You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@ode.apache.org by Pratik Bhatt <bh...@gmail.com> on 2009/12/12 21:25:15 UTC
Large amount of deadlocks occuring when running more then 2 processes
simultanously
Hi,
We have deployed Ode v 1.3.3 onto jboss app server 5 with MS SQL Server 2005
as its persistence store using hibernate.
We are facing a large amount of deadlocks when we run more then a few
processes simultaneously. From the traces on sql server, it looks like the
deadlocks are occurring on the table BPEL_PLNK_VAL. From my initial guess,
it seems like whenever a process is created/deleted it locks the table and
thus causes every other thread which tries to select from that table to
deadlock on it.
This is killing the throughput of the system. As the scheduler retries every
1 minute, the system can execute atmost one process per minute.
So is there any good way of avoiding the deadlocks by using some settings
etc.
Also please find below the ode-axis2.properties files.
ode-axis2.threads.pool.size=100
ode-axis2.db.mode=EXTERNAL
ode-axis2.db.ext.dataSource=java:ApacheODEDS
ode.persistence=hibernate
ode-axis2.tx.factory.class=org.apache.ode.axis2.util.JBossFactory
ode-axis2.dao.factory=org.apache.ode.daohib.bpel.BpelDAOConnectionFactoryImpl
hibernate.dialect=org.hibernate.dialect.SQLServerDialect
hibernate.hbm2ddl.auto=update
hibernate.show_sql=false
hibernate.current_session_context_class=jta
hibernate.transaction.manager_lookup_class=org.hibernate.transaction.JBossTransactionManagerLookup
ode-axis2.process.dehydration.maximum.age=30000000
ode-axis2.process.dehydration.maximum.count=5000
Thanks in advance for your help.
Regards
Re: Large amount of deadlocks occuring when running more then 2
processes simultanously
Posted by Pratik Bhatt <bh...@gmail.com>.
Dear Rafał Rusin,
Thanks a lot for the information. The indexes worked out great. No more
deadlocks on ode.
Thanks
- Pratik
On Sat, Dec 12, 2009 at 12:25 PM, Pratik Bhatt <bh...@gmail.com> wrote:
> Hi,
>
>
>
> We have deployed Ode v 1.3.3 onto jboss app server 5 with MS SQL Server
> 2005 as its persistence store using hibernate.
>
>
> We are facing a large amount of deadlocks when we run more then a few
> processes simultaneously. From the traces on sql server, it looks like the
> deadlocks are occurring on the table BPEL_PLNK_VAL. From my initial guess,
> it seems like whenever a process is created/deleted it locks the table and
> thus causes every other thread which tries to select from that table to
> deadlock on it.
>
>
> This is killing the throughput of the system. As the scheduler retries
> every 1 minute, the system can execute atmost one process per minute.
>
>
> So is there any good way of avoiding the deadlocks by using some settings
> etc.
>
>
> Also please find below the ode-axis2.properties files.
>
>
> ode-axis2.threads.pool.size=100
> ode-axis2.db.mode=EXTERNAL
> ode-axis2.db.ext.dataSource=java:ApacheODEDS
> ode.persistence=hibernate
> ode-axis2.tx.factory.class=org.apache.ode.axis2.util.JBossFactory
>
> ode-axis2.dao.factory=org.apache.ode.daohib.bpel.BpelDAOConnectionFactoryImpl
> hibernate.dialect=org.hibernate.dialect.SQLServerDialect
> hibernate.hbm2ddl.auto=update
> hibernate.show_sql=false
> hibernate.current_session_context_class=jta
>
> hibernate.transaction.manager_lookup_class=org.hibernate.transaction.JBossTransactionManagerLookup
> ode-axis2.process.dehydration.maximum.age=30000000
> ode-axis2.process.dehydration.maximum.count=5000
>
>
> Thanks in advance for your help.
>
>
> Regards
>
Re: Large amount of deadlocks occuring when running more then 2 processes
simultanously
Posted by Veresh Jain <ve...@intalio.com>.
