Files
scylladb/cql3/statements/create_view_statement.cc
Asias He a8ad385ecd repair: Get rid of the gc_grace_seconds
The gc_grace_seconds is a very fragile and broken design inherited from
Cassandra. Deleted data can be resurrected if cluster wide repair is not
performed within gc_grace_seconds. This design pushes the job of making
the database consistency to the user. In practice, it is very hard to
guarantee repair is performed within gc_grace_seconds all the time. For
example, repair workload has the lowest priority in the system which can
be slowed down by the higher priority workload, so that there is no
guarantee when a repair can finish. A gc_grace_seconds value that is
used to work might not work after data volume grows in a cluster. Users
might want to avoid running repair during a specific period where
latency is the top priority for their business.

To solve this problem, an automatic mechanism to protect data
resurrection is proposed and implemented. The main idea is to remove the
tombstone only after the range that covers the tombstone is repaired.

In this patch, a new table option tombstone_gc is added. The option is
used to configure tombstone gc mode. For example:

1) GC a tombstone after gc_grace_seconds

cqlsh> ALTER TABLE ks.cf WITH tombstone_gc = {'mode':'timeout'} ;

This is the default mode. If no tombstone_gc option is specified by the
user. The old gc_grace_seconds based gc will be used.

2) Never GC a tombstone

cqlsh> ALTER TABLE ks.cf WITH tombstone_gc = {'mode':'disabled'};

3) GC a tombstone immediately

cqlsh> ALTER TABLE ks.cf WITH tombstone_gc = {'mode':'immediate'};

4) GC a tombstone after repair

cqlsh> ALTER TABLE ks.cf WITH tombstone_gc = {'mode':'repair'};

In addition to the 'mode' option, another option 'propagation_delay_in_seconds'
is added. It defines the max time a write could possibly delay before it
eventually arrives at a node.

A new gossip feature TOMBSTONE_GC_OPTIONS is added. The new tombstone_gc
option can only be used after the whole cluster supports the new
feature. A mixed cluster works with no problem.

