You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user-zh@flink.apache.org by "Harold.Miao" <mi...@gmail.com> on 2020/09/14 10:44:08 UTC
[sql-client][checkpoint] sql-client提交任务，checkpoint没有写入hdfs
hi all
flink 版本： 1.11.1
我们利用sql-client提交任务， flink-conf.yaml配置如下
state.backend: filesystem
state.backend.fs.checkpointdir:
hdfs:///ai/flink/checkpoint/dataclean/hl-redis0902/checkpoint-data/23252
state.checkpoints.dir:
hdfs:///ai/flink/checkpoint/dataclean/hl-redis0902/checkpoint-meta/23252
state.savepoints.dir:
hdfs:///ai/flink/checkpoint/dataclean/hl-redis0902/savepoint/23252
execution.checkpointing.externalized-checkpoint-retention:
RETAIN_ON_CANCELLATION
execution.checkpointing.interval: 60s
execution.checkpointing.mode: EXACTLY_ONCE
jobmanager.execution.failover-strategy: full
state.backend.incremental: true
任务运行后，在UI界面上看checkpoint都成功了。 但是hdfs上面却一直只有一个meta文件
类似下面：
hdfs://10.218.60.57:8020/ai/flink/checkpoint/dataclean/hl-redis0902/checkpoint-meta/23250/c72c1ee4362c3d0ba72db32698363fcf/chk-5/_metadata
除了这个文件，其他什么都没有。
我们的源是kafka，kafka肯定会保存state的。
请教大家这是什么原因导致的呢
谢谢
--
Best Regards,
Harold Miao
Re: [sql-client][checkpoint] sql-client提交任务，checkpoint没有写入hdfs
Posted by Congxian Qiu <qc...@gmail.com>.
Hi
你可以参考这里[1] 自己进行一些修改尝试，来分析 metadata 文件
[1]
https://github.com/apache/flink/blob/master/flink-runtime/src/test/java/org/apache/flink/runtime/checkpoint/CheckpointMetadataLoadingTest.java
Best,
Congxian
Harold.Miao <mi...@gmail.com> 于2020年9月15日周二 下午1:58写道：
> 是同一个insert任务， 只是重启任务的时候 ，我加了这些代码，构造一个 SavepointRestoreSettings 来恢复cp
>
> 请教 我如何判断cp真正写入了hdfs呢，meta文件有什么工具可以解析吗
>
> 谢谢
>
> Jark Wu <im...@gmail.com> 于2020年9月15日周二 上午11:31写道：
>
> > 是不是你的 cp 恢复的代码，没有执行任何的 insert into 语句？
> >
> > On Mon, 14 Sep 2020 at 20:15, Harold.Miao <mi...@gmail.com> wrote:
> >
> > > 还有一点是 我们修改了sql-client代码， 让任务从cp恢复，修改如下
> > >
> > > private StreamExecutionEnvironment createStreamExecutionEnvironment() {
> > > final StreamExecutionEnvironment env =
> > > StreamExecutionEnvironment.getExecutionEnvironment();
> > >
> > >
> > >
> > >
> > >
> > >
> > >
> > > * LOG.info("restore cp exist: {}",
> > > environment.getExecution().getRestoreSp().isPresent()); if
> > > (environment.getExecution().getRestoreSp().isPresent()) {
> > > LOG.info("restore cp path: {}",
> > > environment.getExecution().getRestoreSp().get()); if
> > > (!environment.getExecution().getRestoreSp().get().contains("none")) {
> > > SavepointRestoreSettings savepointRestoreSettings =
> > >
> > >
> >
> SavepointRestoreSettings.forPath(environment.getExecution().getRestoreSp().get(),
> > > true);
> > >
> >
> env.getStreamGraph().setSavepointRestoreSettings(savepointRestoreSettings);
> > > } }*
> > > // for TimeCharacteristic validation in StreamTableEnvironmentImpl
> > >
> > >
> >
> env.setStreamTimeCharacteristic(environment.getExecution().getTimeCharacteristic());
> > > if (env.getStreamTimeCharacteristic() ==
> > TimeCharacteristic.EventTime) {
> > >
> > >
> >
> env.getConfig().setAutoWatermarkInterval(environment.getExecution().getPeriodicWatermarksInterval());
> > > }
> > > return env;
> > > }
> > >
> > >
> > > 传入上面那个只有meta文件地址的时候报错如下：
> > >
> > > Exception in thread "main"
> > > org.apache.flink.table.client.SqlClientException: Unexpected
> > > exception. This is a bug. Please consider filing an issue.
> > > at
> > org.apache.flink.table.client.SqlClient.main(SqlClient.java:213)
> > > Caused by: org.apache.flink.table.client.gateway.SqlExecutionException:
> > > Could not create execution context.
> > > at
> > >
> >
> org.apache.flink.table.client.gateway.local.ExecutionContext$Builder.build(ExecutionContext.java:879)
> > > at
> > >
> >
> org.apache.flink.table.client.gateway.local.LocalExecutor.openSession(LocalExecutor.java:227)
> > > at
> > > org.apache.flink.table.client.SqlClient.start(SqlClient.java:108)
> > > at
> > org.apache.flink.table.client.SqlClient.main(SqlClient.java:201)
> > > Caused by: java.lang.IllegalStateException: No operators defined in
> > > streaming topology. Cannot execute.
