zhaown

> @chaozwn can you show your task json/sql? { "job": { "content": [ { "reader": { "name": "mysqlreader", "parameter": { "column": [ { "name": "id", "type": "int" }, { "name":...

> connection.isValid(5) 有些jdbc实现类不支持isValid，放在jdbc模块会导致具体的插件实现类报错 sqlserver、DB2驱动包不支持isValid() ok，我想加入一个connection.select 1用来valid connection useable.

> 怎么会被出现这种情况呢，com.dtstack.flinkx.connector.jdbc.sink.JdbcOutputFormat#writeMultipleRecordsInternal，这是jdbc mini batch写入的入口，如果这个位置判断connection失效，是不会有未提交的事务。 kafka -> writeMultipleRecordsInternal -> 写入成功 -> clearbatch -> 出错 -> writeSingleRecordInternal 所以我想不通为啥会有你说的部分数据丢失。

> 如果事务开启了checkPoint 且EXACTLY_ONCE的话，connection 事务是手动提交，只有checkpoint的时候才会提交事务 现在exactly_once提交的时候如果connection是失效的场合会自动rollback，然后重新初始化。

你是es什么版本

> This seems to work well and doesn't require additional dependencies: > > ```yaml > os: > edit: 'nvim --server "$NVIM" --remote-tab {{filename}}' > editAtLine: 'nvim --server "$NVIM" --remote-tab {{filename}};...

it seems run tsserver.reloadProjects have same problem.

> Should be issue with tsserver > Should be issue with tsserver Watchman crashed after instantaneously monitoring a vast number of file changes in node_modules, possibly tens of thousands.

> > Can you try setting `winfixwidth` and `winfixheight` on the neo-tree window? > > No change in behavior. Has the problem been solved?

add `.watchman.json` in your project root. so this issue can close now. ```json { "ignore_dirs": [ "dist", "node_modules" ] } ``` link: https://github.com/neoclide/coc.nvim/wiki/Install-coc.nvim#install-watchman-for-file-watching