Locking/Freezing

[David Shere]

We have a local Java service that imports legacy data from our old
ordering system.  Last night while working on a 130,000 record import,
it was sailing along until it got to 118392, at which time the Ofbiz
Java process jumped to 99%, and the import froze.  No messages from the
service, no notice that the service had stopped.  The load average
hovered around 1.0, and Ofbiz was otherwise working normally.

I'm glad to see that what appears to be a runaway process/infinite loop
didn't stop Ofbiz completely, and was "contained". I stopped and
restarted Ofbiz, the problem went away, and the import picked up where
it left off.  118392 was imported properly.  

In this case, I presume could also have stopped the service to fix the
problem, but that's only because I knew which one it was that was maxing
out the CPU usage.  Is it possible/rare/common that something like this
could happen, and I wouldn't know which service it was?  And in that
case would I have other options than stopping and restarting Ofbiz?

--
David Shere
Information Technology Services
Steele Rubber Products
704-483-9343 x277

[David E Jones]

David,

I have seen this sort of thing happen before, but the trick is there  
are a lot of things that can cause it. One common one is the JDBC  
driver... which database and JDBC driver are you using?

-David J.


On Feb 21, 2007, at 7:05 AM, David Shere wrote:

> We have a local Java service that imports legacy data from our old
> ordering system.  Last night while working on a 130,000 record import,
> it was sailing along until it got to 118392, at which time the Ofbiz
> Java process jumped to 99%, and the import froze.  No messages from  
> the
> service, no notice that the service had stopped.  The load average
> hovered around 1.0, and Ofbiz was otherwise working normally.
>
> I'm glad to see that what appears to be a runaway process/infinite  
> loop
> didn't stop Ofbiz completely, and was "contained". I stopped and
> restarted Ofbiz, the problem went away, and the import picked up where
> it left off.  118392 was imported properly.
>
> In this case, I presume could also have stopped the service to fix the
> problem, but that's only because I knew which one it was that was  
> maxing
> out the CPU usage.  Is it possible/rare/common that something like  
> this
> could happen, and I wouldn't know which service it was?  And in that
> case would I have other options than stopping and restarting Ofbiz?
>
> --
> David Shere
> Information Technology Services
> Steele Rubber Products
> 704-483-9343 x277
>

[David Shere]

I'm using PostgreSQL version 8.2.1 with the driver
postgresql-8.2-504.jdbc3.jar.

On Wed, 2007-02-21 at 17:18 -0700, David E. Jones wrote:

> David,
>
> I have seen this sort of thing happen before, but the trick is there  
> are a lot of things that can cause it. One common one is the JDBC  
> driver... which database and JDBC driver are you using?
>
> -David J.
>
>
> On Feb 21, 2007, at 7:05 AM, David Shere wrote:
>
> > We have a local Java service that imports legacy data from our old
> > ordering system.  Last night while working on a 130,000 record import,
> > it was sailing along until it got to 118392, at which time the Ofbiz
> > Java process jumped to 99%, and the import froze.  No messages from  
> > the
> > service, no notice that the service had stopped.  The load average
> > hovered around 1.0, and Ofbiz was otherwise working normally.
> >
> > I'm glad to see that what appears to be a runaway process/infinite  
> > loop
> > didn't stop Ofbiz completely, and was "contained". I stopped and
> > restarted Ofbiz, the problem went away, and the import picked up where
> > it left off.  118392 was imported properly.
> >
> > In this case, I presume could also have stopped the service to fix the
> > problem, but that's only because I knew which one it was that was  
> > maxing
> > out the CPU usage.  Is it possible/rare/common that something like  
> > this
> > could happen, and I wouldn't know which service it was?  And in that
> > case would I have other options than stopping and restarting Ofbiz?
> >
> > --
> > David Shere
> > Information Technology Services
> > Steele Rubber Products
> > 704-483-9343 x277
> >
>

--
David Shere
Information Technology Services
Steele Rubber Products
704-483-9343 x277

[PRONZATO Cedric RD-BIZZ-GRE]

Hi,

Couldn't it be a VM memory problem? Did you also track (as cpu usage) the memory usage of OFBiz? What are your vm parameters for running OFBiz application?
Perhaps if you give out a little bit more in formations on how you are achieving your imports, someone will be able to provide a more accurate answer.

Bye,
Cédric

-----Message d'origine-----
De : David Shere [mailto:[hidden email]]
Envoyé : mercredi 21 février 2007 15:05
À : [hidden email]
Objet : Locking/Freezing

We have a local Java service that imports legacy data from our old ordering system.  Last night while working on a 130,000 record import, it was sailing along until it got to 118392, at which time the Ofbiz Java process jumped to 99%, and the import froze.  No messages from the service, no notice that the service had stopped.  The load average hovered around 1.0, and Ofbiz was otherwise working normally.

I'm glad to see that what appears to be a runaway process/infinite loop didn't stop Ofbiz completely, and was "contained". I stopped and restarted Ofbiz, the problem went away, and the import picked up where it left off.  118392 was imported properly.  