> > > at
> > >
> >
> org.apache.flink.streaming.api.environment.StreamExecutionEnvironment.getStreamGraphGenerator(StreamExecutionEnvironment.java:1870)
> > > at
> > >
> >
> org.apache.flink.streaming.api.environment.StreamExecutionEnvironment.getStreamGraph(StreamExecutionEnvironment.java:1861)
> > > at
> > >
> >
> org.apache.flink.streaming.api.environment.StreamExecutionEnvironment.getStreamGraph(StreamExecutionEnvironment.java:1846)
> > > at
> > >
> >
> org.apache.flink.streaming.api.environment.StreamExecutionEnvironment.getStreamGraph(StreamExecutionEnvironment.java:1834)
> > > at
> > >
> >
> org.apache.flink.table.client.gateway.local.ExecutionContext.createStreamExecutionEnvironment(ExecutionContext.java:691)
> > > at
> > >
> >
> org.apache.flink.table.client.gateway.local.ExecutionContext.createTableEnvironment(ExecutionContext.java:593)
> > > at
> > >
> >
> org.apache.flink.table.client.gateway.local.ExecutionContext.initializeTableEnvironment(ExecutionContext.java:498)
> > > at
> > >
> >
> org.apache.flink.table.client.gateway.local.ExecutionContext.<init>(ExecutionContext.java:184)
> > > at
> > >
> >
> org.apache.flink.table.client.gateway.local.ExecutionContext.<init>(ExecutionContext.java:137)
> > > at
> > >
> >
> org.apache.flink.table.client.gateway.local.ExecutionContext$Builder.build(ExecutionContext.java:868)
> > > ... 3 more
> > >
> > >
> > > 错误很明显的显示没有算子的state
> > >
> > >
> > >
> > >
> > >
> > >
> > >
> > >
> > > Congxian Qiu <qc...@gmail.com> 于2020年9月14日周一 下午7:53写道：
> > >
> > > > Hi
> > > > 如果你的 state 都非常小的话，可能就会保存在 meta 文件中了，这样的话就只有 _metadata
> > > > 这一个文件的。具体逻辑可以看一下这里[1]
> > > >
> > > > [1]
> > > >
> > > >
> > >
> >
> https://github.com/apache/flink/blob/9b0fb562898b809b860cf0065ded7a45c49300af/flink-runtime/src/main/java/org/apache/flink/runtime/checkpoint/metadata/MetadataV2V3SerializerBase.java#L442
> > > > Best,
> > > > Congxian
> > > >
> > > >
> > > > Harold.Miao <mi...@gmail.com> 于2020年9月14日周一 下午6:44写道：
> > > >
> > > > > hi all
> > > > >
> > > > > flink 版本： 1.11.1
> > > > >
> > > > > 我们利用sql-client提交任务， flink-conf.yaml配置如下
> > > > >
> > > > > state.backend: filesystem
> > > > > state.backend.fs.checkpointdir:
> > > > >
> > >
> hdfs:///ai/flink/checkpoint/dataclean/hl-redis0902/checkpoint-data/23252
> > > > > state.checkpoints.dir:
> > > > >
> > >
> hdfs:///ai/flink/checkpoint/dataclean/hl-redis0902/checkpoint-meta/23252
> > > > > state.savepoints.dir:
> > > > > hdfs:///ai/flink/checkpoint/dataclean/hl-redis0902/savepoint/23252
> > > > >
> > > > > execution.checkpointing.externalized-checkpoint-retention:
> > > > > RETAIN_ON_CANCELLATION
> > > > > execution.checkpointing.interval: 60s
> > > > > execution.checkpointing.mode: EXACTLY_ONCE
> > > > > jobmanager.execution.failover-strategy: full
> > > > > state.backend.incremental: true
> > > > >
> > > > >
> > > > > 任务运行后，在UI界面上看checkpoint都成功了。 但是hdfs上面却一直只有一个meta文件
> > > > >
> > > > > 类似下面：
> > > > >
> > > > > hdfs://
> > > > >
> > > >
> > >
> >
> 10.218.60.57:8020/ai/flink/checkpoint/dataclean/hl-redis0902/checkpoint-meta/23250/c72c1ee4362c3d0ba72db32698363fcf/chk-5/_metadata
> > > > >
> > > > > 除了这个文件，其他什么都没有。
> > > > >
> > > > > 我们的源是kafka，kafka肯定会保存state的。
> > > > >
> > > > >
> > > > > 请教大家这是什么原因导致的呢
> > > > >
> > > > >
> > > > > 谢谢
> > > > >
> > > > >
> > > > >
> > > > >
> > > > >
> > > > >
> > > > >
> > > > >
> > > > >
> > > > >
> > > > > --
> > > > >
> > > > > Best Regards,
> > > > > Harold Miao
> > > > >
> > > >
> > >
> > >
> > > --
> > >
> > > Best Regards,
> > > Harold Miao
> > >
> >
>
>
> --
>
> Best Regards,
> Harold Miao
>
Re: [sql-client][checkpoint] sql-client提交任务，checkpoint没有写入hdfs
Posted by "Harold.Miao" <mi...@gmail.com>.
