FLINK WEEKLY 2019/48 <https://zhuanlan.zhihu.com/p/95448913> 感谢社区同学 forideal 负责编写本期 FLINK WEEKLY!
用户问题 - 如何成为flink的contributor <https://link.zhihu.com/?target=https%3A//lists.apache.org/x/thread.html/199bef8c0abd13db180014fd06c7edb101159e23bab2242ca9916a13%40%253Cdev.flink.apache.org%253E> 用户可以直接在感兴趣的jira下面回复,社区可能会把这个assign给用户 - why operator not chained? <https://link.zhihu.com/?target=https%3A//lists.apache.org/x/thread.html/431edefb1ede9c5238b35f41c191ac82644c23f4ba4e88440e22a9b4%40%253Cuser.flink.apache.org%253E> - Uid and name for Flink sources and sinks <https://link.zhihu.com/?target=https%3A//lists.apache.org/x/thread.html/b5ae277791f14c270409d026d58ed9e963209c84b0adeb851aaef3f3%40%253Cuser.flink.apache.org%253E> 用户设置了operatoruid和name,并针对用户自己的情况提了一些问题,邮件列表讨论了这些问题 - HBase ScannerTimeoutException and double Result processing <https://link.zhihu.com/?target=https%3A//lists.apache.org/x/thread.html/e79a8934f0ed4ab9a2fc0828969b78e97c7b7e097e4b79553dcb8447%40%253Cuser.flink.apache.org%253E> 用户在使用flink读取hbase的数据的时候发生了超时异常,邮件列表从源码层面分析出现这个情况的原因 - Dynamically creating new Task Managers in YARN <https://link.zhihu.com/?target=https%3A//lists.apache.org/x/thread.html/dc95db62ca77bf87066103be0d2b169b6ed0742fc39377d694d65155%40%253Cuser.flink.apache.org%253E>用户想按需动态的申请taskmanager,邮件列表讨论了 Flink on Yarn 的session mode和job mode,并针对问题给出了一些建议 - Per Operator State Monitoring <https://link.zhihu.com/?target=https%3A//lists.apache.org/x/thread.html/46cc075039ce57a71810caed954e4f518c2fa7cc8ae2ef51f5c70f02%40%253Cuser.flink.apache.org%253E> 用户的job被kill了,用户猜测是state导致。邮件列表讨论了某些可能监控的方法,同时讨论了有可能是 state backend为RocksDB导致 - Apache Flink - Throttling stream flow <https://link.zhihu.com/?target=https%3A//lists.apache.org/x/thread.html/2e48ce2f7827c4dcb91739cd64f6298fb60a0a99acee5aecd60b07b7%40%253Cuser.flink.apache.org%253E> 邮件列表讨论了flink source限流的问题 - Read multiline JSON/XML <https://link.zhihu.com/?target=https%3A//lists.apache.org/x/thread.html/1238f5c7b76828161dfd5df5d389f0206aeaa9760987fad3ce5ee10f%40%253Cuser.flink.apache.org%253E> 用户询问在Flink中是否存在像spark读取多行json/xml的api,邮件列表讨论了一些方法 - Apache Flink - Troubleshooting exception while starting the job from savepoint <https://link.zhihu.com/?target=https%3A//lists.apache.org/x/thread.html/1b677fec080ec0aa78e0208b3f3b76c15ef8fa1379eb642e96b4da86%40%253Cuser.flink.apache.org%253E> 用户有一个Job,在没有任何修改的前提下,无法使用 save point重启。经过邮件讨论,用户发现有一个有状态的operator没有被分配name和uid。详细问题排查过程,参考邮件列表。 - How to recover state from savepoint on embedded mode? <https://link.zhihu.com/?target=https%3A//lists.apache.org/x/thread.html/e6a0f5dc8b9dcc22a035dbde5abd236212ca726109d7e45b1c8fa900%40%253Cuser.flink.apache.org%253E> 在flink embedded mode下,用户想从save point中恢复job。邮件列表讨论了可能可行的办法。有兴趣的可以去看看 - Metrics for Task States <https://link.zhihu.com/?target=https%3A//lists.apache.org/x/thread.html/ac96420bd7071c640be47b110f6b4aff6dbbc8f48b028db4e113f4dd%40%253Cuser.flink.apache.org%253E> 开发讨论 - [PROPOSAL/SURVEY] Enable background cleanup for state with TTL by default <https://link.zhihu.com/?target=https%3A//lists.apache.org/x/thread.html/e6390d44975b0768c56d3c3d6ec1cac2aac0626a9d1566d456999d2a%40%253Cuser.flink.apache.org%253E> - [DISCUSS] Disable conversion between TIMESTAMP and Long in parameters and results of UDXs <https://link.zhihu.com/?target=https%3A//lists.apache.org/x/thread.html/db8c9a47a7c13756c7145ef597c94e2217537c9137409c8bb64244dc%40%253Cuser.flink.apache.org%253E> - [DISCUSS] Support JSON functions in Flink SQL <https://link.zhihu.com/?target=https%3A//lists.apache.org/x/thread.html/e259aa70432e4003e1598e8f8db844813a869a0dd96accfc1b73deb6%40%253Cdev.flink.apache.org%253E> - [DISCUSS] Releasing Flink 1.8.3 <https://link.zhihu.com/?target=https%3A//lists.apache.org/x/thread.html/bd80d4bcb8b5ce810b516d5d48812aac51ec1d2df80f67f4fdd6c1ad%40%253Cdev.flink.apache.org%253E> Flink 1.8.3 即将发版了 - set job level TTL <https://link.zhihu.com/?target=https%3A//lists.apache.org/x/thread.html/9812f03735ff258e7a93ede1a0baaee424f343eb64b64e0defad9c6f%40%253Cdev.flink.apache.org%253E> 社区发展 - Apache Flink-shaded 9.0 released <https://link.zhihu.com/?target=https%3A//lists.apache.org/x/thread.html/fca50d391ce2c19934e58b1b9042e2eb5eea2f7e9be8dc0f2bc8c492%40%253Cdev.flink.apache.org%253E> - Flink Forward Asia 2019 <https://link.zhihu.com/?target=https%3A//m.aliyun.com/markets/aliyun/developer/ffa2019> 2019年11月28/29 Flink Forward Asia 2019 在北京召开 -