实验一番，下面将引导 test 数据库中 test 表，首先是准备几条测试用的数据
INSERT INTO `test` VALUES (1, 1, '1'); INSERT INTO `test` VALUES (2, 2, '2'); INSERT INTO `test` VALUES (3, 3, '3'); INSERT INTO `test` VALUES (4, 4, '4');然后 reset master; 清空binlog，删除 maxwell 库中的表。接着使用快速开始中的命令，启动Kafka、Maxwell和Kafka消费者，然后启动 maxwell-bootstrap
docker run -it --rm zendesk/maxwell bin/maxwell-bootstrap --user maxwell \ --password 123456 --host 10.100.97.246 --database test --table test --client_id maxwell注意：--bootstrapper=sync 时，在处理bootstrap时，会阻塞正常的binlog解析；--bootstrapper=async 时，不会阻塞。
也可以执行下面的SQL，在 maxwell.bootstrap 表中插入记录，手动触发
insert into maxwell.bootstrap (database_name, table_name) values ('test', 'test');就可以在 kafka 消费者端看见引导过来的数据了
{"database":"maxwell","table":"bootstrap","type":"insert","ts":1552199115,"xid":36738,"commit":true,"data":{"id":3,"database_name":"test","table_name":"test","where_clause":null,"is_complete":0,"inserted_rows":0,"total_rows":0,"created_at":null,"started_at":null,"completed_at":null,"binlog_file":null,"binlog_position":0,"client_id":"maxwell"}} {"database":"test","table":"test","type":"bootstrap-start","ts":1552199115,"data":{}} {"database":"test","table":"test","type":"bootstrap-insert","ts":1552199115,"data":{"id":1,"age":1,"name":"1"}} {"database":"test","table":"test","type":"bootstrap-insert","ts":1552199115,"data":{"id":2,"age":2,"name":"2"}} {"database":"test","table":"test","type":"bootstrap-insert","ts":1552199115,"data":{"id":3,"age":3,"name":"3"}} {"database":"test","table":"test","type":"bootstrap-insert","ts":1552199115,"data":{"id":4,"age":4,"name":"4"}} {"database":"maxwell","table":"bootstrap","type":"update","ts":1552199115,"xid":36756,"commit":true,"data":{"id":3,"database_name":"test","table_name":"test","where_clause":null,"is_complete":1,"inserted_rows":4,"total_rows":0,"created_at":null,"started_at":"2019-03-10 14:25:15","completed_at":"2019-03-10 14:25:15","binlog_file":"mysql-bin.000001","binlog_position":64446,"client_id":"maxwell"},"old":{"is_complete":0,"inserted_rows":1,"completed_at":null}} {"database":"test","table":"test","type":"bootstrap-complete","ts":1552199115,"data":{}}中间的4条便是 test.test 的binlog数据了，注意这里的 type 是 bootstrap-insert，而不是 insert。
然后再一次查看binlog，show binlog events;，会发现只有与 maxwell 相关的binlog，并没有 test.test 相关的binlog，所以 maxwell-bootstrap 命令并不会产生多余的 binlog，当数据表的数量很大时，这个好处会更加明显
Bootstrap 的过程是 bootstrap-start -> bootstrap-insert -> bootstrap-complete，其中，start和complete的data字段为空，不携带数据。
在进行bootstrap过程中，如果maxwell崩溃，重启时，bootstrap会完全重新开始，不管之前进行到多少，若不希望这样，可以到数据库中设置 is_complete 字段值为1(表示完成)，或者删除该行
Maxwell监控Maxwell 提供了 base logging mechanism, JMX, HTTP or by push to Datadog 这四种监控方式，与监控相关的配置项有下列这些：
选项 参数值 描述 默认值metrics_prefix STRING 指标的前缀 MaxwellMetrics
metrics_type [slf4j | jmx | http | datadog] 发布指标的方式
metrics_jvm BOOLEAN 是否收集JVM信息 false
metrics_slf4j_interval SECONDS 将指标记录到日志的频率，metrics_type须配置为slf4j 60
http_port INT metrics_type为http时，发布指标绑定的端口 8080
http_path_prefix STRING http的路径前缀 /
http_bind_address STRING http发布指标绑定的地址 all addresses
http_diagnostic BOOLEAN http是否开启diagnostic后缀 false
http_diagnostic_timeout MILLISECONDS http diagnostic 响应超时时间 10000
metrics_datadog_type [udp | http] metrics_type为datadog时发布指标的方式 udp
metrics_datadog_tags STRING 提供给 datadog 的标签，如 tag1:value1,tag2:value2
metrics_datadog_interval INT 推指标到datadog的频率，单位秒 60
metrics_datadog_apikey STRING 当 metrics_datadog_type=http 时datadog用的api key
metrics_datadog_host STRING 当metrics_datadog_type=udp时推指标的目标地址 localhost
metrics_datadog_port INT 当metrics_datadog_type=udp 时推指标的端口 8125