Release of Vitess v18.0.0

Summary

Major Changes

In previous releases the local examples were
explicitly using etcd v2 storage (etcd --enable-v2=true) and API (ETCDCTL_API=2) mode. We have now
removed this legacy etcd usage and instead use the new (default) etcd v3 storage and API. Please see
PR #13791 for details. If you are using the local
examples in any sort of long-term non-testing capacity, then you will need to explicitly use the v2 storage
and API mode or migrate your existing data from v2 to v3.

New command line flags and behavior

VTOrc flag `--allow-emergency-reparent`

VTOrc has a new flag --allow-emergency-reparent that specifies whether VTOrc is allowed to run emergency
failover operations. Users that want VTOrc to fix replication issues, but don't want it to run any failovers
should use this flag. This flag defaults to true which corresponds to the default behavior from prior releases.

VTOrc flag `--change-tablets-with-errant-gtid-to-drained`

VTOrc has a new flag --change-tablets-with-errant-gtid-to-drained that allows users to choose whether VTOrc should change the
tablet type of tablets with errant GTIDs to DRAINED. By default, this flag is disabled.

This feature allows users to configure VTOrc such that any tablet that encounters errant GTIDs is automatically taken out of the
serving graph. These tablets can then be inspected for what the errant GTIDs are, and once fixed, they can rejoin the cluster.

ERS sub flag `--wait-for-all-tablets`

vtctldclient command EmergencyReparentShard has a new sub-flag --wait-for-all-tablets that makes EmergencyReparentShard wait
for a response from all the tablets. Originally EmergencyReparentShard was meant only to be run when a primary tablet is unreachable.
We have realized now that there are cases when replication is broken but all tablets are reachable. In these cases, it is advisable to
call EmergencyReparentShard with --wait-for-all-tablets so that it does not ignore any of the tablets.

VTGate GRPC stream execute session flag `--grpc-send-session-in-streaming`

This flag enables transaction support on VTGate's StreamExecute gRPC API.
When this is enabled, StreamExecute will return the session in the last packet of the response.
Users should enable this flag only after client code has been changed to expect such a packet.

It is disabled by default.

Experimental Foreign Key Support

A new optional field foreignKeyMode has been added to the VSchema. This field can be provided for each keyspace. The VTGate flag --foreign_key_mode has been deprecated in favor of this field.

There are 3 foreign key modes now supported in Vitess -

unmanaged -
This mode represents the default behavior in Vitess, where it does not manage foreign key column references. Users are responsible for configuring foreign keys in MySQL in such a way that related rows, as determined by foreign keys, reside within the same shard.
managed [EXPERIMENTAL] -
In this experimental mode, Vitess is fully aware of foreign key relationships and actively tracks foreign key constraints using the schema tracker. VTGate will handle DML operations with foreign keys and correctly cascade updates and deletes.
It will also verify restrict constraints and validate the existence of parent rows before inserting child rows.
This ensures that all child operations are logged in binary logs, unlike the InnoDB implementation of foreign keys.
This allows the usage of VReplication workflows with foreign keys.
Implementation details are documented in the RFC for foreign keys.
disallow -
In this mode Vitess explicitly disallows any DDL statements that try to create a foreign key constraint. This mode is equivalent to running VTGate with the flag --foreign_key_mode=disallow.

In addition to query support, there is a new flag to MoveTables called --atomic-copy which should be used to import data into Vitess from databases which have foreign keys defined in the schema.

Upgrade process

After upgrading from v17 to v18, users should specify the correct foreign key mode for all their keyspaces in the VSchema using the new property.
Once this change has taken effect, the deprecated flag --foreign_key_mode can be dropped from all VTGates. Note that this is only required if running in disallow mode.
No action is needed to use unmanaged mode.

VTAdmin

vtadmin-web updated to node v18.16.0 (LTS)

Building vtadmin-web now requires node >= v18.16.0 (LTS). Breaking changes from v16 to v18 are listed
in https://nodejs.org/en/blog/release/v18, but none apply to VTAdmin. Full details on node v18.16.0 are listed
on https://nodejs.org/en/blog/release/v18.16.0.