In this case, I presume could also have stopped the service to fix the problem, but that's only because I knew which one it was that was maxing out the CPU usage.  Is it possible/rare/common that something like this could happen, and I wouldn't know which service it was?  And in that case would I have other options than stopping and restarting Ofbiz?

--
David Shere
Information Technology Services
Steele Rubber Products
704-483-9343 x277

[Jacques Le Roux]

Jconsole may used for that

Jacques


> Hi,
>
> Couldn't it be a VM memory problem? Did you also track (as cpu usage) the memory usage of OFBiz? What are your vm parameters for
running OFBiz application?
> Perhaps if you give out a little bit more in formations on how you are achieving your imports, someone will be able to provide a
more accurate answer.

>
> Bye,
> Cédric
>
> -----Message d'origine-----
> De : David Shere [mailto:[hidden email]]
> Envoyé : mercredi 21 février 2007 15:05
> À : [hidden email]
> Objet : Locking/Freezing
>
> We have a local Java service that imports legacy data from our old ordering system.  Last night while working on a 130,000 record

import, it was sailing along until it got to 118392, at which time the Ofbiz Java process jumped to 99%, and the import froze.  No
messages from the service, no notice that the service had stopped.  The load average hovered around 1.0, and Ofbiz was otherwise
working normally.
>
> I'm glad to see that what appears to be a runaway process/infinite loop didn't stop Ofbiz completely, and was "contained". I
stopped and restarted Ofbiz, the problem went away, and the import picked up where it left off.  118392 was imported properly.
>
> In this case, I presume could also have stopped the service to fix the problem, but that's only because I knew which one it was
that was maxing out the CPU usage.  Is it possible/rare/common that something like this could happen, and I wouldn't know which
service it was?  And in that case would I have other options than stopping and restarting Ofbiz?
>
> --
> David Shere
> Information Technology Services
> Steele Rubber Products
> 704-483-9343 x277
>

[David Shere]

My VM settings are at 512 MB.  

Our import uses an intermediate entity.  Our data is imported into it,
and from there a java service parses and manipulates the data before
creating many entries in many tables with the new data.

I'm not sure what my memory usage was at the time this process froze.
I'm using Ubuntu, but ofbiz is run under the root user, which to my
understanding has no memory usage limits.  Gnome was not running at the
time.  

On Thu, 2007-02-22 at 18:27 +0100, PRONZATO Cedric RD-BIZZ-GRE wrote:

> Hi,
>
> Couldn't it be a VM memory problem? Did you also track (as cpu usage) the memory usage of OFBiz? What are your vm parameters for running OFBiz application?
> Perhaps if you give out a little bit more in formations on how you are achieving your imports, someone will be able to provide a more accurate answer.
>
> Bye,
> Cédric
>
> -----Message d'origine-----
> De : David Shere [mailto:[hidden email]]
> Envoyé : mercredi 21 février 2007 15:05
> À : [hidden email]
> Objet : Locking/Freezing
>
> We have a local Java service that imports legacy data from our old ordering system.  Last night while working on a 130,000 record import, it was sailing along until it got to 118392, at which time the Ofbiz Java process jumped to 99%, and the import froze.  No messages from the service, no notice that the service had stopped.  The load average hovered around 1.0, and Ofbiz was otherwise working normally.
>
> I'm glad to see that what appears to be a runaway process/infinite loop didn't stop Ofbiz completely, and was "contained". I stopped and restarted Ofbiz, the problem went away, and the import picked up where it left off.  118392 was imported properly.  
>
> In this case, I presume could also have stopped the service to fix the problem, but that's only because I knew which one it was that was maxing out the CPU usage.  Is it possible/rare/common that something like this could happen, and I wouldn't know which service it was?  And in that case would I have other options than stopping and restarting Ofbiz?
>
> --
> David Shere
> Information Technology Services
> Steele Rubber Products
> 704-483-9343 x277
>
>

--
David Shere
Information Technology Services
Steele Rubber Products
704-483-9343 x277

[Brett]

David,

I have seen this problem too if the service engine is saturated with
failed services that it has to continually retry.  We worked around
these issues by setting the maximum retry count to something like 3.

I have also seen this error when the jdbc driver version is incorrect
for the specified DB.


Brett

On 2/21/07, David Shere <[hidden email]> wrote:

> We have a local Java service that imports legacy data from our old
> ordering system.  Last night while working on a 130,000 record import,
> it was sailing along until it got to 118392, at which time the Ofbiz
> Java process jumped to 99%, and the import froze.  No messages from the
> service, no notice that the service had stopped.  The load average
> hovered around 1.0, and Ofbiz was otherwise working normally.
>
> I'm glad to see that what appears to be a runaway process/infinite loop
> didn't stop Ofbiz completely, and was "contained". I stopped and
> restarted Ofbiz, the problem went away, and the import picked up where
> it left off.  118392 was imported properly.
>
> In this case, I presume could also have stopped the service to fix the
> problem, but that's only because I knew which one it was that was maxing
> out the CPU usage.  Is it possible/rare/common that something like this
> could happen, and I wouldn't know which service it was?  And in that
> case would I have other options than stopping and restarting Ofbiz?
>
> --
> David Shere
> Information Technology Services
> Steele Rubber Products
> 704-483-9343 x277
>
>