[ 
https://issues.apache.org/jira/browse/HUDI-7392?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

voon updated HUDI-7392:
-----------------------
    Description: 
When consistent_hashing is enabled and a long running Spark job (Deltastreamer) 
is created, we noticed that there is a gradual increase in CLOSE_WAIT 
connections originating from the AM -> HDFS DN. 

 

Command to check for close waits
{code:java}
netstat -anlpt | grep CLOSE_WAIT | grep 50010{code}
Result
{code:java}
tcp6       1      0 10.1.2.3:45994      10.5.4.3:50010      CLOSE_WAIT  
2446/java
tcp6       1      0 10.1.2.3:48478      10.6.5.4:50010      CLOSE_WAIT  
2446/java
tcp6       1      0 10.1.2.3:49542      10.7.6.5:50010      CLOSE_WAIT  
2446/java
tcp6       1      0 10.1.2.3:47220      10.8.7.6:50010      CLOSE_WAIT  
2446/java
tcp6       1      0 10.1.2.3:49786      10.9.8.7:50010      CLOSE_WAIT  
2446/java {code}
 

To reproduce this:
 
{code:java}
CREATE TABLE dev_hudi.close_wait_issue_investigation (
    id INT,
    name STRING,
    date_col STRING,
    grass_region STRING
) USING hudi
PARTITIONED BY (grass_region)
tblproperties (
    primaryKey = 'id',
    type = 'mor',
    precombineField = 'id',
    hoodie.index.type = 'BUCKET',
    hoodie.index.bucket.engine = 'CONSISTENT_HASHING',     
    hoodie.compact.inline = 'true'
)
LOCATION 'hdfs://DEV/close_wait_issue_investigation';

INSERT INTO dev_hudi.close_wait_issue_investigation VALUES (1, 'alex1', 
'2023-12-22', 'SG');
INSERT INTO dev_hudi.close_wait_issue_investigation VALUES (2, 'alex2', 
'2023-12-22', 'SG');
INSERT INTO dev_hudi.close_wait_issue_investigation VALUES (3, 'alex3', 
'2023-12-22', 'SG');
INSERT INTO dev_hudi.close_wait_issue_investigation VALUES (4, 'alex4', 
'2023-12-22', 'SG');
INSERT INTO dev_hudi.close_wait_issue_investigation VALUES (5, 'alex5', 
'2023-12-22', 'SG');{code}
 
 

 

  was:
When consistent_hashing is enabled and a long running Spark job (Deltastreamer) 
is created, we noticed that there is a gradual increase in CLOSE_WAIT 
connections originating from the AM -> HDFS DN. 

 

Command to check for close waits
{code:java}
netstat -anlpt | grep CLOSE_WAIT | grep 50010{code}
Result

 
{code:java}
tcp6       1      0 10.1.2.3:45994      10.5.4.3:50010      CLOSE_WAIT  
2446/java
tcp6       1      0 10.1.2.3:48478      10.6.5.4:50010      CLOSE_WAIT  
2446/java
tcp6       1      0 10.1.2.3:49542      10.7.6.5:50010      CLOSE_WAIT  
2446/java
tcp6       1      0 10.1.2.3:47220      10.8.7.6:50010      CLOSE_WAIT  
2446/java
tcp6       1      0 10.1.2.3:49786      10.9.8.7:50010      CLOSE_WAIT  
2446/java {code}
 

 


> Fix Connection leak causing CLOSE_WAIT TCP connections
> ------------------------------------------------------
>
>                 Key: HUDI-7392
>                 URL: https://issues.apache.org/jira/browse/HUDI-7392
>             Project: Apache Hudi
>          Issue Type: Bug
>            Reporter: voon
>            Assignee: voon
>            Priority: Major
>
> When consistent_hashing is enabled and a long running Spark job 
> (Deltastreamer) is created, we noticed that there is a gradual increase in 
> CLOSE_WAIT connections originating from the AM -> HDFS DN. 
>  
> Command to check for close waits
> {code:java}
> netstat -anlpt | grep CLOSE_WAIT | grep 50010{code}
> Result
> {code:java}
> tcp6       1      0 10.1.2.3:45994      10.5.4.3:50010      CLOSE_WAIT  
> 2446/java
> tcp6       1      0 10.1.2.3:48478      10.6.5.4:50010      CLOSE_WAIT  
> 2446/java
> tcp6       1      0 10.1.2.3:49542      10.7.6.5:50010      CLOSE_WAIT  
> 2446/java
> tcp6       1      0 10.1.2.3:47220      10.8.7.6:50010      CLOSE_WAIT  
> 2446/java
> tcp6       1      0 10.1.2.3:49786      10.9.8.7:50010      CLOSE_WAIT  
> 2446/java {code}
>  
> To reproduce this:
>  
> {code:java}
> CREATE TABLE dev_hudi.close_wait_issue_investigation (
>     id INT,
>     name STRING,
>     date_col STRING,
>     grass_region STRING
> ) USING hudi
> PARTITIONED BY (grass_region)
> tblproperties (
>     primaryKey = 'id',
>     type = 'mor',
>     precombineField = 'id',
>     hoodie.index.type = 'BUCKET',
>     hoodie.index.bucket.engine = 'CONSISTENT_HASHING',     
>     hoodie.compact.inline = 'true'
> )
> LOCATION 'hdfs://DEV/close_wait_issue_investigation';
> INSERT INTO dev_hudi.close_wait_issue_investigation VALUES (1, 'alex1', 
> '2023-12-22', 'SG');
> INSERT INTO dev_hudi.close_wait_issue_investigation VALUES (2, 'alex2', 
> '2023-12-22', 'SG');
> INSERT INTO dev_hudi.close_wait_issue_investigation VALUES (3, 'alex3', 
> '2023-12-22', 'SG');
> INSERT INTO dev_hudi.close_wait_issue_investigation VALUES (4, 'alex4', 
> '2023-12-22', 'SG');
> INSERT INTO dev_hudi.close_wait_issue_investigation VALUES (5, 'alex5', 
> '2023-12-22', 'SG');{code}
>  
>  
>  



--
This message was sent by Atlassian Jira
(v8.20.10#820010)

Reply via email to