Deprecations and Deletions

Legacy Client Binaries

vtctldclient is our new modern Vitess controller daemon (vtctld) client – which you will use to perform commands
and take actions in your Vitess clusters. It is replacing the legacy vtctl/vtctlclient binaries.
Some of the benefits are:

Dedicated RPCs for each command
that are used between vtctldclient and vtctld – this offers clean separation of commands and makes it easier to
develop new features without impacting other commands. This also presents an API that other clients (both Vitess and
3rd party) can use to interface with Vitess.
Use of modern frameworks: pFlag, Cobra, and Viper.
This makes development easier while also offering a better UX. For example, this offers a way to use
configuration files with support for
dynamic configuration (see also).
The reference documentation is now built through code. This
removes a burden from developers while helping users by ensuring the docs are always correct and up-to-date.

In Vitess 18 we have completed migrating all client commands to vtctldclient – the last ones being the OnlineDDL
and VReplication commands. With this work now completed, the
legacy vtctl/vtctlclient binaries are now fully deprecated and we plan to remove them in Vitess 19. You should
begin your transition before upgrading to Vitess 18.

Deprecated Command Line Flags

Throttler related vttablet flags:

--throttle_threshold is deprecated and will be removed in v19
--throttle_metrics_query is deprecated and will be removed in v19
--throttle_metrics_threshold is deprecated and will be removed in v19
--throttle_check_as_check_self is deprecated and will be removed in v19
--throttler-config-via-topo is deprecated after assumed true in v17. It will be removed in a future version.

Cache related vttablet flags:

--queryserver-config-query-cache-lfu is deprecated and will be removed in v19. The query cache always uses a LFU implementation now.
--queryserver-config-query-cache-size is deprecated and will be removed in v19. This option only applied to LRU caches, which are now unsupported.

Buffering related vtgate flags:

--buffer_implementation is deprecated and will be removed in v19

Cache related vtgate flags:

--gate_query_cache_lfu is deprecated and will be removed in v19. The query cache always uses a LFU implementation now.
--gate_query_cache_size is deprecated and will be removed in v19. This option only applied to LRU caches, which are now unsupported.

VTGate flags:

--schema_change_signal_user is deprecated and will be removed in v19
--foreign_key_mode is deprecated and will be removed in v19. For more detail read the foreign keys section.

VDiff v1:

VDiff v2 was added in Vitess 15 and marked as GA in 16.
The legacy v1 client command is now deprecated in Vitess 18 and will be removed in Vitess 19.
Please switch all of your usage to the new VDiff client command ASAP.

Deprecated Stats

The following EmergencyReparentShard stats are deprecated in v18 and will be removed in v19:

ers_counter
ers_success_counter
ers_failure_counter

These metrics are replaced by new reparenting stats introduced in v18.

VTBackup stat DurationByPhase is deprecated. Use the binary-valued Phase stat instead.

Deleted Command Line Flags

Flags in vtcombo:

--vtctld_addr

Flags in vtctldclient ApplySchema:

--skip-preflight

Flags in vtctl ApplySchema:

--skip_preflight

Flags in vtgate:

--vtctld_addr

Flags in vttablet:

--vtctld_addr
--use_super_read_only
--disable-replication-manager
--init_populate_metadata
--queryserver-config-pool-prefill-parallelism
--queryserver-config-stream-pool-prefill-parallelism
--queryserver-config-transaction-pool-prefill-parallelism
--queryserver-config-schema-change-signal-interval
--enable-lag-throttler

Flags in vtctld:

--vtctld_show_topology_crud
--durability_policy

Flags in vtorc:

--lock-shard-timeout
--orc_web_dir

Deleted `v3` planner

The Gen4 planner has been the default planner since Vitess 14. The v3 planner was deprecated in Vitess 15 and has been removed in Vitess 18.

Deleted `k8stopo`

k8stopo was deprecated in Vitess 17, see #13298. It has now been removed.

Deleted `vtgr`

vtgr was deprecated in Vitess 17, see #13300. It has now been removed.

Deleted `query_analyzer`

The undocumented query_analyzer binary has been removed in Vitess 18, see #14054.

New stats

VTGate Vindex unknown parameters

The VTGate stat VindexUnknownParameters gauges unknown Vindex parameters found in the latest VSchema pulled from the topology.

VTBackup `Phase` stat

In v17, the vtbackup stat DurationByPhase stat was added to measure the time spent by vtbackup in each phase. This stat turned out to be awkward to use in production, and has been replaced in v18 by a binary-valued Phase stat.