Hello
If problem is not yet resolved then I would recommend keeping the
threads to low ( say 20) and see how system behaves , current value of
100 is too high.
ode-axis2.threads.pool.size=100
HTH
Veresh
On 03/31/2011 02:12 PM, raf123 wrote:
> Hi,
>
> I have a problem with deadlocks. Can somebody tell me what additional
> indexes I need to create?
> The link below doesn't work.
>
> Thanks,
>
> Rafal
>
>
> Rafal Rusin wrote:
>> Possibly you need to create some additional indexes.
>> For oracle + hibernate, you can check out those:
>> http://top.touk.pl/git?p=ode/odeo/.git;a=blob;f=touk/oracle/ode_indexes.sql;h=de532c17c654e6acb7ad405fa259b27a583f5d2f;hb=HEAD
>>
>> I believe it will be easy to convert them to MSSQL.
>> There are also some indexes available in archives, so you can try to
>> search for it.
>>
>> 2009/12/12 Pratik Bhatt<bh...@gmail.com>:
>>> Hi,
>>>
>>>
>>>
>>> We have deployed Ode v 1.3.3 onto jboss app server 5 with MS SQL Server
>>> 2005
>>> as its persistence store using hibernate.
>>>
>>>
>>> We are facing a large amount of deadlocks when we run more then a few
>>> processes simultaneously. From the traces on sql server, it looks like
>>> the
>>> deadlocks are occurring on the table BPEL_PLNK_VAL. From my initial
>>> guess,
>>> it seems like whenever a process is created/deleted it locks the table
>>> and
>>> thus causes every other thread which tries to select from that table to
>>> deadlock on it.
>>>
>>>
>>> This is killing the throughput of the system. As the scheduler retries
>>> every
>>> 1 minute, the system can execute atmost one process per minute.
>>>
>>>
>>> So is there any good way of avoiding the deadlocks by using some settings
>>> etc.
>>>
>>>
>>> Also please find below the ode-axis2.properties files.
>>>
>>>
>>> ode-axis2.threads.pool.size=100
>>> ode-axis2.db.mode=EXTERNAL
>>> ode-axis2.db.ext.dataSource=java:ApacheODEDS
>>> ode.persistence=hibernate
>>> ode-axis2.tx.factory.class=org.apache.ode.axis2.util.JBossFactory
>>> ode-axis2.dao.factory=org.apache.ode.daohib.bpel.BpelDAOConnectionFactoryImpl
>>> hibernate.dialect=org.hibernate.dialect.SQLServerDialect
>>> hibernate.hbm2ddl.auto=update
>>> hibernate.show_sql=false
>>> hibernate.current_session_context_class=jta
>>> hibernate.transaction.manager_lookup_class=org.hibernate.transaction.JBossTransactionManagerLookup
>>> ode-axis2.process.dehydration.maximum.age=30000000
>>> ode-axis2.process.dehydration.maximum.count=5000
>>>
>>>
>>> Thanks in advance for your help.
>>>
>>>
>>> Regards
>>>
>>
>> Regards,
>> --
>> Rafał Rusin
>> http://rrusin.blogspot.com
>> http://www.touk.pl
>> http://top.touk.pl
>>
>>
Re: Large amount of deadlocks occuring when running more then 2 processes
simultanously
Posted by Tammo van Lessen <tv...@gmail.com>.
Hi,
schemas with indexes are also listed here:
http://ode.apache.org/ode-schema.html
hth,
Tammo
On 31.03.2011 23:12, raf123 wrote:
>
> Hi,
>
> I have a problem with deadlocks. Can somebody tell me what additional
> indexes I need to create?
> The link below doesn't work.
>
> Thanks,
>
> Rafal
>
>
> Rafal Rusin wrote:
>>
>> Possibly you need to create some additional indexes.
>> For oracle + hibernate, you can check out those:
>> http://top.touk.pl/git?p=ode/odeo/.git;a=blob;f=touk/oracle/ode_indexes.sql;h=de532c17c654e6acb7ad405fa259b27a583f5d2f;hb=HEAD
>>
>> I believe it will be easy to convert them to MSSQL.
