You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@ode.apache.org by Sathwik B P <sa...@gmail.com> on 2014/04/18 07:33:24 UTC

Re: Re:Re: Management API function be blocked until all the instances run to the end

Hi Vito,

I was on vacation and hence could not respond.

Are you on Mysql or Postgres?

I suspect that bitronix might be using the default isolation setting of the
underlying database. Try with explicit setting of the isolation level in
resource.properties file and let us know how it goes.

regards,
sathwik


On Thu, Mar 27, 2014 at 10:38 PM, Tammo van Lessen <tv...@gmail.com>wrote:

> Hi Vito,
>
> the PMAPI runs in a transaction as well, but actually I should run smoothly
> between process instances. Are your instances long or short running? Are
> you using a synchronous request response operation? Without the transaction
> manager you will probably run into weird situations like race conditions
> when using message correlation, so this is actually not an option.
>
> Best,
>   Tammo
>
>
> On Thu, Mar 27, 2014 at 9:26 AM, Vito <te...@163.com> wrote:
>
> > Hi all,
> >
> > These days I tried to invoke listInstancesSummary functions of Management
> > API after invoke many instances, but the functions were all blocked until
> > all the instances ran to the end. I thought of abandoning the bitronix
> > transaction manager, and it works. The functions are not blocked any
> more.
> > Do you know why bitronix block my functions, and can I do some
> modification
> > to the bitronix configuration in ODE?
> >
> > Regards,
> > Vito
> >
> >
> >
> >
> >
> >
> > At 2014-03-25 19:54:51,Vito <te...@163.com> wrote:
> > >Hi Sathwik,
> > >
> > >Actually I did not solve the problem of combining openjpa and
> postgresql,
> > and the two communities also did not give me much information of the
> > problem. What I do is to use hibernate DAO to combine ODE and postgresql,
> > and it's working just fine.
> > >
> > >The listInstancesSummary function is already in the Management API. I
> > just invoked it and found it blocked sometimes when I had many instances
> > running. After looking into the ODE source code, I find that the function
> > listAllInstances() actually have little difference from
> > listInstancesSummary(), while the former function have few filters than
> the
> > latter.
> > >
> > >Have you met with the problem that the Management API be blocked? I
> think
> > it may be caused by some transaction problems.
> > >
> > >Regards,
> > >Vito
> > >
> > >
> > >
> > >
> > >
> > >
> > >
> > >
> > >At 2014-03-25 17:02:24,"Sathwik B P" <sa...@gmail.com> wrote:
> > >>Vito,
> > >>
> > >>Good to know you could solve the issue with Postgresql. It would be
> great
> > >>if you could help the community know about the changes you did to make
> it
> > >>work :)
> > >>
> > >>You might have a look at listAllInstances which will list all the
> > instances.
> > >>
> > >>What is the query that used in listInstancesSummary?
> > >>
> > >>regards,
> > >>sathwik
> > >>
> > >>
> > >>On Tue, Mar 25, 2014 at 11:33 AM, Vito <te...@163.com> wrote:
> > >>
> > >>> Hi all,
> > >>> Thank you for your kindly help these days. I just make ODE up and
> > running
> > >>> with mysql and postgresql finally! But when I try some experiments,
> > other
> > >>> problems seem to come up. Actually, it may not be called an error
> > because
> > >>> there is no error or exception.
> > >>> The problem came out when I ran multiple instances. It met no problem
> > >>> until I try to list instances summary from ODE Management API to get
> > the
> > >>> instance number in the engine. The list function returned no results
> > until
> > >>> all the instances ran to the end. I think it may be blocked by the
> > >>> transactions of running instances.
> > >>> Do you think the transactions really block the management api
> > function? If
> > >>> this really happens, how can I get the number of instance number of
> the
> > >>> current engine?
> > >>> Best,
> > >>> Vito
> > >>>
> >
>
>
>
> --
> Tammo van Lessen - http://www.taval.de
>

Re:Re: Re: Re:Re: Management API function be blocked until all the instances run to the end

Posted by Vito <te...@163.com>.
Hi Sathwik,

I've run with Postgres with hibernate well for a couple of days, and delete my file and configuration of Postgres and openjpa.
I may try to reproduce the problem next week. If the problem occurs again, I'll file a Jira and inform you of the issue.