是同一个insert任务， 只是重启任务的时候 ，我加了这些代码，构造一个 SavepointRestoreSettings 来恢复cp
请教 我如何判断cp真正写入了hdfs呢，meta文件有什么工具可以解析吗
谢谢
Jark Wu <im...@gmail.com> 于2020年9月15日周二 上午11:31写道：
> 是不是你的 cp 恢复的代码，没有执行任何的 insert into 语句？
>
> On Mon, 14 Sep 2020 at 20:15, Harold.Miao <mi...@gmail.com> wrote:
>
> > 还有一点是 我们修改了sql-client代码， 让任务从cp恢复，修改如下
> >
> > private StreamExecutionEnvironment createStreamExecutionEnvironment() {
> > final StreamExecutionEnvironment env =
> > StreamExecutionEnvironment.getExecutionEnvironment();
> >
> >
> >
> >
> >
> >
> >
> > * LOG.info("restore cp exist: {}",
> > environment.getExecution().getRestoreSp().isPresent()); if
> > (environment.getExecution().getRestoreSp().isPresent()) {
> > LOG.info("restore cp path: {}",
> > environment.getExecution().getRestoreSp().get()); if
> > (!environment.getExecution().getRestoreSp().get().contains("none")) {
> > SavepointRestoreSettings savepointRestoreSettings =
> >
> >
> SavepointRestoreSettings.forPath(environment.getExecution().getRestoreSp().get(),
> > true);
> >
> env.getStreamGraph().setSavepointRestoreSettings(savepointRestoreSettings);
> > } }*
> > // for TimeCharacteristic validation in StreamTableEnvironmentImpl
> >
> >
> env.setStreamTimeCharacteristic(environment.getExecution().getTimeCharacteristic());
> > if (env.getStreamTimeCharacteristic() ==
> TimeCharacteristic.EventTime) {
> >
> >
> env.getConfig().setAutoWatermarkInterval(environment.getExecution().getPeriodicWatermarksInterval());
> > }
> > return env;
> > }
> >
> >
> > 传入上面那个只有meta文件地址的时候报错如下：
> >
> > Exception in thread "main"
> > org.apache.flink.table.client.SqlClientException: Unexpected
> > exception. This is a bug. Please consider filing an issue.
> > at
> org.apache.flink.table.client.SqlClient.main(SqlClient.java:213)
> > Caused by: org.apache.flink.table.client.gateway.SqlExecutionException:
> > Could not create execution context.
> > at
> >
> org.apache.flink.table.client.gateway.local.ExecutionContext$Builder.build(ExecutionContext.java:879)
> > at
> >
> org.apache.flink.table.client.gateway.local.LocalExecutor.openSession(LocalExecutor.java:227)
> > at
> > org.apache.flink.table.client.SqlClient.start(SqlClient.java:108)
> > at
> org.apache.flink.table.client.SqlClient.main(SqlClient.java:201)
> > Caused by: java.lang.IllegalStateException: No operators defined in
> > streaming topology. Cannot execute.
> > at
> >
> org.apache.flink.streaming.api.environment.StreamExecutionEnvironment.getStreamGraphGenerator(StreamExecutionEnvironment.java:1870)
> > at
> >
> org.apache.flink.streaming.api.environment.StreamExecutionEnvironment.getStreamGraph(StreamExecutionEnvironment.java:1861)
> > at
> >
> org.apache.flink.streaming.api.environment.StreamExecutionEnvironment.getStreamGraph(StreamExecutionEnvironment.java:1846)
> > at
> >
> org.apache.flink.streaming.api.environment.StreamExecutionEnvironment.getStreamGraph(StreamExecutionEnvironment.java:1834)
> > at
> >
> org.apache.flink.table.client.gateway.local.ExecutionContext.createStreamExecutionEnvironment(ExecutionContext.java:691)
> > at
> >
> org.apache.flink.table.client.gateway.local.ExecutionContext.createTableEnvironment(ExecutionContext.java:593)
> > at
> >
> org.apache.flink.table.client.gateway.local.ExecutionContext.initializeTableEnvironment(ExecutionContext.java:498)
> > at
> >
> org.apache.flink.table.client.gateway.local.ExecutionContext.<init>(ExecutionContext.java:184)
> > at
> >
> org.apache.flink.table.client.gateway.local.ExecutionContext.<init>(ExecutionContext.java:137)
> > at
> >
> org.apache.flink.table.client.gateway.local.ExecutionContext$Builder.build(ExecutionContext.java:868)
> > ... 3 more
> >
> >
> > 错误很明显的显示没有算子的state
> >
> >
> >
> >
> >
