[ 
https://issues.apache.org/jira/browse/FLINK-25296?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17468375#comment-17468375
 ] 

chenqizhu edited comment on FLINK-25296 at 1/4/22, 6:13 AM:
------------------------------------------------------------

Hi [~pnowojski] You are right. i mean adding -Dstate.checkpoints.num-retained 
via command-Line Interface, like this : 
{code:java}
bin/flink run -t yarn-per-job -Dexecution.checkpointing.interval=60s 
-Dstate.checkpoints.num-retained=1 examples/streaming/TopSpeedWindowing.jar.
{code}
This configuration options is valid.

checkpoints files:
hdfs://flinkcluster/flink/checkpoint/0291345f4624476a24dcb5c03e3a303e/chk-3
..
{color:red}As specified in the configuration, only one file is retained
{color}
But When I remove this configuration,now the command line becomes like this: 
{code:java}
bin/flink run -t yarn-per-job -Dexecution.checkpointing.interval=60s 
examples/streaming/TopSpeedWindowing.jar.
{code}
{color:red}The default value of this configuration options didn't take effect.
{color}

checkpoints files:
hdfs://flinkcluster/flink/checkpoint/c03ac3a71c52f2f33fd518089cc78c76/chk-1
hdfs://flinkcluster/flink/checkpoint/c03ac3a71c52f2f33fd518089cc78c76/chk-2
......

I checked the flink source code for this configuration, and the default is 
indeed 1.

{code:java}
    public static final ConfigOption<Integer> MAX_RETAINED_CHECKPOINTS =
            ConfigOptions.key("state.checkpoints.num-retained")
                    .defaultValue(1)
                    .withDescription("The maximum number of completed 
checkpoints to retain.");
{code}




was (Author: libra_816):
Hi [~pnowojski] You are right. i mean adding -Dstate.checkpoints.num-retained 
via command-Line Interface, like this : 
{code:java}
bin/flink run -t yarn-per-job -Dexecution.checkpointing.interval=60s 
-Dstate.checkpoints.num-retained=1 examples/streaming/TopSpeedWindowing.jar.
{code}
This configuration options is valid.

checkpoints files:
hdfs://flinkcluster/flink/checkpoint/0291345f4624476a24dcb5c03e3a303e/chk-3
..
{color:red}As specified in the configuration, only one file is retained
{color}
But When I remove this configuration,now the command line becomes like this: 
{code:java}
bin/flink run -t yarn-per-job -Dexecution.checkpointing.interval=60s 
examples/streaming/TopSpeedWindowing.jar.
{code}
{color:red}The default value of this configuration options didn't take effect.
{color}checkpoints files:
hdfs://flinkcluster/flink/checkpoint/c03ac3a71c52f2f33fd518089cc78c76/chk-1
hdfs://flinkcluster/flink/checkpoint/c03ac3a71c52f2f33fd518089cc78c76/chk-2
......

I checked the flink source code for this configuration, and the default is 
indeed 1.

{code:java}
    public static final ConfigOption<Integer> MAX_RETAINED_CHECKPOINTS =
            ConfigOptions.key("state.checkpoints.num-retained")
                    .defaultValue(1)
                    .withDescription("The maximum number of completed 
checkpoints to retain.");
{code}



> [state.checkpoints.num-retained ]The default value does not take effect
> -----------------------------------------------------------------------
>
>                 Key: FLINK-25296
>                 URL: https://issues.apache.org/jira/browse/FLINK-25296
>             Project: Flink
>          Issue Type: Bug
>          Components: Runtime / Checkpointing, Runtime / Coordination
>    Affects Versions: 1.13.3
>            Reporter: chenqizhu
>            Priority: Major
>
> The configration is valid only when I add -Dstate.checkpoints.num-retained=n 
> via command-Line Interface
>  If I do not specify this configuration, the default value does not take 
> effect , is it a bug ?



--
This message was sent by Atlassian Jira
(v8.20.1#820001)

Reply via email to