Regards,
Vito





At 2014-04-21 15:28:48,"Sathwik B P" <sa...@gmail.com> wrote:
>Hi Vito,
>
>Can you file a Jira for Postgres with openjpa.
>Kindly include your configurations for bitronix, ode-axis2.properties along
>with exact version of ODE, Postgres used and if possible the process
>definition. Also the description of the problem you faced.
>
>Let me try replicating the environment.
>
>regards,
>sathwik
>
>
>On Mon, Apr 21, 2014 at 9:55 AM, Vito <te...@163.com> wrote:
>
>> Hi Sathwik,
>>
>>
>> Thank you for your reply.
>>
>>
>> I'm working on both Mysql and Postgres. The problem with isolation level
>> happened on Postgres with openjpa. I tried the explicit setting of the
>> isolation level, but it didn't solve the problem. So I used hibernate both
>> on Mysql and Postgres.
>> I slowed down all the requests on both databases, and they succeeded to
>> function well most of the time except for some transaction-intensive
>> periods.
>>
>>
>> regards,
>> Vito
>>
>>
>>
>>
>> At 2014-04-18 13:33:24,"Sathwik B P" <sa...@gmail.com> wrote:
>> >Hi Vito,
>> >
>> >I was on vacation and hence could not respond.
>> >
>> >Are you on Mysql or Postgres?
>> >
>> >I suspect that bitronix might be using the default isolation setting of
>> the
>> >underlying database. Try with explicit setting of the isolation level in
>> >resource.properties file and let us know how it goes.
>> >
>> >regards,
>> >sathwik
>> >
>> >
>> >On Thu, Mar 27, 2014 at 10:38 PM, Tammo van Lessen <tvanlessen@gmail.com
>> >wrote:
>> >
>> >> Hi Vito,
>> >>
>> >> the PMAPI runs in a transaction as well, but actually I should run
>> smoothly
>> >> between process instances. Are your instances long or short running? Are
>> >> you using a synchronous request response operation? Without the
>> transaction
>> >> manager you will probably run into weird situations like race conditions
>> >> when using message correlation, so this is actually not an option.
>> >>
>> >> Best,
>> >>   Tammo
>> >>
>> >>
>> >> On Thu, Mar 27, 2014 at 9:26 AM, Vito <te...@163.com> wrote:
>> >>
>> >> > Hi all,
>> >> >
>> >> > These days I tried to invoke listInstancesSummary functions of
>> Management
>> >> > API after invoke many instances, but the functions were all blocked
>> until
>> >> > all the instances ran to the end. I thought of abandoning the bitronix
>> >> > transaction manager, and it works. The functions are not blocked any
>> >> more.
>> >> > Do you know why bitronix block my functions, and can I do some
>> >> modification
>> >> > to the bitronix configuration in ODE?
>> >> >
>> >> > Regards,
>> >> > Vito
>> >> >
>> >> >
>> >> >
>> >> >
>> >> >
>> >> >
>> >> > At 2014-03-25 19:54:51,Vito <te...@163.com> wrote:
>> >> > >Hi Sathwik,
>> >> > >