Tests: compaction_test.py, ninja test

Fixes #3560

[avi: resolve conflicts vs data_dictionary]
2022-01-04 19:48:14 +02:00

385 lines
18 KiB
C++

/*
* Licensed to the Apache Software Foundation (ASF) under one
* or more contributor license agreements. See the NOTICE file
* distributed with this work for additional information
* regarding copyright ownership. The ASF licenses this file
* to you under the Apache License, Version 2.0 (the
* "License"); you may not use this file except in compliance
* with the License. You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
/*
* Copyright (C) 2016-present ScyllaDB
*
* Modified by ScyllaDB
*/
/*
* This file is part of Scylla.
*
* Scylla is free software: you can redistribute it and/or modify
* it under the terms of the GNU Affero General Public License as published by
* the Free Software Foundation, either version 3 of the License, or
* (at your option) any later version.
*
* Scylla is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with Scylla. If not, see <http://www.gnu.org/licenses/>.
*/
#include <unordered_set>
#include <vector>
#include <boost/range/iterator_range.hpp>
#include <boost/range/join.hpp>
#include <boost/range/adaptor/map.hpp>
#include <boost/range/adaptor/transformed.hpp>
#include <seastar/core/coroutine.hh>
#include "cql3/column_identifier.hh"
#include "cql3/restrictions/statement_restrictions.hh"
#include "cql3/statements/create_view_statement.hh"
#include "cql3/statements/prepared_statement.hh"
#include "cql3/statements/select_statement.hh"
#include "cql3/statements/raw/select_statement.hh"
#include "cql3/selection/selectable.hh"
#include "cql3/selection/selectable_with_field_selection.hh"
#include "cql3/selection/selection.hh"
#include "cql3/selection/writetime_or_ttl.hh"
#include "cql3/query_processor.hh"
#include "cql3/util.hh"
#include "schema_builder.hh"
#include "service/storage_proxy.hh"
#include "validation.hh"
#include "db/extensions.hh"
#include "data_dictionary/data_dictionary.hh"
#include "gms/feature_service.hh"
#include "db/view/view.hh"
#include "service/migration_manager.hh"
namespace cql3 {
namespace statements {
create_view_statement::create_view_statement(
cf_name view_name,
cf_name base_name,
std::vector<::shared_ptr<selection::raw_selector>> select_clause,
std::vector<::shared_ptr<relation>> where_clause,
std::vector<::shared_ptr<cql3::column_identifier::raw>> partition_keys,
std::vector<::shared_ptr<cql3::column_identifier::raw>> clustering_keys,
bool if_not_exists)
: schema_altering_statement{view_name}
, _base_name{base_name}
, _select_clause{select_clause}
, _where_clause{where_clause}
, _partition_keys{partition_keys}
, _clustering_keys{clustering_keys}
, _if_not_exists{if_not_exists}
{
}
future<> create_view_statement::check_access(query_processor& qp, const service::client_state& state) const {
return state.has_column_family_access(qp.proxy().local_db(), keyspace(), _base_name.get_column_family(), auth::permission::ALTER);
}
void create_view_statement::validate(query_processor& qp, const service::client_state& state) const {
}
static const column_definition* get_column_definition(const schema& schema, column_identifier::raw& identifier) {
auto prepared = identifier.prepare(schema);
assert(dynamic_pointer_cast<column_identifier>(prepared));
auto id = static_pointer_cast<column_identifier>(prepared);
return schema.get_column_definition(id->name());
}
static bool validate_primary_key(
const schema& schema,
const column_definition* def,
const std::unordered_set<const column_definition*>& base_pk,
bool has_non_pk_column,
const restrictions::statement_restrictions& restrictions) {
if (def->type->is_multi_cell()) {
throw exceptions::invalid_request_exception(format("Cannot use MultiCell column '{}' in PRIMARY KEY of materialized view", def->name_as_text()));
}
if (def->type->references_duration()) {
throw exceptions::invalid_request_exception(format("Cannot use Duration column '{}' in PRIMARY KEY of materialized view", def->name_as_text()));
}
if (def->is_static()) {
throw exceptions::invalid_request_exception(format("Cannot use Static column '{}' in PRIMARY KEY of materialized view", def->name_as_text()));
}
bool new_non_pk_column = false;
if (!base_pk.contains(def)) {
if (has_non_pk_column) {
throw exceptions::invalid_request_exception(format("Cannot include more than one non-primary key column '{}' in materialized view primary key", def->name_as_text()));
}
new_non_pk_column = true;
}
// We don't need to include the "IS NOT NULL" filter on a non-composite partition key
// because we will never allow a single partition key to be NULL
bool is_non_composite_partition_key = def->is_partition_key() &&
schema.partition_key_columns().size() == 1;
if (!is_non_composite_partition_key && !restrictions.is_restricted(def)) {
throw exceptions::invalid_request_exception(format("Primary key column '{}' is required to be filtered by 'IS NOT NULL'", def->name_as_text()));
}
return new_non_pk_column;
}
view_ptr create_view_statement::prepare_view(data_dictionary::database db) const {
// We need to make sure that:
// - primary key includes all columns in base table's primary key
// - make sure that the select statement does not have anything other than columns
// and their names match the base table's names
// - make sure that primary key does not include any collections
// - make sure there is no where clause in the select statement
// - make sure there is not currently a table or view
// - make sure base_table gc_grace_seconds > 0
auto schema_extensions = _properties.properties()->make_schema_extensions(db.extensions());
