mirror of
https://github.com/ClickHouse/ClickHouse.git
synced 2026-10-28 16:18:23 +08:00
Michael KolupaevandClaude Fable 5.1
7b210c6b7f
The previous commit added `TopChildrenTracker` and the `--top-children` option but did not wire them into `analyzeSnapshot`, so the build failed with `-Wmissing-prototypes` (the definition did not match the declaration) and clang-tidy reported the tracker as unused: https://s3.amazonaws.com/clickhouse-test-reports/praktika.html?PR=106270&sha=d74d5b7fedcbac408f79a2a316b08d673d2406f7&name_0=PR&name_1=Fast%20test https://s3.amazonaws.com/clickhouse-test-reports/praktika.html?PR=106270&sha=d74d5b7fedcbac408f79a2a316b08d673d2406f7&name_0=PR&name_1=Build%20%28arm_tidy%29 PR: https://github.com/ClickHouse/ClickHouse/pull/106270 Also address review comments: - A failure to analyze a single snapshot file now propagates, so the command exits with a non-zero code instead of just printing to stderr. In directory mode the remaining snapshots are still analyzed, but an exception is thrown at the end if any of them failed. - Snapshots in a directory are sorted by the log index parsed from the file name (`getLogIdxFromSnapshotPath`) instead of lexicographically, so `snapshot_10.bin` really comes before `snapshot_9.bin`. - Whether `--snapshot-path` is a file or a directory is decided by `is_directory` instead of the file extension. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
29 lines
924 B
C++
29 lines
924 B
C++
#pragma once
|
|
|
|
#include <cstddef>
|
|
#include <string>
|
|
|
|
namespace DB
|
|
{
|
|
|
|
/// Analyze a Keeper snapshot file (or every snapshot in a directory) by streaming it
|
|
/// from disk with O(1) memory and printing summary information and statistics.
|
|
///
|
|
/// If `with_node_stats` is set, the set of all paths is additionally kept in memory
|
|
/// (O(number of nodes)) to compute the biggest subtrees - this is the only part that
|
|
/// is not O(1) memory.
|
|
///
|
|
/// `sample_size` random nodes and the `top_children_limit` nodes with the most children are
|
|
/// printed too (0 disables).
|
|
///
|
|
/// Throws if `snapshot_path` is a file that can't be analyzed, or a directory in which some
|
|
/// snapshot can't be analyzed (the remaining snapshots in the directory are still analyzed).
|
|
void analyzeSnapshot(
|
|
const std::string & snapshot_path,
|
|
bool with_node_stats,
|
|
size_t subtrees_limit,
|
|
size_t sample_size,
|
|
size_t top_children_limit);
|
|
|
|
}
|