>> >> > >Actually I did not solve the problem of combining openjpa and
>> >> postgresql,
>> >> > and the two communities also did not give me much information of the
>> >> > problem. What I do is to use hibernate DAO to combine ODE and
>> postgresql,
>> >> > and it's working just fine.
>> >> > >
>> >> > >The listInstancesSummary function is already in the Management API. I
>> >> > just invoked it and found it blocked sometimes when I had many
>> instances
>> >> > running. After looking into the ODE source code, I find that the
>> function
>> >> > listAllInstances() actually have little difference from
>> >> > listInstancesSummary(), while the former function have few filters
>> than
>> >> the
>> >> > latter.
>> >> > >
>> >> > >Have you met with the problem that the Management API be blocked? I
>> >> think
>> >> > it may be caused by some transaction problems.
>> >> > >
>> >> > >Regards,
>> >> > >Vito
>> >> > >
>> >> > >
>> >> > >
>> >> > >
>> >> > >
>> >> > >
>> >> > >
>> >> > >
>> >> > >At 2014-03-25 17:02:24,"Sathwik B P" <sa...@gmail.com> wrote:
>> >> > >>Vito,
>> >> > >>
>> >> > >>Good to know you could solve the issue with Postgresql. It would be
>> >> great
>> >> > >>if you could help the community know about the changes you did to
>> make
>> >> it
>> >> > >>work :)
>> >> > >>
>> >> > >>You might have a look at listAllInstances which will list all the
>> >> > instances.
>> >> > >>
>> >> > >>What is the query that used in listInstancesSummary?
>> >> > >>
>> >> > >>regards,
>> >> > >>sathwik
>> >> > >>
>> >> > >>
>> >> > >>On Tue, Mar 25, 2014 at 11:33 AM, Vito <te...@163.com> wrote:
>> >> > >>
>> >> > >>> Hi all,
>> >> > >>> Thank you for your kindly help these days. I just make ODE up and
>> >> > running
>> >> > >>> with mysql and postgresql finally! But when I try some
>> experiments,
>> >> > other
>> >> > >>> problems seem to come up. Actually, it may not be called an error
>> >> > because
>> >> > >>> there is no error or exception.
>> >> > >>> The problem came out when I ran multiple instances. It met no
>> problem
>> >> > >>> until I try to list instances summary from ODE Management API to
>> get
>> >> > the
>> >> > >>> instance number in the engine. The list function returned no
>> results
>> >> > until
>> >> > >>> all the instances ran to the end. I think it may be blocked by the
>> >> > >>> transactions of running instances.
>> >> > >>> Do you think the transactions really block the management api
>> >> > function? If
>> >> > >>> this really happens, how can I get the number of instance number
>> of
>> >> the
>> >> > >>> current engine?
>> >> > >>> Best,
>> >> > >>> Vito
>> >> > >>>
>> >> >
>> >>
>> >>
>> >>
>> >> --
>> >> Tammo van Lessen - http://www.taval.de
>> >>
>>