> >
> >
> >
> > Congxian Qiu <qc...@gmail.com> 于2020年9月14日周一 下午7:53写道：
> >
> > > Hi
> > > 如果你的 state 都非常小的话，可能就会保存在 meta 文件中了，这样的话就只有 _metadata
> > > 这一个文件的。具体逻辑可以看一下这里[1]
> > >
> > > [1]
> > >
> > >
> >
> https://github.com/apache/flink/blob/9b0fb562898b809b860cf0065ded7a45c49300af/flink-runtime/src/main/java/org/apache/flink/runtime/checkpoint/metadata/MetadataV2V3SerializerBase.java#L442
> > > Best,
> > > Congxian
> > >
> > >
> > > Harold.Miao <mi...@gmail.com> 于2020年9月14日周一 下午6:44写道：
> > >
> > > > hi all
> > > >
> > > > flink 版本： 1.11.1
> > > >
> > > > 我们利用sql-client提交任务， flink-conf.yaml配置如下
> > > >
> > > > state.backend: filesystem
> > > > state.backend.fs.checkpointdir:
> > > >
> > hdfs:///ai/flink/checkpoint/dataclean/hl-redis0902/checkpoint-data/23252
> > > > state.checkpoints.dir:
> > > >
> > hdfs:///ai/flink/checkpoint/dataclean/hl-redis0902/checkpoint-meta/23252
> > > > state.savepoints.dir:
> > > > hdfs:///ai/flink/checkpoint/dataclean/hl-redis0902/savepoint/23252
> > > >
> > > > execution.checkpointing.externalized-checkpoint-retention:
> > > > RETAIN_ON_CANCELLATION
> > > > execution.checkpointing.interval: 60s
> > > > execution.checkpointing.mode: EXACTLY_ONCE
> > > > jobmanager.execution.failover-strategy: full
> > > > state.backend.incremental: true
> > > >
> > > >
> > > > 任务运行后，在UI界面上看checkpoint都成功了。 但是hdfs上面却一直只有一个meta文件
> > > >
> > > > 类似下面：
> > > >
> > > > hdfs://
> > > >
> > >
> >
> 10.218.60.57:8020/ai/flink/checkpoint/dataclean/hl-redis0902/checkpoint-meta/23250/c72c1ee4362c3d0ba72db32698363fcf/chk-5/_metadata
> > > >
> > > > 除了这个文件，其他什么都没有。
> > > >
> > > > 我们的源是kafka，kafka肯定会保存state的。
> > > >
> > > >
> > > > 请教大家这是什么原因导致的呢
> > > >
> > > >
> > > > 谢谢
> > > >
> > > >
> > > >
> > > >
> > > >
> > > >
> > > >
> > > >
> > > >
> > > >
> > > > --
> > > >
> > > > Best Regards,
> > > > Harold Miao
> > > >
> > >
> >
> >
> > --
> >
> > Best Regards,
> > Harold Miao
> >
>
--
Best Regards,
Harold Miao
Re: [sql-client][checkpoint] sql-client提交任务，checkpoint没有写入hdfs
Posted by "Harold.Miao" <mi...@gmail.com>.
是我的代码问题，我set sp的时候streamGraph里面的算子还没有构建出来，正确的做法是在生成jobGraph的时候set 进去。 感谢
Jark Wu <im...@gmail.com> 于2020年9月15日周二 上午11:31写道：
> 是不是你的 cp 恢复的代码，没有执行任何的 insert into 语句？
>
> On Mon, 14 Sep 2020 at 20:15, Harold.Miao <mi...@gmail.com> wrote:
>
> > 还有一点是 我们修改了sql-client代码， 让任务从cp恢复，修改如下
> >
> > private StreamExecutionEnvironment createStreamExecutionEnvironment() {
> > final StreamExecutionEnvironment env =
> > StreamExecutionEnvironment.getExecutionEnvironment();
> >
> >
> >
> >
> >
> >
> >
> > * LOG.info("restore cp exist: {}",
> > environment.getExecution().getRestoreSp().isPresent()); if
> > (environment.getExecution().getRestoreSp().isPresent()) {
> > LOG.info("restore cp path: {}",
> > environment.getExecution().getRestoreSp().get()); if
> > (!environment.getExecution().getRestoreSp().get().contains("none")) {
> > SavepointRestoreSettings savepointRestoreSettings =
> >
> >
> SavepointRestoreSettings.forPath(environment.getExecution().getRestoreSp().get(),
> > true);
> >
> env.getStreamGraph().setSavepointRestoreSettings(savepointRestoreSettings);
> > } }*
> > // for TimeCharacteristic validation in StreamTableEnvironmentImpl
> >
> >
> env.setStreamTimeCharacteristic(environment.getExecution().getTimeCharacteristic());
> > if (env.getStreamTimeCharacteristic() ==
> TimeCharacteristic.EventTime) {
> >
> >
> env.getConfig().setAutoWatermarkInterval(environment.getExecution().getPeriodicWatermarksInterval());
> > }
> > return env;
> > }
> >
> >
> > 传入上面那个只有meta文件地址的时候报错如下：
> >
> > Exception in thread "main"
> > org.apache.flink.table.client.SqlClientException: Unexpected
> > exception. This is a bug. Please consider filing an issue.
