Processing many small inserts over the native protocol is suboptimal because
the server replies with the table structure and column defaults for every query.
This adds:
- A setting `send_table_structure_on_insert_with_inline_data` (default true). When disabled,
the server will parse inline data in INSERT queries itself instead of sending
table structure back to the client.
- A `--inline-insert-data` command-line option for `clickhouse-client`. When set,
the client sends the data as is in the INSERT query text instead of converting
it to blocks in the native format, and routes the query through the ordinary
query path (no round-trip for table structure).
The server already has the machinery to process inline INSERT data (used by the
HTTP interface and async inserts). This change reuses it for the native protocol.
https://github.com/ClickHouse/ClickHouse/issues/91630
Co-Authored-By: Claude Opus 4.6 (1M context) <[email protected]>
Add named `Throttler` instances with LOG_TRACE logging to diagnose
throttling issues (follow-up to #95127). The `Throttler` class now
takes a `std::string_view` name as its first constructor parameter.
Logging covers all state mutations:
- Sleeping: amount, count, tokens, max_speed, block_ns, max_block_ns, sleep_ns
- Not sleeping (when throttler is active): amount, count, tokens, max_speed
- `reset`: count, tokens, max_burst before clearing
- `setMaxSpeed`: old/new speed, tokens
Fix the flaky test by removing the `native_copy` case — with
`allow_s3_native_copy=1` S3 performs a server-side copy and ClickHouse
never reads/writes the data, so throttling cannot apply and wall clock
time is unpredictable.
https://s3.amazonaws.com/clickhouse-test-reports/json.html?PR=95118&sha=ab1adaa5d88f38a4934fac9fc8ffb91e97ef5a6f&name_0=PR&name_1=Stateless%20tests%20%28arm_binary%2C%20parallel%29
Co-Authored-By: Claude Opus 4.6 (1M context) <[email protected]>
- Remove `CurrentThread.h` from high-fan-out headers: `ProcessList.h`,
`IQueryPlanStep.h`, `DDLWorker.h`, `TCPHandler.h`, `AggregatingTransform.h`,
`ParallelParsingInputFormat.h`, `StorageTableFunction.h`, `parseSyslogLevel.h`,
`EmbeddedRocksDBBulkSink.h`, `BlockIO.h`, `LocalConnection.h`,
`StorageMaterializedView.h`, `HTTPHandler.h`
- Extract `QueryScope` from `CurrentThread` into standalone `Common/QueryScope.h`
with its own `.cpp`, replacing the nested `CurrentThread::QueryScope` class
- Extract `getCurrentThreadGroup` free function into `Common/ThreadGroupSwitcher.h`,
replacing `CurrentThread::getGroup()` in template headers:
`threadPoolCallbackRunner.h`, `ParallelFormattingOutputFormat.h`,
`HashedDictionaryParallelLoader.h`, `UniqExactSet.h`, `HashedDictionary.h`,
`StatusRequestsPool.h`
- De-inline `MemoryTrackerSwitcher` ctor/dtor into `MemoryTrackerSwitcher.cpp`
- De-inline `ResourceGuard::Request::wait` into `ResourceGuard.cpp`
- De-inline `ManyAggregatedData` destructor into `AggregatingTransform.cpp`
- De-inline `QueryStatus::updateProgressIn/Out` into `ProcessList.cpp`
- Add explicit `CurrentThread.h` includes to `.cpp` files that relied on
transitive includes through the trimmed headers
Co-Authored-By: Claude Opus 4.6 <[email protected]>
clang-tidy `bugprone-empty-catch` fires on catch blocks with empty
bodies regardless of the exception type. Re-add the NOLINT annotation
on the catch lines that were previously suppressed with `catch (...)`
and now use specific exception types.
Co-Authored-By: Claude Opus 4.6 <[email protected]>
Replace blanket `catch (...)` with specific exception types based on
analysis of what each try block can actually throw:
- `const Exception &` for catch blocks where only ClickHouse operations
are involved (Field operations, expression evaluation, parsing, etc.)