Re: Re: Re:Re: Management API function be blocked until all the instances run to the end

Posted by Sathwik B P <sa...@gmail.com>.
Hi Vito,

Can you file a Jira for Postgres with openjpa.
Kindly include your configurations for bitronix, ode-axis2.properties along
with exact version of ODE, Postgres used and if possible the process
definition. Also the description of the problem you faced.

Let me try replicating the environment.

regards,
sathwik


On Mon, Apr 21, 2014 at 9:55 AM, Vito <te...@163.com> wrote:

> Hi Sathwik,
>
>
> Thank you for your reply.
>
>
> I'm working on both Mysql and Postgres. The problem with isolation level
> happened on Postgres with openjpa. I tried the explicit setting of the
> isolation level, but it didn't solve the problem. So I used hibernate both
> on Mysql and Postgres.
> I slowed down all the requests on both databases, and they succeeded to
> function well most of the time except for some transaction-intensive
> periods.
>
>
> regards,
> Vito
>
>
>
>
> At 2014-04-18 13:33:24,"Sathwik B P" <sa...@gmail.com> wrote:
> >Hi Vito,
> >
> >I was on vacation and hence could not respond.
> >
> >Are you on Mysql or Postgres?
> >
> >I suspect that bitronix might be using the default isolation setting of
> the
> >underlying database. Try with explicit setting of the isolation level in
> >resource.properties file and let us know how it goes.
> >
> >regards,
> >sathwik
> >
> >
> >On Thu, Mar 27, 2014 at 10:38 PM, Tammo van Lessen <tvanlessen@gmail.com
> >wrote:
> >
> >> Hi Vito,
> >>
> >> the PMAPI runs in a transaction as well, but actually I should run
> smoothly
> >> between process instances. Are your instances long or short running? Are
> >> you using a synchronous request response operation? Without the
> transaction
> >> manager you will probably run into weird situations like race conditions
> >> when using message correlation, so this is actually not an option.
> >>
> >> Best,
> >>   Tammo
> >>
> >>
> >> On Thu, Mar 27, 2014 at 9:26 AM, Vito <te...@163.com> wrote:
> >>
> >> > Hi all,
> >> >
> >> > These days I tried to invoke listInstancesSummary functions of
> Management
> >> > API after invoke many instances, but the functions were all blocked
> until
> >> > all the instances ran to the end. I thought of abandoning the bitronix
> >> > transaction manager, and it works. The functions are not blocked any
> >> more.
> >> > Do you know why bitronix block my functions, and can I do some
> >> modification
> >> > to the bitronix configuration in ODE?
> >> >
> >> > Regards,
> >> > Vito
> >> >
> >> >
> >> >
> >> >
> >> >
> >> >
> >> > At 2014-03-25 19:54:51,Vito <te...@163.com> wrote:
> >> > >Hi Sathwik,
> >> > >
> >> > >Actually I did not solve the problem of combining openjpa and
> >> postgresql,
> >> > and the two communities also did not give me much information of the
> >> > problem. What I do is to use hibernate DAO to combine ODE and
> postgresql,
> >> > and it's working just fine.
> >> > >
> >> > >The listInstancesSummary function is already in the Management API. I
> >> > just invoked it and found it blocked sometimes when I had many
> instances
> >> > running. After looking into the ODE source code, I find that the
> function
> >> > listAllInstances() actually have little difference from
> >> > listInstancesSummary(), while the former function have few filters
> than
> >> the
> >> > latter.
> >> > >
> >> > >Have you met with the problem that the Management API be blocked? I
> >> think
> >> > it may be caused by some transaction problems.
> >> > >
> >> > >Regards,
> >> > >Vito
> >> > >
> >> > >
> >> > >
> >> > >
> >> > >
> >> > >
> >> > >
> >> > >
> >> > >At 2014-03-25 17:02:24,"Sathwik B P" <sa...@gmail.com> wrote:
> >> > >>Vito,
> >> > >>
> >> > >>Good to know you could solve the issue with Postgresql. It would be
> >> great
> >> > >>if you could help the community know about the changes you did to
> make
> >> it
> >> > >>work :)
> >> > >>
> >> > >>You might have a look at listAllInstances which will list all the
> >> > instances.
> >> > >>
> >> > >>What is the query that used in listInstancesSummary?
> >> > >>
> >> > >>regards,
> >> > >>sathwik
> >> > >>
> >> > >>
> >> > >>On Tue, Mar 25, 2014 at 11:33 AM, Vito <te...@163.com> wrote:
> >> > >>
> >> > >>> Hi all,
> >> > >>> Thank you for your kindly help these days. I just make ODE up and
> >> > running
> >> > >>> with mysql and postgresql finally! But when I try some
> experiments,
> >> > other
> >> > >>> problems seem to come up. Actually, it may not be called an error
> >> > because
> >> > >>> there is no error or exception.
> >> > >>> The problem came out when I ran multiple instances. It met no
> problem
> >> > >>> until I try to list instances summary from ODE Management API to
> get
> >> > the
> >> > >>> instance number in the engine. The list function returned no
> results
> >> > until
> >> > >>> all the instances ran to the end. I think it may be blocked by the
> >> > >>> transactions of running instances.
> >> > >>> Do you think the transactions really block the management api
> >> > function? If
> >> > >>> this really happens, how can I get the number of instance number
> of
> >> the
> >> > >>> current engine?
> >> > >>> Best,
> >> > >>> Vito
> >> > >>>
> >> >
> >>
> >>
> >>
> >> --
> >> Tammo van Lessen - http://www.taval.de
> >>
>

Re:Re: Re:Re: Management API function be blocked until all the instances run to the end

Posted by Vito <te...@163.com>.
Hi Sathwik,


Thank you for your reply.