> > at
> org.apache.flink.table.client.SqlClient.main(SqlClient.java:213)
> > Caused by: org.apache.flink.table.client.gateway.SqlExecutionException:
> > Could not create execution context.
> > at
> >
> org.apache.flink.table.client.gateway.local.ExecutionContext$Builder.build(ExecutionContext.java:879)
> > at
> >
> org.apache.flink.table.client.gateway.local.LocalExecutor.openSession(LocalExecutor.java:227)
> > at
> > org.apache.flink.table.client.SqlClient.start(SqlClient.java:108)
> > at
> org.apache.flink.table.client.SqlClient.main(SqlClient.java:201)
> > Caused by: java.lang.IllegalStateException: No operators defined in
> > streaming topology. Cannot execute.
> > at
> >
> org.apache.flink.streaming.api.environment.StreamExecutionEnvironment.getStreamGraphGenerator(StreamExecutionEnvironment.java:1870)
> > at
> >
> org.apache.flink.streaming.api.environment.StreamExecutionEnvironment.getStreamGraph(StreamExecutionEnvironment.java:1861)
> > at
> >
> org.apache.flink.streaming.api.environment.StreamExecutionEnvironment.getStreamGraph(StreamExecutionEnvironment.java:1846)
> > at
> >
> org.apache.flink.streaming.api.environment.StreamExecutionEnvironment.getStreamGraph(StreamExecutionEnvironment.java:1834)
> > at
> >
> org.apache.flink.table.client.gateway.local.ExecutionContext.createStreamExecutionEnvironment(ExecutionContext.java:691)
> > at
> >
> org.apache.flink.table.client.gateway.local.ExecutionContext.createTableEnvironment(ExecutionContext.java:593)
> > at
> >
> org.apache.flink.table.client.gateway.local.ExecutionContext.initializeTableEnvironment(ExecutionContext.java:498)
> > at
> >
> org.apache.flink.table.client.gateway.local.ExecutionContext.<init>(ExecutionContext.java:184)
> > at
> >
> org.apache.flink.table.client.gateway.local.ExecutionContext.<init>(ExecutionContext.java:137)
> > at
> >
> org.apache.flink.table.client.gateway.local.ExecutionContext$Builder.build(ExecutionContext.java:868)
> > ... 3 more
> >
> >
> > 错误很明显的显示没有算子的state
> >
> >
> >
> >
> >
> >
> >
> >
> > Congxian Qiu <qc...@gmail.com> 于2020年9月14日周一 下午7:53写道：
> >
> > > Hi
> > > 如果你的 state 都非常小的话，可能就会保存在 meta 文件中了，这样的话就只有 _metadata
> > > 这一个文件的。具体逻辑可以看一下这里[1]
> > >
> > > [1]
> > >
> > >
> >
> https://github.com/apache/flink/blob/9b0fb562898b809b860cf0065ded7a45c49300af/flink-runtime/src/main/java/org/apache/flink/runtime/checkpoint/metadata/MetadataV2V3SerializerBase.java#L442
> > > Best,
> > > Congxian
> > >
> > >
> > > Harold.Miao <mi...@gmail.com> 于2020年9月14日周一 下午6:44写道：
> > >
> > > > hi all
> > > >
> > > > flink 版本： 1.11.1
> > > >
> > > > 我们利用sql-client提交任务， flink-conf.yaml配置如下
> > > >
> > > > state.backend: filesystem
> > > > state.backend.fs.checkpointdir:
> > > >
> > hdfs:///ai/flink/checkpoint/dataclean/hl-redis0902/checkpoint-data/23252
> > > > state.checkpoints.dir:
> > > >
> > hdfs:///ai/flink/checkpoint/dataclean/hl-redis0902/checkpoint-meta/23252
> > > > state.savepoints.dir:
> > > > hdfs:///ai/flink/checkpoint/dataclean/hl-redis0902/savepoint/23252
> > > >
> > > > execution.checkpointing.externalized-checkpoint-retention:
> > > > RETAIN_ON_CANCELLATION
> > > > execution.checkpointing.interval: 60s
> > > > execution.checkpointing.mode: EXACTLY_ONCE
> > > > jobmanager.execution.failover-strategy: full
> > > > state.backend.incremental: true
> > > >
> > > >
> > > > 任务运行后，在UI界面上看checkpoint都成功了。 但是hdfs上面却一直只有一个meta文件
> > > >
> > > > 类似下面：
> > > >
> > > > hdfs://
> > > >
> > >
> >
> 10.218.60.57:8020/ai/flink/checkpoint/dataclean/hl-redis0902/checkpoint-meta/23250/c72c1ee4362c3d0ba72db32698363fcf/chk-5/_metadata
> > > >
> > > > 除了这个文件，其他什么都没有。
> > > >
> > > > 我们的源是kafka，kafka肯定会保存state的。
> > > >
> > > >
> > > > 请教大家这是什么原因导致的呢
> > > >
> > > >
> > > > 谢谢
> > > >
> > > >
> > > >
> > > >
> > > >
> > > >
> > > >
> > > >
> > > >
> > > >
> > > > --
> > > >
> > > > Best Regards,
> > > > Harold Miao
> > > >
> > >
> >
> >
> > --
> >
> > Best Regards,
> > Harold Miao
> >
>
--
Best Regards,
Harold Miao
Re: [sql-client][checkpoint] sql-client提交任务，checkpoint没有写入hdfs
Posted by Jark Wu <im...@gmail.com>.