_properties.validate(db, keyspace(), schema_extensions);
if (_properties.use_compact_storage()) {
throw exceptions::invalid_request_exception(format("Cannot use 'COMPACT STORAGE' when defining a materialized view"));
}
if (_properties.properties()->get_cdc_options(schema_extensions)) {
throw exceptions::invalid_request_exception("Cannot enable CDC for a materialized view");
}
// View and base tables must be in the same keyspace, to ensure that RF
// is the same (because we assign a view replica to each base replica).
// If a keyspace was not specified for the base table name, it is assumed
// it is in the same keyspace as the view table being created (which
// itself might be the current USEd keyspace, or explicitly specified).
if (!_base_name.has_keyspace()) {
_base_name.set_keyspace(keyspace(), true);
}
if (_base_name.get_keyspace() != keyspace()) {
throw exceptions::invalid_request_exception(format("Cannot create a materialized view on a table in a separate keyspace ('{}' != '{}')",
_base_name.get_keyspace(), keyspace()));
}
schema_ptr schema = validation::validate_column_family(db.real_database(), _base_name.get_keyspace(), _base_name.get_column_family());
if (schema->is_counter()) {
throw exceptions::invalid_request_exception(format("Materialized views are not supported on counter tables"));
}
if (schema->is_view()) {
throw exceptions::invalid_request_exception(format("Materialized views cannot be created against other materialized views"));
}
if (db.get_cdc_base_table(*schema)) {
throw exceptions::invalid_request_exception(format("Materialized views cannot be created on CDC Log tables"));
}
if (schema->gc_grace_seconds().count() == 0) {
throw exceptions::invalid_request_exception(fmt::format(
"Cannot create materialized view '{}' for base table "
"'{}' with gc_grace_seconds of 0, since this value is "
"used to TTL undelivered updates. Setting gc_grace_seconds "
"too low might cause undelivered updates to expire "
"before being replayed.", column_family(), _base_name.get_column_family()));
}
// Gather all included columns, as specified by the select clause
auto included = boost::copy_range<std::unordered_set<const column_definition*>>(_select_clause | boost::adaptors::transformed([&](auto&& selector) {
if (selector->alias) {
throw exceptions::invalid_request_exception(format("Cannot use alias when defining a materialized view"));
}
auto& selectable = selector->selectable_;
shared_ptr<column_identifier::raw> identifier;
expr::visit(overloaded_functor{
[&] (const expr::unresolved_identifier& ui) { identifier = ui.ident; },
[] (const auto& default_case) -> void { throw exceptions::invalid_request_exception(format("Cannot use general expressions when defining a materialized view")); },
}, selectable);
auto* def = get_column_definition(*schema, *identifier);
if (!def) {
throw exceptions::invalid_request_exception(format("Unknown column name detected in CREATE MATERIALIZED VIEW statement: {}", identifier));
}
return def;
}));
auto parameters = make_lw_shared<raw::select_statement::parameters>(raw::select_statement::parameters::orderings_type(), false, true);
raw::select_statement raw_select(_base_name, std::move(parameters), _select_clause, _where_clause, std::nullopt, std::nullopt, {}, std::make_unique<cql3::attributes::raw>());
raw_select.prepare_keyspace(keyspace());
raw_select.set_bound_variables({});
cql_stats ignored;
auto prepared = raw_select.prepare(db, ignored, true);
auto restrictions = static_pointer_cast<statements::select_statement>(prepared->statement)->get_restrictions();
auto base_primary_key_cols = boost::copy_range<std::unordered_set<const column_definition*>>(
boost::range::join(schema->partition_key_columns(), schema->clustering_key_columns())
| boost::adaptors::transformed([](auto&& def) { return &def; }));
// Validate the primary key clause, ensuring only one non-PK base column is used in the view's PK.
bool has_non_pk_column = false;
std::unordered_set<const column_definition*> target_primary_keys;
std::vector<const column_definition*> target_partition_keys;
std::vector<const column_definition*> target_clustering_keys;
auto validate_pk = [&] (const std::vector<::shared_ptr<cql3::column_identifier::raw>>& keys, std::vector<const column_definition*>& target_keys) mutable {
for (auto&& identifier : keys) {
auto* def = get_column_definition(*schema, *identifier);
if (!def) {
throw exceptions::invalid_request_exception(format("Unknown column name detected in CREATE MATERIALIZED VIEW statement: {}", identifier));
}
if (!target_primary_keys.insert(def).second) {
throw exceptions::invalid_request_exception(format("Duplicate entry found in PRIMARY KEY: {}", identifier));
}
target_keys.push_back(def);
has_non_pk_column |= validate_primary_key(*schema, def, base_primary_key_cols, has_non_pk_column, *restrictions);
}
};
validate_pk(_partition_keys, target_partition_keys);
validate_pk(_clustering_keys, target_clustering_keys);
std::vector<const column_definition*> missing_pk_columns;
std::vector<const column_definition*> target_non_pk_columns;
std::vector<const column_definition*> unselected_columns;
// We need to include all of the primary key columns from the base table in order to make sure that we do not
// overwrite values in the view. We cannot support "collapsing" the base table into a smaller number of rows in
// the view because if we need to generate a tombstone, we have no way of knowing which value is currently being
// used in the view and whether or not to generate a tombstone. In order to not surprise our users, we require
