thevar1able/ClickHouse
1
0
Fork 0
mirror of https://github.com/ClickHouse/ClickHouse.git synced 2024-11-10 09:32:06 +00:00
Code Issues Packages Projects Releases Wiki Activity

Query Cache: Reject queries with non-deterministic functions by default

Browse Source 
https://github.com/ClickHouse/support-escalation/issues/963
This commit is contained in:
Robert Schulze 2023-09-19 16:29:02 +00:00
parent 1556cdd21a
commit a15029ecb5
No known key found for this signature in database
GPG Key ID: 26703B55FB13728A
5 changed files with 15 additions and 11 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
docs/en/operations/query-cache.md
View File

@ -142,7 +142,7 @@ As a result, the query cache stores for each query multiple (partial)
result blocks. While this behavior is a good default, it can be suppressed using setting
[query_cache_squash_partial_results](settings/settings.md#query-cache-squash-partial-results).
Also, results of queries with non-deterministic functions are not cached. Such functions include
Also, results of queries with non-deterministic functions are not cached by default. Such functions include
- functions for accessing dictionaries: [`dictGet()`](../sql-reference/functions/ext-dict-functions.md#dictGet) etc.
- [user-defined functions](../sql-reference/statements/create/function.md),
- functions which return the current date or time: [`now()`](../sql-reference/functions/date-time-functions.md#now),
@ -158,7 +158,7 @@ Also, results of queries with non-deterministic functions are not cached. Such f
- functions which depend on the environment: [`currentUser()`](../sql-reference/functions/other-functions.md#currentUser),
[`queryID()`](../sql-reference/functions/other-functions.md#queryID),
[`getMacro()`](../sql-reference/functions/other-functions.md#getMacro) etc.
Caching of non-deterministic functions can be forced regardless using setting
To force caching of results of queries with non-deterministic functionsregardless, using setting
[query_cache_store_results_of_queries_with_nondeterministic_functions](settings/settings.md#query-cache-store-results-of-queries-with-nondeterministic-functions).
Finally, entries in the query cache are not shared between users due to security reasons. For example, user A must not be able to bypass a

1
src/Common/ErrorCodes.cpp
View File

@ -585,6 +585,7 @@
M(700, USER_SESSION_LIMIT_EXCEEDED) \
M(701, CLUSTER_DOESNT_EXIST) \
M(702, CLIENT_INFO_DOES_NOT_MATCH) \
M(703, CANNOT_USE_QUERY_CACHE_WITH_NONDETERMINISTIC_FUNCTIONS) \
\
M(999, KEEPER_EXCEPTION) \
M(1000, POCO_EXCEPTION) \

14
src/Interpreters/executeQuery.cpp
View File

@ -94,11 +94,12 @@ namespace DB
namespace ErrorCodes
{
extern const int CANNOT_USE_QUERY_CACHE_WITH_NONDETERMINISTIC_FUNCTIONS;
extern const int INTO_OUTFILE_NOT_ALLOWED;
extern const int QUERY_WAS_CANCELLED;
extern const int INVALID_TRANSACTION;
extern const int LOGICAL_ERROR;
extern const int NOT_IMPLEMENTED;
extern const int QUERY_WAS_CANCELLED;
}
@ -991,7 +992,7 @@ static std::tuple<ASTPtr, BlockIO> executeQueryImpl(
if (!async_insert)
{
/// If it is a non-internal SELECT, and passive/read use of the query cache is enabled, and the cache knows the query, then set
/// If it is a non-internal SELECT, and passive (read) use of the query cache is enabled, and the cache knows the query, then set
/// a pipeline with a source populated by the query cache.
auto get_result_from_query_cache = [&]()
{
@ -1091,11 +1092,14 @@ static std::tuple<ASTPtr, BlockIO> executeQueryImpl(
res = interpreter->execute();
/// If it is a non-internal SELECT query, and active/write use of the query cache is enabled, then add a processor on
/// If it is a non-internal SELECT query, and active (write) use of the query cache is enabled, then add a processor on
/// top of the pipeline which stores the result in the query cache.
if (can_use_query_cache && settings.enable_writes_to_query_cache
&& (!astContainsNonDeterministicFunctions(ast, context) || settings.query_cache_store_results_of_queries_with_nondeterministic_functions))
if (can_use_query_cache && settings.enable_writes_to_query_cache)
{
if (astContainsNonDeterministicFunctions(ast, context) && !settings.query_cache_store_results_of_queries_with_nondeterministic_functions)
throw Exception(ErrorCodes::CANNOT_USE_QUERY_CACHE_WITH_NONDETERMINISTIC_FUNCTIONS,
"Unable to cache the query result because the query contains a non-deterministic function. Use setting query_cache_store_results_of_queries_with_nondeterministic_functions = 1 to store the query result regardless.");
QueryCache::Key key(
ast, res.pipeline.getHeader(),
context->getUserName(), settings.query_cache_share_between_users,

1
tests/queries/0_stateless/02494_query_cache_nondeterministic_functions.reference
View File

@ -1,4 +1,3 @@
1
0
---
1

6
tests/queries/0_stateless/02494_query_cache_nondeterministic_functions.sql
View File

@ -3,13 +3,13 @@
SYSTEM DROP QUERY CACHE;
-- rand() is non-deterministic, with default settings no entry in the query cache should be created
SELECT COUNT(rand(1)) SETTINGS use_query_cache = true;
-- rand() is non-deterministic, the query is rejected by default
SELECT COUNT(rand(1)) SETTINGS use_query_cache = true; -- { serverError 703 }
SELECT COUNT(*) FROM system.query_cache;
SELECT '---';
-- But an entry can be forced using a setting
-- Force caching using a setting
SELECT COUNT(RAND(1)) SETTINGS use_query_cache = true, query_cache_store_results_of_queries_with_nondeterministic_functions = true;
SELECT COUNT(*) FROM system.query_cache;