openjpa-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Oleg Lyalikov (JIRA)" <j...@apache.org>
Subject [jira] [Updated] (OPENJPA-2019) 2 connections can be opened simultaneously during some DataCache work inside single query.getResultList()
Date Mon, 20 Jun 2011 13:51:47 GMT

     [ https://issues.apache.org/jira/browse/OPENJPA-2019?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

Oleg Lyalikov updated OPENJPA-2019:
-----------------------------------

    Description: 
Test configuration:

@Entity
public class Document { ... }

@Entity
public class Person {
    ...
    @JoinTable(name="person_document", joinColumns = {@JoinColumn(name="person_id", referencedColumnName="ID")},
inverseJoinColumns = {@JoinColumn(name="documents_id", referencedColumnName="ID")})
    @OneToMany(fetch = FetchType.EAGER)
    private List<Document> documents;
    ...
}

Test case:
0. Set DataCache to: true(CacheSize=40, SoftReferenceSize=0), and limit connection pool size
by 1 connection
1. Create 20 objects of Person entity (+ 3 documents per each Person): 80 objects total (only
half of objects can be located in data cache)
2. Clear entity manager and run query "SELECT p FROM Person p WHERE p.id IN (?1, ?2, ... ,
?20)"
Openjpa tries to open second connection at some time and fails.

The problem is significant for me because I have lots of threads executing queries like in
2. point and I have hard limits on connection pool size. I cannot have connection_pool_size
> possible_threads_count. So at some time I get connection pool exhausted and application
stops working.

It seems that the problem is related to DataCache work and L1 cache work (I didn't reproduce
the problem with big sized or small sized DataCache - to reproduce it's better to assign DataCache
size = 1/2 of total query objects count. Also I didn't reproduce it with Ehcache and openjpa-2.2.0
with Lru=true).

Attached test project and stack trace of getting second connection.

Sometimes test can successfully finish (~5-10% of all runs) - it's hard to make it deterministic.

  was:
Test configuration:

@Entity
public class Document { ... }

@Entity
public class Person {
    ...
    @JoinTable(name="person_document", joinColumns = {@JoinColumn(name="person_id", referencedColumnName="ID")},
inverseJoinColumns = {@JoinColumn(name="documents_id", referencedColumnName="ID")})
    @OneToMany(fetch = FetchType.EAGER)
    private List<Document> documents;
    ...
}

Test case:
0. Set DataCache to: true(CacheSize=40, SoftReferenceSize=0), and limit connection pool size
by 1 connection
1. Create 20 objects of Person entity (+ 3 documents per each Person): 80 objects total (only
half of objects can be located in data cache)
2. Clear entity manager and run query "SELECT p FROM Person p WHERE p.id IN (?1, ?2, ... ,
?20)"
Openjpa tries to open second connection at some time and fails.

Attached test project and stack trace of getting second connection.

The problem is significant for me because I have lots of threads executing queries like in
2. point and I have hard limits on connection pool size. I cannot have connection_pool_size
> possible_threads_count. So at some time I get connection pool exhausted and application
stops working.

It seems that the problem is related to DataCache work and L1 cache work (I didn't reproduce
the problem with big sized or small sized DataCache - to reproduce it's better to assign DataCache
size = 1/2 of total query objects count. Also I didn't reproduce it with Ehcache and openjpa-2.2.0
with Lru=true).


> 2 connections can be opened simultaneously during some DataCache work inside single query.getResultList()
> ---------------------------------------------------------------------------------------------------------
>
>                 Key: OPENJPA-2019
>                 URL: https://issues.apache.org/jira/browse/OPENJPA-2019
>             Project: OpenJPA
>          Issue Type: Bug
>          Components: datacache, kernel, query
>    Affects Versions: 2.0.0
>         Environment: WinXP SP3, JDK 1.6.0_21, Postgres 
>            Reporter: Oleg Lyalikov
>         Attachments: jpa-data-cache.zip, jpa_datacache_stack_trace.txt
>
>
> Test configuration:
> @Entity
> public class Document { ... }
> @Entity
> public class Person {
>     ...
>     @JoinTable(name="person_document", joinColumns = {@JoinColumn(name="person_id", referencedColumnName="ID")},
inverseJoinColumns = {@JoinColumn(name="documents_id", referencedColumnName="ID")})
>     @OneToMany(fetch = FetchType.EAGER)
>     private List<Document> documents;
>     ...
> }
> Test case:
> 0. Set DataCache to: true(CacheSize=40, SoftReferenceSize=0), and limit connection pool
size by 1 connection
> 1. Create 20 objects of Person entity (+ 3 documents per each Person): 80 objects total
(only half of objects can be located in data cache)
> 2. Clear entity manager and run query "SELECT p FROM Person p WHERE p.id IN (?1, ?2,
... , ?20)"
> Openjpa tries to open second connection at some time and fails.
> The problem is significant for me because I have lots of threads executing queries like
in 2. point and I have hard limits on connection pool size. I cannot have connection_pool_size
> possible_threads_count. So at some time I get connection pool exhausted and application
stops working.
> It seems that the problem is related to DataCache work and L1 cache work (I didn't reproduce
the problem with big sized or small sized DataCache - to reproduce it's better to assign DataCache
size = 1/2 of total query objects count. Also I didn't reproduce it with Ehcache and openjpa-2.2.0
with Lru=true).
> Attached test project and stack trace of getting second connection.
> Sometimes test can successfully finish (~5-10% of all runs) - it's hard to make it deterministic.

--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira

        

Mime
View raw message