassandra and tested with
it.
The original second sentence implied per-table scope, which contradicted
the first sentence. Replaced with explicit keyspace-scope wording and
removed the redundant vector/non-vector qualifier.
The original second sentence implied per-table scope, which contradicted
the first sentence. Replaced with explicit keyspace-scope wording and
removed the redundant vector/non-vector qualifier.