是不是你的 cp 恢复的代码，没有执行任何的 insert into 语句？
On Mon, 14 Sep 2020 at 20:15, Harold.Miao <mi...@gmail.com> wrote:
> 还有一点是 我们修改了sql-client代码， 让任务从cp恢复，修改如下
>
> private StreamExecutionEnvironment createStreamExecutionEnvironment() {
> final StreamExecutionEnvironment env =
> StreamExecutionEnvironment.getExecutionEnvironment();
>
>
>
>
>
>
>
> * LOG.info("restore cp exist: {}",
> environment.getExecution().getRestoreSp().isPresent()); if
> (environment.getExecution().getRestoreSp().isPresent()) {
> LOG.info("restore cp path: {}",
> environment.getExecution().getRestoreSp().get()); if
> (!environment.getExecution().getRestoreSp().get().contains("none")) {
> SavepointRestoreSettings savepointRestoreSettings =
>
> SavepointRestoreSettings.forPath(environment.getExecution().getRestoreSp().get(),
> true);
> env.getStreamGraph().setSavepointRestoreSettings(savepointRestoreSettings);
> } }*
> // for TimeCharacteristic validation in StreamTableEnvironmentImpl
>
> env.setStreamTimeCharacteristic(environment.getExecution().getTimeCharacteristic());
> if (env.getStreamTimeCharacteristic() == TimeCharacteristic.EventTime) {
>
> env.getConfig().setAutoWatermarkInterval(environment.getExecution().getPeriodicWatermarksInterval());
> }
> return env;
> }
>
>
> 传入上面那个只有meta文件地址的时候报错如下：
>
> Exception in thread "main"
> org.apache.flink.table.client.SqlClientException: Unexpected
> exception. This is a bug. Please consider filing an issue.
> at org.apache.flink.table.client.SqlClient.main(SqlClient.java:213)
> Caused by: org.apache.flink.table.client.gateway.SqlExecutionException:
> Could not create execution context.
> at
> org.apache.flink.table.client.gateway.local.ExecutionContext$Builder.build(ExecutionContext.java:879)
> at
> org.apache.flink.table.client.gateway.local.LocalExecutor.openSession(LocalExecutor.java:227)
> at
> org.apache.flink.table.client.SqlClient.start(SqlClient.java:108)
> at org.apache.flink.table.client.SqlClient.main(SqlClient.java:201)
> Caused by: java.lang.IllegalStateException: No operators defined in
> streaming topology. Cannot execute.
> at
> org.apache.flink.streaming.api.environment.StreamExecutionEnvironment.getStreamGraphGenerator(StreamExecutionEnvironment.java:1870)
> at
> org.apache.flink.streaming.api.environment.StreamExecutionEnvironment.getStreamGraph(StreamExecutionEnvironment.java:1861)
> at
> org.apache.flink.streaming.api.environment.StreamExecutionEnvironment.getStreamGraph(StreamExecutionEnvironment.java:1846)
> at
> org.apache.flink.streaming.api.environment.StreamExecutionEnvironment.getStreamGraph(StreamExecutionEnvironment.java:1834)
> at
> org.apache.flink.table.client.gateway.local.ExecutionContext.createStreamExecutionEnvironment(ExecutionContext.java:691)
> at
> org.apache.flink.table.client.gateway.local.ExecutionContext.createTableEnvironment(ExecutionContext.java:593)
> at
> org.apache.flink.table.client.gateway.local.ExecutionContext.initializeTableEnvironment(ExecutionContext.java:498)
> at
> org.apache.flink.table.client.gateway.local.ExecutionContext.<init>(ExecutionContext.java:184)
> at
> org.apache.flink.table.client.gateway.local.ExecutionContext.<init>(ExecutionContext.java:137)
> at
> org.apache.flink.table.client.gateway.local.ExecutionContext$Builder.build(ExecutionContext.java:868)
> ... 3 more
>
>
> 错误很明显的显示没有算子的state
>
>
>
>
>
>
>
>
> Congxian Qiu <qc...@gmail.com> 于2020年9月14日周一 下午7:53写道：
>
> > Hi
> > 如果你的 state 都非常小的话，可能就会保存在 meta 文件中了，这样的话就只有 _metadata
> > 这一个文件的。具体逻辑可以看一下这里[1]
> >
> > [1]
> >
> >
