Re: flink sink kafka from checkpoint run failed

[tianxy]

我也遇到了 请问你解决了没




--
Sent from: http://apache-flink.147419.n8.nabble.com/

temporarily unavailable due to an ongoing leader election

[tianxy]

flink1.12.2 on yarn 3.1.5  作业手动触发savepoint 成功   然后使用run -s 启动成功  但是访问flinkui
报service temporarily unavailable due to an ongoing leader election




--
Sent from: http://apache-flink.147419.n8.nabble.com/

flink1.12.2，interval join并没有 inProcessingTime() and inEventTime()

[tianxy]

 

FLIP-134: Batch execution for the DataStream API
Allow explicitly configuring time behaviour on KeyedStream.intervalJoin()
FLINK-19479

Before Flink 1.12 the KeyedStream.intervalJoin() operation was changing
behavior based on the globally set Stream TimeCharacteristic. In Flink 1.12
we introduced explicit inProcessingTime() and inEventTime() methods on
IntervalJoin and the join no longer changes behaviour based on the global
characteristic.


flink1.12.2，interval join并没有 inProcessingTime() and inEventTime()并没有找到对应的方法 
是不是还不支持 求解答！！！



--
Sent from: http://apache-flink.147419.n8.nabble.com/

flink1.12.2使用rocksdb状态后端，checkpoint size变大

[tianxy]

452 
COMPLETED
103/103 2021-04-22 17:29:12 2021-04-22 17:29:12 325ms   4.40 MB 0 B 
(5.39 KB)
451 
COMPLETED
103/103 2021-04-22 17:28:12 2021-04-22 17:28:12 122ms   4.43 MB 9.36 KB 
(15.2
KB)
450 
COMPLETED
103/103 2021-04-22 17:27:12 2021-04-22 17:27:12 124ms   4.43 MB 0 B 
(5.39 KB)
449 
COMPLETED
103/103 2021-04-22 17:26:12 2021-04-22 17:26:12 112ms   4.48 MB 0 B 
(5.39 KB)
448 
COMPLETED
103/103 2021-04-22 17:25:12 2021-04-22 17:25:12 117ms   4.35 MB 0 B 
(5.39 KB)
447 
COMPLETED
103/103 2021-04-22 17:24:12 2021-04-22 17:24:12 113ms   4.41 MB 113 B 
(5.71
KB)
446 
COMPLETED
103/103 2021-04-22 17:23:12 2021-04-22 17:23:12 108ms   4.42 MB 9 B 
(5.86 KB)
445 
COMPLETED
103/103 2021-04-22 17:22:12 2021-04-22 17:22:12 130ms   4.47 MB 0 B 
(5.39 KB)
444 
COMPLETED
103/103 2021-04-22 17:21:12 2021-04-22 17:21:12 155ms   4.35 MB 0 B 
(5.39 KB)
443 
COMPLETED
103/103 2021-04-22 17:20:12 2021-04-22 17:20:12 119ms   4.45 MB 19.0 KB 
(46.1
KB)


flink1.12.2使用rocksdb作为后端存储时，发现ck size不断变大，我使用的是interval join
按理是不断失效的，应该在某个值附近摆动，同样的程序换成使用 fsstatebackend时发现就一直维持在几百k附近 不会一直变大。
请问这个是什么原因呢？ 本来采取fsstatebackend，结果发现 运行一段时间（比如几个小时后）就会突然出现ck
失败，5分钟内无法完成ck超时，看日志并不是作业报错，只是单纯的 报以下错误：

Checkpoint 19 of job dd7b3ab0ec365d23c5dfa25dcf53a730 expired before complet
java.util.concurrent.CancellationException: null

请教大佬，求解答！



--
Sent from: http://apache-flink.147419.n8.nabble.com/

Re: 疑问：当开启state.backend.incremental 后 Checkpointed Data Size 会不断变大

[tianxy]

你好  我也遇到了 所以这个问题你知道原因了没



--
Sent from: http://apache-flink.147419.n8.nabble.com/