Phase reports a 1 (active) or a 0 (inactive) for each of the following phases:

CatchupReplication
InitialBackup
RestoreLastBackup
TakeNewBackup

To calculate how long vtbackup has spent in a given phase, sum the 1-valued data points over time and multiply by the data collection or reporting interval. For example, in Prometheus:

sum_over_time(vtbackup_phase{phase="TakeNewBackup"}) * <interval>

VTBackup `PhaseStatus` stat

PhaseStatus reports a 1 (active) or a 0 (inactive) for each of the following phases and statuses:

CatchupReplication phase has statuses Stalled and Stopped.
- Stalled is set to 1 when replication stops advancing.
- Stopped is set to 1 when replication stops before vtbackup catches up with the primary.

Backup and restore metrics for AWS S3

Requests to AWS S3 are instrumented in backup and restore metrics. For example:

vtbackup_backup_count{component="BackupStorage",implementation="S3",operation="AWS:Request:Send"} 823
vtbackup_backup_duration_nanoseconds{component="BackupStorage",implementation="S3",operation="AWS:Request:Send"} 1.33632421437e+11
vtbackup_restore_count{component="BackupStorage",implementation="S3",operation="AWS:Request:Send"} 165
vtbackup_restore_count{component="BackupStorage",implementation="S3",operation="AWS:Request:Send"} 165

VTCtld and VTOrc reparenting stats

New VTCtld and VTOrc stats were added to measure frequency of reparents by keyspace/shard:

emergency_reparent_counts - Number of times EmergencyReparentShard has been run. It is further subdivided by the keyspace, shard and the result of the operation.
planned_reparent_counts - Number of times PlannedReparentShard has been run. It is further subdivided by the keyspace, shard and the result of the operation.

Also, the reparent_shard_operation_timings stat was added to provide per-operation timings of reparent operations.

VTTablet

New ResetSequences rpc

A new VTTablet RPC ResetSequences has been added, which is being used by MoveTables and Migrate for workflows
where a sequence table is being moved (#13238). This has an impact on the
Vitess upgrade process from an earlier version if you need to use such a workflow before the entire cluster is upgraded.

Any MoveTables or Migrate workflow that moves a sequence table should only be run after all vitess components have been
upgraded, and no upgrade should be done while such a workflow is in progress.

New Dry-run/monitoring-only mode for the transaction throttler

A new CLI flag --tx-throttler-dry-run to set the Transaction Throttler to monitoring-only/dry-run mode has been added.
If the transaction throttler is enabled with --enable-tx-throttler and the new dry-run flag is also specified, the
tablet will not actually throttle any transactions; however, it will increase the counters for transactions throttled
(vttablet_transaction_throttler_throttled). This allows users to deploy the transaction throttler in production and
gain observability on how much throttling would take place, without actually throttling any requests.

Two new built-in durability policies have been added in Vitess 18: `semi_sync_with_rdonly_ack` and `cross_cell_with_rdonly_ack`.
These policies are similar to `semi_sync` and `cross_cell` respectively, the only difference is that `rdonly` tablets can also send semi-sync ACKs.

The entire changelog for this release can be found here.

The release includes 420 merged Pull Requests.

Thanks to all our contributors: @GuptaManan100, @Juneezee, @L3o-pold, @adsr, @ajm188, @app/dependabot, @app/github-actions, @app/vitess-bot, @arvind-murty, @austenLacy, @brendar, @davidpiegza, @dbussink, @deepthi, @derekperkins, @ejortegau, @frouioui, @harshit-gangal, @hkdsun, @jfg956, @jspawar, @mattlord, @maxenglander, @mdlayher, @notfelineit, @olyazavr, @pbibra, @peterlyoo, @rafer, @rohit-nayak-ps, @shlomi-noach, @systay, @timvaillancourt, @vmg, @yields

Vitess v18.0.0

Release of Vitess v18.0.0

Summary

Table of Contents

Upgrade process

Two new built-in durability policies have been added in Vitess 18: semi_sync_with_rdonly_ack and cross_cell_with_rdonly_ack. These policies are similar to semi_sync and cross_cell respectively, the only difference is that rdonly tablets can also send semi-sync ACKs.

Contributors

Two new built-in durability policies have been added in Vitess 18: `semi_sync_with_rdonly_ack` and `cross_cell_with_rdonly_ack`.
These policies are similar to `semi_sync` and `cross_cell` respectively, the only difference is that `rdonly` tablets can also send semi-sync ACKs.