- `const std::exception &` for catch blocks involving standard library,
Poco, or third-party code (filesystem, HTTP, thread creation, etc.)
Blocks that genuinely need `catch (...)` are left unchanged:
- try-pattern functions (`tryGetPosition`, `tryDeserializeText`, etc.)
- Exception formatting utilities in `Exception.cpp`
- Signal/terminate handlers
- FFI callbacks (DeltaLake)
- Dual-handler patterns (already have specific + catch-all)
- `noexcept` functions with user-supplied callbacks
Co-Authored-By: Claude Opus 4.6 <[email protected]>
Enhance the `catch_all` style check in two ways:
1. Recognize more legitimate exception-handling patterns as "handled":
`getCurrentExceptionMessage`, `getCurrentExceptionCode`,
`getCurrentExceptionMessageAndPattern`, `ExecutionStatus::fromCurrentException`,
`onBackgroundException`, `storeException`, `STDERR_FILENO`, `writeRetry`,
`handle_exception`, `handleException`, `finishWithException`.
2. Fix function-try-block detection for destructors like
`Exception::~Exception() try { } catch (...) { }` — the enclosing-scope
scan now falls through to the function-try-block scan when it hits a
namespace/class scope (no parentheses in signature).
For the remaining 108 catch-all blocks, add `// Ok: <reason>` comments
explaining why each is intentional. Categories include:
- Try-patterns (`tryDeserializeText`, `tryGetPosition`, etc.)
- Exception handling/reporting utilities (`Exception.cpp`)
- Non-essential system probes (`Server.cpp` Linux /proc and /sys checks)
- C callbacks where exceptions cannot propagate (zip archives)
- Background thread exception saving (`onBackgroundException`)
- Fallback/retry logic (`DatabaseOverlay`, `ReadSchemaUtils`)
- Best-effort cleanup or error reporting
- Tests that expect exceptions
Replace `// NOLINT(bugprone-empty-catch)` with `// Ok:` comments on all
catch blocks that were flagged by the new check. Fix Allman-style brace
violations in `convertFieldToType.cpp` and `ClientBase.cpp`.
Co-Authored-By: Claude Opus 4.6 <[email protected]>
Adapted the `ssh-key-file` check to the new retry loop structure:
added `!config().has("ssh-key-file")` to the `should_ask_password` condition.
Co-Authored-By: Claude Opus 4.6 <[email protected]>
- Wrap long lines in `getHelpHeader` and `getHelpFooter` for `clickhouse client`
and `clickhouse local` to keep lines within reasonable width,
fixing the 00910_client_window_size_detection test
- Use `createOptionsDescription` instead of raw `po::options_description` constructor
in `git-import` and `static-files-disk-uploader` to handle zero terminal width
properly, fixing the `git-import --help` abort in debug builds
(03823_clickhouse_help_return_codes test)
Co-Authored-By: Claude Opus 4.6 <[email protected]>
This commit improves the help output for all ClickHouse applications:
- Add `clickhouse help` command that lists all available subcommands
- Return exit code 0 for all help messages (previously some returned 1)
- Output help messages to stdout instead of stderr
- Use "clickhouse <cmd>" format instead of argv[0] in help messages
- Add dynamic `app_name` detection for `clickhouse client` and `clickhouse local`
to show correct invocation name based on how the binary was called
- Add `--no-sudo` option for `clickhouse start` and `clickhouse restart`
for environments without sudo (e.g., Docker containers)
- Add detailed help headers and footers for client and local with usage examples
Closes#58244
Co-Authored-By: Claude Opus 4.5 <[email protected]>
When connecting to Clickhouse Cloud (or any similar proxied
setup) over a tunnel it is required to correctly specify SNI
on the TLS connection. This allows overriding SNI if the DNS
is something else.
Signed-off-by: Matt Klein <[email protected]>
Simply checking for `~/` should be enough, since we do not support
anything except expanding current's user `$HOME` (i.e. we do not support
`~another_user` syntax).
Alexey Milovidov