> https://github.com/apache/flink/blob/9b0fb562898b809b860cf0065ded7a45c49300af/flink-runtime/src/main/java/org/apache/flink/runtime/checkpoint/metadata/MetadataV2V3SerializerBase.java#L442
> > Best,
> > Congxian
> >
> >
> > Harold.Miao <mi...@gmail.com> 于2020年9月14日周一 下午6:44写道：
> >
> > > hi all
> > >
> > > flink 版本： 1.11.1
> > >
> > > 我们利用sql-client提交任务， flink-conf.yaml配置如下
> > >
> > > state.backend: filesystem
> > > state.backend.fs.checkpointdir:
> > >
> hdfs:///ai/flink/checkpoint/dataclean/hl-redis0902/checkpoint-data/23252
> > > state.checkpoints.dir:
> > >
> hdfs:///ai/flink/checkpoint/dataclean/hl-redis0902/checkpoint-meta/23252
> > > state.savepoints.dir:
> > > hdfs:///ai/flink/checkpoint/dataclean/hl-redis0902/savepoint/23252
> > >
> > > execution.checkpointing.externalized-checkpoint-retention:
> > > RETAIN_ON_CANCELLATION
> > > execution.checkpointing.interval: 60s
> > > execution.checkpointing.mode: EXACTLY_ONCE
> > > jobmanager.execution.failover-strategy: full
> > > state.backend.incremental: true
> > >
> > >
> > > 任务运行后，在UI界面上看checkpoint都成功了。 但是hdfs上面却一直只有一个meta文件
> > >
> > > 类似下面：
> > >
> > > hdfs://
> > >
> >
> 10.218.60.57:8020/ai/flink/checkpoint/dataclean/hl-redis0902/checkpoint-meta/23250/c72c1ee4362c3d0ba72db32698363fcf/chk-5/_metadata
> > >
> > > 除了这个文件，其他什么都没有。
> > >
> > > 我们的源是kafka，kafka肯定会保存state的。
> > >
> > >
> > > 请教大家这是什么原因导致的呢
> > >
> > >
> > > 谢谢
> > >
> > >
> > >
> > >
> > >
> > >
> > >
> > >
> > >
> > >
> > > --
> > >
> > > Best Regards,
> > > Harold Miao
> > >
> >
>
>
> --
>
> Best Regards,
> Harold Miao
>
Re: [sql-client][checkpoint] sql-client提交任务，checkpoint没有写入hdfs
Posted by "Harold.Miao" <mi...@gmail.com>.
还有一点是 我们修改了sql-client代码， 让任务从cp恢复，修改如下
private StreamExecutionEnvironment createStreamExecutionEnvironment() {
final StreamExecutionEnvironment env =
StreamExecutionEnvironment.getExecutionEnvironment();
* LOG.info("restore cp exist: {}",
environment.getExecution().getRestoreSp().isPresent()); if
(environment.getExecution().getRestoreSp().isPresent()) {
LOG.info("restore cp path: {}",
environment.getExecution().getRestoreSp().get()); if
(!environment.getExecution().getRestoreSp().get().contains("none")) {
SavepointRestoreSettings savepointRestoreSettings =
SavepointRestoreSettings.forPath(environment.getExecution().getRestoreSp().get(),
true); env.getStreamGraph().setSavepointRestoreSettings(savepointRestoreSettings);
} }*
// for TimeCharacteristic validation in StreamTableEnvironmentImpl
env.setStreamTimeCharacteristic(environment.getExecution().getTimeCharacteristic());
if (env.getStreamTimeCharacteristic() == TimeCharacteristic.EventTime) {
env.getConfig().setAutoWatermarkInterval(environment.getExecution().getPeriodicWatermarksInterval());
}
return env;
}
传入上面那个只有meta文件地址的时候报错如下：
Exception in thread "main"
org.apache.flink.table.client.SqlClientException: Unexpected
exception. This is a bug. Please consider filing an issue.
at org.apache.flink.table.client.SqlClient.main(SqlClient.java:213)
Caused by: org.apache.flink.table.client.gateway.SqlExecutionException:
Could not create execution context.
at org.apache.flink.table.client.gateway.local.ExecutionContext$Builder.build(ExecutionContext.java:879)
at org.apache.flink.table.client.gateway.local.LocalExecutor.openSession(LocalExecutor.java:227)
at org.apache.flink.table.client.SqlClient.start(SqlClient.java:108)
at org.apache.flink.table.client.SqlClient.main(SqlClient.java:201)
Caused by: java.lang.IllegalStateException: No operators defined in
streaming topology. Cannot execute.