I'm working on both Mysql and Postgres. The problem with isolation level happened on Postgres with openjpa. I tried the explicit setting of the isolation level, but it didn't solve the problem. So I used hibernate both on Mysql and Postgres.
I slowed down all the requests on both databases, and they succeeded to function well most of the time except for some transaction-intensive periods.


regards,
Vito




At 2014-04-18 13:33:24,"Sathwik B P" <sa...@gmail.com> wrote:
>Hi Vito,
>
>I was on vacation and hence could not respond.
>
>Are you on Mysql or Postgres?
>
>I suspect that bitronix might be using the default isolation setting of the
>underlying database. Try with explicit setting of the isolation level in
>resource.properties file and let us know how it goes.
>
>regards,
>sathwik
>
>
>On Thu, Mar 27, 2014 at 10:38 PM, Tammo van Lessen <tv...@gmail.com>wrote:
>
>> Hi Vito,
>>
>> the PMAPI runs in a transaction as well, but actually I should run smoothly
>> between process instances. Are your instances long or short running? Are
>> you using a synchronous request response operation? Without the transaction
>> manager you will probably run into weird situations like race conditions
>> when using message correlation, so this is actually not an option.
>>
>> Best,
>>   Tammo
>>
>>
>> On Thu, Mar 27, 2014 at 9:26 AM, Vito <te...@163.com> wrote:
>>
>> > Hi all,
>> >
>> > These days I tried to invoke listInstancesSummary functions of Management
>> > API after invoke many instances, but the functions were all blocked until
>> > all the instances ran to the end. I thought of abandoning the bitronix
>> > transaction manager, and it works. The functions are not blocked any
>> more.
>> > Do you know why bitronix block my functions, and can I do some
>> modification
>> > to the bitronix configuration in ODE?
>> >
>> > Regards,
>> > Vito
>> >
>> >
>> >
>> >
>> >
>> >
>> > At 2014-03-25 19:54:51,Vito <te...@163.com> wrote:
>> > >Hi Sathwik,
>> > >
>> > >Actually I did not solve the problem of combining openjpa and
>> postgresql,
>> > and the two communities also did not give me much information of the
>> > problem. What I do is to use hibernate DAO to combine ODE and postgresql,
>> > and it's working just fine.
>> > >
>> > >The listInstancesSummary function is already in the Management API. I
>> > just invoked it and found it blocked sometimes when I had many instances
>> > running. After looking into the ODE source code, I find that the function
>> > listAllInstances() actually have little difference from
>> > listInstancesSummary(), while the former function have few filters than
>> the
>> > latter.
>> > >
>> > >Have you met with the problem that the Management API be blocked? I
>> think
>> > it may be caused by some transaction problems.
>> > >
>> > >Regards,
>> > >Vito
>> > >
>> > >
>> > >
>> > >
>> > >
>> > >
>> > >
>> > >
>> > >At 2014-03-25 17:02:24,"Sathwik B P" <sa...@gmail.com> wrote:
>> > >>Vito,
>> > >>
>> > >>Good to know you could solve the issue with Postgresql. It would be
>> great
>> > >>if you could help the community know about the changes you did to make
>> it
>> > >>work :)
>> > >>
>> > >>You might have a look at listAllInstances which will list all the
>> > instances.
>> > >>
>> > >>What is the query that used in listInstancesSummary?
>> > >>
>> > >>regards,
>> > >>sathwik
>> > >>
>> > >>
>> > >>On Tue, Mar 25, 2014 at 11:33 AM, Vito <te...@163.com> wrote:
>> > >>
>> > >>> Hi all,
>> > >>> Thank you for your kindly help these days. I just make ODE up and
>> > running
>> > >>> with mysql and postgresql finally! But when I try some experiments,
>> > other
>> > >>> problems seem to come up. Actually, it may not be called an error
>> > because
>> > >>> there is no error or exception.
>> > >>> The problem came out when I ran multiple instances. It met no problem
>> > >>> until I try to list instances summary from ODE Management API to get
>> > the
>> > >>> instance number in the engine. The list function returned no results
>> > until
>> > >>> all the instances ran to the end. I think it may be blocked by the
>> > >>> transactions of running instances.
>> > >>> Do you think the transactions really block the management api
>> > function? If
>> > >>> this really happens, how can I get the number of instance number of
>> the
>> > >>> current engine?
>> > >>> Best,
>> > >>> Vito
>> > >>>
>> >
>>
>>
>>
>> --
>> Tammo van Lessen - http://www.taval.de
>>