[ https://issues.apache.org/jira/browse/CASSANDRA-18887?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Marcus Eriksson updated CASSANDRA-18887: ---------------------------------------- Description: [CEP-21] CASSANDRA-18816 rebase fixes https://github.com/krummas/cassandra/commit/4d06d73c7e make the 18816 tests pass with TCM [CEP-21] fix cqlshlib tests https://github.com/krummas/cassandra/commit/a90cc1b98a Add cluster_metadata keyspace where needed [CEP-21] remove authsetup https://github.com/krummas/cassandra/commit/e5a8ac2607 Avoid calling doAuthSetup twice [CEP-21] serialize MemtableParams when writing TableParams https://github.com/krummas/cassandra/commit/e81ddb4e31 Missed in earlier rebase, need to serialize MemtableParams when saving a snapshot [CEP-21] Upgrading a one node cluster to TCM fails attempting Gossip shadow round https://github.com/krummas/cassandra/commit/8bcd581fc4 Upgrading fails due to first upgraded node waiting for gossip state from peers. [CEP-21] Implement replacement with same address https://github.com/krummas/cassandra/commit/c1e94aa21a Add back missing replace with same address feature [CEP-21] fix nodetool bootstrap resume https://github.com/krummas/cassandra/commit/2265c4b615 Add back missing nodetool bootstrap resume feature [CEP-21] Remove LEFT peers from system tables and exclude them during startup https://github.com/krummas/cassandra/commit/2aef6f6a96 Avoid trying to gossip with LEFT peers [CEP-21] Fix flaky distributed log test. While it fails very infrequently on CI, it consistently fails locally. https://github.com/krummas/cassandra/commit/6ea5042618 Test fix [CEP-21] Retry indefinitely for STARTUP messages. https://github.com/krummas/cassandra/commit/5924404bd6 If the CMS is down and we try to upgrade an instance we should block until the CMS comes back [CEP-21] Implement versioning for ranges https://github.com/krummas/cassandra/commit/ef1ad17669 To be able to know if the coordinator has the correct ring view we keep track of when a range was last updated. If coordinator has not seen this update we fail the request. [CEP-21] Handle case where removenode requires no streaming to restore RF https://github.com/krummas/cassandra/commit/47769f0ba2 [CEP-21] Remove redundant Keyspaces arg from SchemaTransformation::apply https://github.com/krummas/cassandra/commit/d0367189da Code cleanup [CEP-21] Don't trigger client warnings or guardrails when executing AlterSchema transformations https://github.com/krummas/cassandra/commit/ce7511f2ec Avoid getting multiple client warnings [CEP-21] Validate schema alterations on coordinator https://github.com/krummas/cassandra/commit/bf9169683b [CEP-21] Add a script for simulation running https://github.com/krummas/cassandra/commit/599d6f70a3 [CEP-21] Add nextId to Directory serialization and make it possible to bump metadata serialization version https://github.com/krummas/cassandra/commit/6dd4c92797 Previously, on bounce, each instance calculated which nodeId we should give a new node by counting the number of instances in the cluster, this can diverge so that different nodes have different ids. Fixed by serializing the current nextId when storing snapshot. [CEP-21] Add metric for CMS membership https://github.com/krummas/cassandra/commit/5dfb270d92 [CEP-21] Reenable starting without joining (-Dcassandra.join_ring=false) https://github.com/krummas/cassandra/commit/bd1e9da2f8 Add back missing join_ring=false feature. [CEP-21] Preparing DDL statements should capture the CQL string https://github.com/krummas/cassandra/commit/c312fd5273 Avoid NPE was: [CEP-21] fix cqlshlib tests https://github.com/krummas/cassandra/commit/31a4f59449 Add cluster_metadata keyspace where needed [CEP-21] remove authsetup https://github.com/krummas/cassandra/commit/4cab51d76b Avoid calling doAuthSetup twice [CEP-21] serialize MemtableParams when writing TableParams https://github.com/krummas/cassandra/commit/602c01a7b4 Missed in earlier rebase, need to serialize MemtableParams when saving a snapshot [CEP-21] Upgrading a one node cluster to TCM fails attempting Gossip shadow round https://github.com/krummas/cassandra/commit/3feb3b1a58 Upgrading fails due to first upgraded node waiting for gossip state from peers. [CEP-21] Implement replacement with same address https://github.com/krummas/cassandra/commit/a05bbc4870 Add back missing replace with same address feature [CEP-21] fix nodetool bootstrap resume https://github.com/krummas/cassandra/commit/061b835dae Add back missing nodetool bootstrap resume feature [CEP-21] Remove LEFT peers from system tables and exclude them during startup https://github.com/krummas/cassandra/commit/4bc9dc674b Avoid trying to gossip with LEFT peers [CEP-21] Fix flaky distributed log test. While it fails very infrequently on CI, it consistently fails locally. https://github.com/krummas/cassandra/commit/f231e9dda9 Test fix [CEP-21] Retry indefinitely for STARTUP messages. https://github.com/krummas/cassandra/commit/7f9e88857e If the CMS is down and we try to upgrade an instance we should block until the CMS comes back [CEP-21] Implement versioning for ranges https://github.com/krummas/cassandra/commit/a170b9f349 To be able to know if the coordinator has the correct ring view we keep track of when a range was last updated. If coordinator has not seen this update we fail the request. [CEP-21] Handle case where removenode requires no streaming to restore RF https://github.com/krummas/cassandra/commit/68054d3ec0 [CEP-21] Remove redundant Keyspaces arg from SchemaTransformation::apply https://github.com/krummas/cassandra/commit/fe1c132946 Code cleanup [CEP-21] Don't trigger client warnings or guardrails when executing AlterSchema transformations https://github.com/krummas/cassandra/commit/dbbd96c9ad Avoid getting multiple client warnings [CEP-21] Validate schema alterations on coordinator https://github.com/krummas/cassandra/commit/4711053513 [CEP-21] Add a script for simulation running https://github.com/krummas/cassandra/commit/94543e9230 [CEP-21] Add nextId to Directory serialization and make it possible to bump metadata serialization version https://github.com/krummas/cassandra/commit/6d5d6a8bc6 Previously, on bounce, each instance calculated which nodeId we should give a new node by counting the number of instances in the cluster, this can diverge so that different nodes have different ids. Fixed by serializing the current nextId when storing snapshot. [CEP-21] Add metric for CMS membership https://github.com/krummas/cassandra/commit/bbdd165013 [CEP-21] Reenable starting without joining (-Dcassandra.join_ring=false) https://github.com/krummas/cassandra/commit/565c3747c5 Add back missing join_ring=false feature. [CEP-21] Preparing DDL statements should capture the CQL string https://github.com/krummas/cassandra/commit/4608a74c64 Avoid NPE > CEP-21 - Implement missing features and various other fixes > ----------------------------------------------------------- > > Key: CASSANDRA-18887 > URL: https://issues.apache.org/jira/browse/CASSANDRA-18887 > Project: Cassandra > Issue Type: Improvement > Components: Consistency/Coordination > Reporter: Marcus Eriksson > Assignee: Marcus Eriksson > Priority: Normal > > [CEP-21] CASSANDRA-18816 rebase fixes > https://github.com/krummas/cassandra/commit/4d06d73c7e > make the 18816 tests pass with TCM > [CEP-21] fix cqlshlib tests > https://github.com/krummas/cassandra/commit/a90cc1b98a > Add cluster_metadata keyspace where needed > [CEP-21] remove authsetup > https://github.com/krummas/cassandra/commit/e5a8ac2607 > Avoid calling doAuthSetup twice > [CEP-21] serialize MemtableParams when writing TableParams > https://github.com/krummas/cassandra/commit/e81ddb4e31 > Missed in earlier rebase, need to serialize MemtableParams when saving a > snapshot > [CEP-21] Upgrading a one node cluster to TCM fails attempting Gossip shadow > round > https://github.com/krummas/cassandra/commit/8bcd581fc4 > Upgrading fails due to first upgraded node waiting for gossip state from > peers. > [CEP-21] Implement replacement with same address > https://github.com/krummas/cassandra/commit/c1e94aa21a > Add back missing replace with same address feature > [CEP-21] fix nodetool bootstrap resume > https://github.com/krummas/cassandra/commit/2265c4b615 > Add back missing nodetool bootstrap resume feature > [CEP-21] Remove LEFT peers from system tables and exclude them during startup > https://github.com/krummas/cassandra/commit/2aef6f6a96 > Avoid trying to gossip with LEFT peers > [CEP-21] Fix flaky distributed log test. While it fails very infrequently on > CI, it consistently fails locally. > https://github.com/krummas/cassandra/commit/6ea5042618 > Test fix > [CEP-21] Retry indefinitely for STARTUP messages. > https://github.com/krummas/cassandra/commit/5924404bd6 > If the CMS is down and we try to upgrade an instance we should block until > the CMS comes back > [CEP-21] Implement versioning for ranges > https://github.com/krummas/cassandra/commit/ef1ad17669 > To be able to know if the coordinator has the correct ring view we keep track > of when a range was last updated. If coordinator has not seen this update we > fail the request. > [CEP-21] Handle case where removenode requires no streaming to restore RF > https://github.com/krummas/cassandra/commit/47769f0ba2 > [CEP-21] Remove redundant Keyspaces arg from SchemaTransformation::apply > https://github.com/krummas/cassandra/commit/d0367189da > Code cleanup > [CEP-21] Don't trigger client warnings or guardrails when executing > AlterSchema transformations > https://github.com/krummas/cassandra/commit/ce7511f2ec > Avoid getting multiple client warnings > [CEP-21] Validate schema alterations on coordinator > https://github.com/krummas/cassandra/commit/bf9169683b > [CEP-21] Add a script for simulation running > https://github.com/krummas/cassandra/commit/599d6f70a3 > [CEP-21] Add nextId to Directory serialization and make it possible to bump > metadata serialization version > https://github.com/krummas/cassandra/commit/6dd4c92797 > Previously, on bounce, each instance calculated which nodeId we should give a > new node by counting the number of instances in the cluster, this can diverge > so that different nodes have different ids. Fixed by serializing the current > nextId when storing snapshot. > [CEP-21] Add metric for CMS membership > https://github.com/krummas/cassandra/commit/5dfb270d92 > [CEP-21] Reenable starting without joining (-Dcassandra.join_ring=false) > https://github.com/krummas/cassandra/commit/bd1e9da2f8 > Add back missing join_ring=false feature. > [CEP-21] Preparing DDL statements should capture the CQL string > https://github.com/krummas/cassandra/commit/c312fd5273 > Avoid NPE -- This message was sent by Atlassian Jira (v8.20.10#820010) --------------------------------------------------------------------- To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org For additional commands, e-mail: commits-h...@cassandra.apache.org