>> There are also some indexes available in archives, so you can try to
>> search for it.
>>
>> 2009/12/12 Pratik Bhatt <bh...@gmail.com>:
>>> Hi,
>>>
>>>
>>>
>>> We have deployed Ode v 1.3.3 onto jboss app server 5 with MS SQL Server
>>> 2005
>>> as its persistence store using hibernate.
>>>
>>>
>>> We are facing a large amount of deadlocks when we run more then a few
>>> processes simultaneously. From the traces on sql server, it looks like
>>> the
>>> deadlocks are occurring on the table BPEL_PLNK_VAL. From my initial
>>> guess,
>>> it seems like whenever a process is created/deleted it locks the table
>>> and
>>> thus causes every other thread which tries to select from that table to
>>> deadlock on it.
>>>
>>>
>>> This is killing the throughput of the system. As the scheduler retries
>>> every
>>> 1 minute, the system can execute atmost one process per minute.
>>>
>>>
>>> So is there any good way of avoiding the deadlocks by using some settings
>>> etc.
>>>
>>>
>>> Also please find below the ode-axis2.properties files.
>>>
>>>
>>> ode-axis2.threads.pool.size=100
>>> ode-axis2.db.mode=EXTERNAL
>>> ode-axis2.db.ext.dataSource=java:ApacheODEDS
>>> ode.persistence=hibernate
>>> ode-axis2.tx.factory.class=org.apache.ode.axis2.util.JBossFactory
>>> ode-axis2.dao.factory=org.apache.ode.daohib.bpel.BpelDAOConnectionFactoryImpl
>>> hibernate.dialect=org.hibernate.dialect.SQLServerDialect
>>> hibernate.hbm2ddl.auto=update
>>> hibernate.show_sql=false
>>> hibernate.current_session_context_class=jta
>>> hibernate.transaction.manager_lookup_class=org.hibernate.transaction.JBossTransactionManagerLookup
>>> ode-axis2.process.dehydration.maximum.age=30000000
>>> ode-axis2.process.dehydration.maximum.count=5000
>>>
>>>
>>> Thanks in advance for your help.
>>>
>>>
>>> Regards
>>>
>>
>>
>> Regards,
>> --
>> Rafał Rusin
>> http://rrusin.blogspot.com
>> http://www.touk.pl
>> http://top.touk.pl
>>
>>
>
--
Tammo van Lessen - http://www.taval.de
Re: Large amount of deadlocks occuring when running more then 2
processes simultanously
Posted by raf123 <ra...@gmail.com>.
Hi,
I have a problem with deadlocks. Can somebody tell me what additional
indexes I need to create?
The link below doesn't work.
Thanks,
Rafal
Rafal Rusin wrote:
>
> Possibly you need to create some additional indexes.
> For oracle + hibernate, you can check out those:
> http://top.touk.pl/git?p=ode/odeo/.git;a=blob;f=touk/oracle/ode_indexes.sql;h=de532c17c654e6acb7ad405fa259b27a583f5d2f;hb=HEAD
>
> I believe it will be easy to convert them to MSSQL.
> There are also some indexes available in archives, so you can try to
> search for it.
>
> 2009/12/12 Pratik Bhatt <bh...@gmail.com>:
>> Hi,
>>
>>
>>
>> We have deployed Ode v 1.3.3 onto jboss app server 5 with MS SQL Server
>> 2005
>> as its persistence store using hibernate.
>>
>>
>> We are facing a large amount of deadlocks when we run more then a few
>> processes simultaneously. From the traces on sql server, it looks like
>> the
>> deadlocks are occurring on the table BPEL_PLNK_VAL. From my initial
>> guess,
>> it seems like whenever a process is created/deleted it locks the table
>> and
>> thus causes every other thread which tries to select from that table to
>> deadlock on it.
>>
>>
>> This is killing the throughput of the system. As the scheduler retries
>> every
>> 1 minute, the system can execute atmost one process per minute.
