/clickhouse-bulk

Collects many small inserts to ClickHouse and send in big inserts

Primary LanguageGoApache License 2.0Apache-2.0

ClickHouse-Bulk

build codecov download binaries Go Report Card godoc

Simple Yandex ClickHouse insert collector. It collect requests and send to ClickHouse servers.

Installation

Download binary for you platorm

or

Use docker image

or from sources (Go 1.13+):

git clone https://github.com/nikepan/clickhouse-bulk
cd clickhouse-bulk
go build

Features

  • Group n requests and send to any of ClickHouse server
  • Sending collected data by interval
  • Tested with VALUES, TabSeparated formats
  • Supports many servers to send
  • Supports query in query parameters and in body
  • Supports other query parameters like username, password, database
  • Supports basic authentication

For example:

INSERT INTO table3 (c1, c2, c3) VALUES ('v1', 'v2', 'v3')
INSERT INTO table3 (c1, c2, c3) VALUES ('v4', 'v5', 'v6')

sends as

INSERT INTO table3 (c1, c2, c3) VALUES ('v1', 'v2', 'v3')('v4', 'v5', 'v6')

Options

  • -config - config file (json); default config.json

Configuration file

{
  "listen": ":8124",
  "flush_count": 10000, // check by \n char
  "flush_interval": 1000, // milliseconds
  "clean_interval": 0, // how often cleanup internal tables - e.g. inserts to different temporary tables, or as workaround for query_id etc. milliseconds
  "remove_query_id": true, // some drivers sends query_id which prevents inserts to be batched
  "dump_check_interval": 300, // interval for try to send dumps (seconds); -1 to disable
  "debug": false, // log incoming requests
  "dump_dir": "dumps", // directory for dump unsended data (if clickhouse errors)
  "clickhouse": {
    "down_timeout": 60, // wait if server in down (seconds)
    "connect_timeout": 10, // wait for server connect (seconds)
    "tls_server_name": "", // override TLS serverName for certificate verification (e.g. in cases you share same "cluster" certificate across multiple nodes)
    "insecure_tls_skip_verify": false, // INSECURE - skip certificate verification at all
    "servers": [
      "http://127.0.0.1:8123"
    ]
  },
  "metrics_prefix": "prefix"
}

Environment variables (used for docker image)

  • CLICKHOUSE_BULK_DEBUG - enable debug logging
  • CLICKHOUSE_SERVERS - comma separated list of servers
  • CLICKHOUSE_FLUSH_COUNT - count of rows for insert
  • CLICKHOUSE_FLUSH_INTERVAL - insert interval
  • CLICKHOUSE_CLEAN_INTERVAL - internal tables clean interval
  • DUMP_CHECK_INTERVAL - interval of resend dumps
  • CLICKHOUSE_DOWN_TIMEOUT - wait time if server is down
  • CLICKHOUSE_CONNECT_TIMEOUT - clickhouse server connect timeout
  • CLICKHOUSE_TLS_SERVER_NAME - server name for TLS certificate verification
  • CLICKHOUSE_INSECURE_TLS_SKIP_VERIFY - skip certificate verification at all
  • METRICS_PREFIX - prefix for prometheus metrics

Quickstart

./clickhouse-bulk and send queries to :8124

Metrics

manual check main metrics curl -s http://127.0.0.1:8124/metrics | grep "^ch_"

  • ch_bad_servers 0 - actual count of bad servers
  • ch_dump_count 0 - dumps saved from launch
  • ch_queued_dumps 0 - actual dump files id directory
  • ch_good_servers 1 - actual good servers count
  • ch_received_count 40 - received requests count from launch
  • ch_sent_count 1 - sent request count from launch

Tips

For better performance words FORMAT and VALUES must be uppercase.