// that they include all of the columns. We provide them with a list of all of the columns left to include.
for (auto& def : schema->all_columns()) {
bool included_def = included.empty() || included.contains(&def);
if (included_def && def.is_static()) {
throw exceptions::invalid_request_exception(format("Unable to include static column '{}' which would be included by Materialized View SELECT * statement", def));
}
bool def_in_target_pk = target_primary_keys.contains(&def);
if (included_def && !def_in_target_pk) {
target_non_pk_columns.push_back(&def);
}
if (!included_def && !def_in_target_pk && !def.is_static()) {
unselected_columns.push_back(&def);
}
if (def.is_primary_key() && !def_in_target_pk) {
missing_pk_columns.push_back(&def);
}
}
if (!missing_pk_columns.empty()) {
auto column_names = ::join(", ", missing_pk_columns | boost::adaptors::transformed(std::mem_fn(&column_definition::name_as_text)));
throw exceptions::invalid_request_exception(format("Cannot create Materialized View {} without primary key columns from base {} ({})",
column_family(), _base_name.get_column_family(), column_names));
}
if (_partition_keys.empty()) {
throw exceptions::invalid_request_exception(format("Must select at least a column for a Materialized View"));
}
// The unique feature of a filter by a non-key column is that the
// value of such column can be updated - and also be expired with TTL
// and cause the view row to appear and disappear. We don't currently
// support support this case - see issue #3430, and neither does
// Cassandra - see see CASSANDRA-13798 and CASSANDRA-13832.
// Actually, as CASSANDRA-13798 explains, the problem is "the liveness of
// view row is now depending on multiple base columns (multiple filtered
// non-pk base column + base column used in view pk)". When the filtered
// column *is* the base column added to the view pk, we don't have this
// problem. And this case actually works correctly.
auto non_pk_restrictions = restrictions->get_non_pk_restriction();
if (non_pk_restrictions.size() == 1 && has_non_pk_column &&
target_primary_keys.contains(non_pk_restrictions.cbegin()->first)) {
// This case (filter by new PK column of the view) works, as explained above
} else if (!non_pk_restrictions.empty()) {
auto column_names = ::join(", ", non_pk_restrictions | boost::adaptors::map_keys | boost::adaptors::transformed(std::mem_fn(&column_definition::name_as_text)));
throw exceptions::invalid_request_exception(format("Non-primary key columns cannot be restricted in the SELECT statement used for materialized view {} creation (got restrictions on: {})",
column_family(), column_names));
}
schema_builder builder{keyspace(), column_family()};
auto add_columns = [this, &builder] (std::vector<const column_definition*>& defs, column_kind kind) mutable {
for (auto* def : defs) {
auto&& type = _properties.get_reversable_type(*def->column_specification->name, def->type);
builder.with_column(def->name(), type, kind);
}
};
add_columns(target_partition_keys, column_kind::partition_key);
add_columns(target_clustering_keys, column_kind::clustering_key);
add_columns(target_non_pk_columns, column_kind::regular_column);
// Add all unselected columns (base-table columns which are not selected
// in the view) as "virtual columns" - columns which have timestamp and
// ttl information, but an empty value. These are needed to keep view
// rows alive when the base row is alive, even if the view row has no
// data, just a key (see issue #3362). The virtual columns are not needed
// when the view pk adds a regular base column (i.e., has_non_pk_column)
// because in that case, the liveness of that base column is what
// determines the liveness of the view row.
if (!has_non_pk_column) {
for (auto* def : unselected_columns) {
db::view::create_virtual_column(builder, def->name(), def->type);
}
}
_properties.properties()->apply_to_builder(builder, std::move(schema_extensions));
if (builder.default_time_to_live().count() > 0) {
throw exceptions::invalid_request_exception(
"Cannot set or alter default_time_to_live for a materialized view. "
"Data in a materialized view always expire at the same time than "
"the corresponding data in the parent table.");
}
auto where_clause_text = util::relations_to_where_clause(_where_clause);
builder.with_view_info(schema->id(), schema->cf_name(), included.empty(), std::move(where_clause_text));
return view_ptr(builder.build());
}
future<std::pair<::shared_ptr<cql_transport::event::schema_change>, std::vector<mutation>>>
create_view_statement::prepare_schema_mutations(query_processor& qp) const {
::shared_ptr<cql_transport::event::schema_change> ret;
std::vector<mutation> m;
auto definition = prepare_view(qp.db());
try {
m = co_await qp.get_migration_manager().prepare_new_view_announcement(std::move(definition));
using namespace cql_transport;
ret = ::make_shared<event::schema_change>(
event::schema_change::change_type::CREATED,
event::schema_change::target_type::TABLE,
keyspace(),
column_family());
} catch (const exceptions::already_exists_exception& e) {
if (!_if_not_exists) {
co_return coroutine::exception(std::current_exception());
}
}
co_return std::make_pair(std::move(ret), std::move(m));
}
std::unique_ptr<cql3::statements::prepared_statement>
create_view_statement::prepare(data_dictionary::database db, cql_stats& stats) {
if (!_prepare_ctx.get_variable_specifications().empty()) {
throw exceptions::invalid_request_exception(format("Cannot use query parameters in CREATE MATERIALIZED VIEW statements"));
}
return std::make_unique<prepared_statement>(make_shared<create_view_statement>(*this));
}
}
}