>>
>>
>> So is there any good way of avoiding the deadlocks by using some settings
>> etc.
>>
>>
>> Also please find below the ode-axis2.properties files.
>>
>>
>> ode-axis2.threads.pool.size=100
>> ode-axis2.db.mode=EXTERNAL
>> ode-axis2.db.ext.dataSource=java:ApacheODEDS
>> ode.persistence=hibernate
>> ode-axis2.tx.factory.class=org.apache.ode.axis2.util.JBossFactory
>> ode-axis2.dao.factory=org.apache.ode.daohib.bpel.BpelDAOConnectionFactoryImpl
>> hibernate.dialect=org.hibernate.dialect.SQLServerDialect
>> hibernate.hbm2ddl.auto=update
>> hibernate.show_sql=false
>> hibernate.current_session_context_class=jta
>> hibernate.transaction.manager_lookup_class=org.hibernate.transaction.JBossTransactionManagerLookup
>> ode-axis2.process.dehydration.maximum.age=30000000
>> ode-axis2.process.dehydration.maximum.count=5000
>>
>>
>> Thanks in advance for your help.
>>
>>
>> Regards
>>
>
>
> Regards,
> --
> Rafał Rusin
> http://rrusin.blogspot.com
> http://www.touk.pl
> http://top.touk.pl
>
>
--
View this message in context: http://old.nabble.com/Large-amount-of-deadlocks-occuring-when-running-more-then-2-processes--simultanously-tp26763513p31290289.html
Sent from the Apache Ode User mailing list archive at Nabble.com.
Re: Large amount of deadlocks occuring when running more then 2
processes simultanously
Posted by Rafal Rusin <ra...@gmail.com>.
Possibly you need to create some additional indexes.
For oracle + hibernate, you can check out those:
http://top.touk.pl/git?p=ode/odeo/.git;a=blob;f=touk/oracle/ode_indexes.sql;h=de532c17c654e6acb7ad405fa259b27a583f5d2f;hb=HEAD
I believe it will be easy to convert them to MSSQL.
There are also some indexes available in archives, so you can try to
search for it.
2009/12/12 Pratik Bhatt <bh...@gmail.com>:
> Hi,
>
>
>
> We have deployed Ode v 1.3.3 onto jboss app server 5 with MS SQL Server 2005
> as its persistence store using hibernate.
>
>
> We are facing a large amount of deadlocks when we run more then a few
> processes simultaneously. From the traces on sql server, it looks like the
> deadlocks are occurring on the table BPEL_PLNK_VAL. From my initial guess,
> it seems like whenever a process is created/deleted it locks the table and
> thus causes every other thread which tries to select from that table to
> deadlock on it.
>
>
> This is killing the throughput of the system. As the scheduler retries every
> 1 minute, the system can execute atmost one process per minute.
>
>
> So is there any good way of avoiding the deadlocks by using some settings
> etc.
>
>
> Also please find below the ode-axis2.properties files.
>
>
> ode-axis2.threads.pool.size=100
> ode-axis2.db.mode=EXTERNAL
> ode-axis2.db.ext.dataSource=java:ApacheODEDS
> ode.persistence=hibernate
> ode-axis2.tx.factory.class=org.apache.ode.axis2.util.JBossFactory
> ode-axis2.dao.factory=org.apache.ode.daohib.bpel.BpelDAOConnectionFactoryImpl
> hibernate.dialect=org.hibernate.dialect.SQLServerDialect
> hibernate.hbm2ddl.auto=update
> hibernate.show_sql=false
> hibernate.current_session_context_class=jta
> hibernate.transaction.manager_lookup_class=org.hibernate.transaction.JBossTransactionManagerLookup
> ode-axis2.process.dehydration.maximum.age=30000000
> ode-axis2.process.dehydration.maximum.count=5000
>
>
> Thanks in advance for your help.
>
>
> Regards
>
Regards,
--
Rafał Rusin
http://rrusin.blogspot.com
http://www.touk.pl
http://top.touk.pl