at org.apache.flink.streaming.api.environment.StreamExecutionEnvironment.getStreamGraphGenerator(StreamExecutionEnvironment.java:1870)
at org.apache.flink.streaming.api.environment.StreamExecutionEnvironment.getStreamGraph(StreamExecutionEnvironment.java:1861)
at org.apache.flink.streaming.api.environment.StreamExecutionEnvironment.getStreamGraph(StreamExecutionEnvironment.java:1846)
at org.apache.flink.streaming.api.environment.StreamExecutionEnvironment.getStreamGraph(StreamExecutionEnvironment.java:1834)
at org.apache.flink.table.client.gateway.local.ExecutionContext.createStreamExecutionEnvironment(ExecutionContext.java:691)
at org.apache.flink.table.client.gateway.local.ExecutionContext.createTableEnvironment(ExecutionContext.java:593)
at org.apache.flink.table.client.gateway.local.ExecutionContext.initializeTableEnvironment(ExecutionContext.java:498)
at org.apache.flink.table.client.gateway.local.ExecutionContext.<init>(ExecutionContext.java:184)
at org.apache.flink.table.client.gateway.local.ExecutionContext.<init>(ExecutionContext.java:137)
at org.apache.flink.table.client.gateway.local.ExecutionContext$Builder.build(ExecutionContext.java:868)
... 3 more
错误很明显的显示没有算子的state
Congxian Qiu <qc...@gmail.com> 于2020年9月14日周一 下午7:53写道：
> Hi
> 如果你的 state 都非常小的话，可能就会保存在 meta 文件中了，这样的话就只有 _metadata
> 这一个文件的。具体逻辑可以看一下这里[1]
>
> [1]
>
> https://github.com/apache/flink/blob/9b0fb562898b809b860cf0065ded7a45c49300af/flink-runtime/src/main/java/org/apache/flink/runtime/checkpoint/metadata/MetadataV2V3SerializerBase.java#L442
> Best,
> Congxian
>
>
> Harold.Miao <mi...@gmail.com> 于2020年9月14日周一 下午6:44写道：
>
> > hi all
> >
> > flink 版本： 1.11.1
> >
> > 我们利用sql-client提交任务， flink-conf.yaml配置如下
> >
> > state.backend: filesystem
> > state.backend.fs.checkpointdir:
> > hdfs:///ai/flink/checkpoint/dataclean/hl-redis0902/checkpoint-data/23252
> > state.checkpoints.dir:
> > hdfs:///ai/flink/checkpoint/dataclean/hl-redis0902/checkpoint-meta/23252
> > state.savepoints.dir:
> > hdfs:///ai/flink/checkpoint/dataclean/hl-redis0902/savepoint/23252
> >
> > execution.checkpointing.externalized-checkpoint-retention:
> > RETAIN_ON_CANCELLATION
> > execution.checkpointing.interval: 60s
> > execution.checkpointing.mode: EXACTLY_ONCE
> > jobmanager.execution.failover-strategy: full
> > state.backend.incremental: true
> >
> >
> > 任务运行后，在UI界面上看checkpoint都成功了。 但是hdfs上面却一直只有一个meta文件
> >
> > 类似下面：
> >
> > hdfs://
> >
> 10.218.60.57:8020/ai/flink/checkpoint/dataclean/hl-redis0902/checkpoint-meta/23250/c72c1ee4362c3d0ba72db32698363fcf/chk-5/_metadata
> >
> > 除了这个文件，其他什么都没有。
> >
> > 我们的源是kafka，kafka肯定会保存state的。
> >
> >
> > 请教大家这是什么原因导致的呢
> >
> >
> > 谢谢
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> > --
> >
> > Best Regards,
> > Harold Miao
> >
>
--
Best Regards,
Harold Miao
Re: [sql-client][checkpoint] sql-client提交任务，checkpoint没有写入hdfs
Posted by Congxian Qiu <qc...@gmail.com>.
Hi
如果你的 state 都非常小的话，可能就会保存在 meta 文件中了，这样的话就只有 _metadata
这一个文件的。具体逻辑可以看一下这里[1]
[1]
https://github.com/apache/flink/blob/9b0fb562898b809b860cf0065ded7a45c49300af/flink-runtime/src/main/java/org/apache/flink/runtime/checkpoint/metadata/MetadataV2V3SerializerBase.java#L442
Best,
Congxian
Harold.Miao <mi...@gmail.com> 于2020年9月14日周一 下午6:44写道：
> hi all
>
> flink 版本： 1.11.1
>
> 我们利用sql-client提交任务， flink-conf.yaml配置如下
>
> state.backend: filesystem
> state.backend.fs.checkpointdir:
> hdfs:///ai/flink/checkpoint/dataclean/hl-redis0902/checkpoint-data/23252
> state.checkpoints.dir:
> hdfs:///ai/flink/checkpoint/dataclean/hl-redis0902/checkpoint-meta/23252
> state.savepoints.dir:
> hdfs:///ai/flink/checkpoint/dataclean/hl-redis0902/savepoint/23252
>
> execution.checkpointing.externalized-checkpoint-retention:
> RETAIN_ON_CANCELLATION
> execution.checkpointing.interval: 60s
> execution.checkpointing.mode: EXACTLY_ONCE
> jobmanager.execution.failover-strategy: full
> state.backend.incremental: true
>
>
> 任务运行后，在UI界面上看checkpoint都成功了。 但是hdfs上面却一直只有一个meta文件
>
> 类似下面：
>
> hdfs://
> 10.218.60.57:8020/ai/flink/checkpoint/dataclean/hl-redis0902/checkpoint-meta/23250/c72c1ee4362c3d0ba72db32698363fcf/chk-5/_metadata
>
> 除了这个文件，其他什么都没有。
>
> 我们的源是kafka，kafka肯定会保存state的。
>
>
> 请教大家这是什么原因导致的呢
>
>
> 谢谢
>
>
>
>
>
>
>
>
>
>
> --
>
> Best Regards,
> Harold Miao
>