GetJMS Queue Processor not releasing connections

classic Classic list List threaded Threaded
8 messages Options
Reply | Threaded
Open this post in threaded view
|

GetJMS Queue Processor not releasing connections

Rob Weiss
All,
What is the expectation of the connection state when the processor is not
in a running state? We have seen that the connection does not release and
has to be forced to release by shutting the NiFi process down. This is
causing a major headache when trying to consume the queue via another
process (either within NiFi or outside of NiFi).

Thoughts?

Thanks,
Rob.
Reply | Threaded
Open this post in threaded view
|

Re: GetJMS Queue Processor not releasing connections

Matt Gilman
Rob,

It looks as though the connection is closed when the processor is stopped.
If that is not the behavior you're seeing can you check the logs for any
warnings that contain the message

"unable to close connection to JMS Server due to "

followed by a more detailed explanation. Thanks.

Matt

On Mon, May 4, 2015 at 11:59 AM, Rob Weiss <[hidden email]> wrote:

> All,
> What is the expectation of the connection state when the processor is not
> in a running state? We have seen that the connection does not release and
> has to be forced to release by shutting the NiFi process down. This is
> causing a major headache when trying to consume the queue via another
> process (either within NiFi or outside of NiFi).
>
> Thoughts?
>
> Thanks,
> Rob.
>
Reply | Threaded
Open this post in threaded view
|

Re: GetJMS Queue Processor not releasing connections

Mark Payne
In reply to this post by Rob Weiss
Rob,


The connection should be closed. Do you see any of the following statements in your logs:​


unable to close connection to JMS Server due to


If so, there should be an Exception there indicating the issue.


Thanks

-Mark








From: Rob Weiss
Sent: ‎Monday‎, ‎May‎ ‎4‎, ‎2015 ‎12‎:‎00‎ ‎PM
To: [hidden email]





All,
What is the expectation of the connection state when the processor is not
in a running state? We have seen that the connection does not release and
has to be forced to release by shutting the NiFi process down. This is
causing a major headache when trying to consume the queue via another
process (either within NiFi or outside of NiFi).

Thoughts?

Thanks,
Rob.
Reply | Threaded
Open this post in threaded view
|

Re: GetJMS Queue Processor not releasing connections

Rob Weiss
In reply to this post by Matt Gilman
Matt,
I am not seeing any error messages at all related to the JMS server
connection. I am also not seeing and connection tear down messages. This is
really odd.

It seems fairly easy to replicate, just add a GetJMS processor, then remove
it then add another one. I am using ActiveMQ as the broker.

On Mon, May 4, 2015 at 12:05 PM, Matt Gilman <[hidden email]>
wrote:

> Rob,
>
> It looks as though the connection is closed when the processor is stopped.
> If that is not the behavior you're seeing can you check the logs for any
> warnings that contain the message
>
> "unable to close connection to JMS Server due to "
>
> followed by a more detailed explanation. Thanks.
>
> Matt
>
> On Mon, May 4, 2015 at 11:59 AM, Rob Weiss <[hidden email]> wrote:
>
> > All,
> > What is the expectation of the connection state when the processor is not
> > in a running state? We have seen that the connection does not release and
> > has to be forced to release by shutting the NiFi process down. This is
> > causing a major headache when trying to consume the queue via another
> > process (either within NiFi or outside of NiFi).
> >
> > Thoughts?
> >
> > Thanks,
> > Rob.
> >
>
Reply | Threaded
Open this post in threaded view
|

Re: GetJMS Queue Processor not releasing connections

Matt Gilman
Rob,

I was able to replicate the behavior your seeing locally. We are in the
process of cutting a release but I think we can get this addressed and
included.

Matt

On Mon, May 4, 2015 at 12:16 PM, Rob Weiss <[hidden email]> wrote:

> Matt,
> I am not seeing any error messages at all related to the JMS server
> connection. I am also not seeing and connection tear down messages. This is
> really odd.
>
> It seems fairly easy to replicate, just add a GetJMS processor, then remove
> it then add another one. I am using ActiveMQ as the broker.
>
> On Mon, May 4, 2015 at 12:05 PM, Matt Gilman <[hidden email]>
> wrote:
>
> > Rob,
> >
> > It looks as though the connection is closed when the processor is
> stopped.
> > If that is not the behavior you're seeing can you check the logs for any
> > warnings that contain the message
> >
> > "unable to close connection to JMS Server due to "
> >
> > followed by a more detailed explanation. Thanks.
> >
> > Matt
> >
> > On Mon, May 4, 2015 at 11:59 AM, Rob Weiss <[hidden email]> wrote:
> >
> > > All,
> > > What is the expectation of the connection state when the processor is
> not
> > > in a running state? We have seen that the connection does not release
> and
> > > has to be forced to release by shutting the NiFi process down. This is
> > > causing a major headache when trying to consume the queue via another
> > > process (either within NiFi or outside of NiFi).
> > >
> > > Thoughts?
> > >
> > > Thanks,
> > > Rob.
> > >
> >
>
Reply | Threaded
Open this post in threaded view
|

Re: GetJMS Queue Processor not releasing connections

Rob Weiss
Matt,
Thanks! Glad I am not crazy!

On Mon, May 4, 2015 at 1:06 PM, Matt Gilman <[hidden email]> wrote:

> Rob,
>
> I was able to replicate the behavior your seeing locally. We are in the
> process of cutting a release but I think we can get this addressed and
> included.
>
> Matt
>
> On Mon, May 4, 2015 at 12:16 PM, Rob Weiss <[hidden email]> wrote:
>
> > Matt,
> > I am not seeing any error messages at all related to the JMS server
> > connection. I am also not seeing and connection tear down messages. This
> is
> > really odd.
> >
> > It seems fairly easy to replicate, just add a GetJMS processor, then
> remove
> > it then add another one. I am using ActiveMQ as the broker.
> >
> > On Mon, May 4, 2015 at 12:05 PM, Matt Gilman <[hidden email]>
> > wrote:
> >
> > > Rob,
> > >
> > > It looks as though the connection is closed when the processor is
> > stopped.
> > > If that is not the behavior you're seeing can you check the logs for
> any
> > > warnings that contain the message
> > >
> > > "unable to close connection to JMS Server due to "
> > >
> > > followed by a more detailed explanation. Thanks.
> > >
> > > Matt
> > >
> > > On Mon, May 4, 2015 at 11:59 AM, Rob Weiss <[hidden email]>
> wrote:
> > >
> > > > All,
> > > > What is the expectation of the connection state when the processor is
> > not
> > > > in a running state? We have seen that the connection does not release
> > and
> > > > has to be forced to release by shutting the NiFi process down. This
> is
> > > > causing a major headache when trying to consume the queue via another
> > > > process (either within NiFi or outside of NiFi).
> > > >
> > > > Thoughts?
> > > >
> > > > Thanks,
> > > > Rob.
> > > >
> > >
> >
>
Reply | Threaded
Open this post in threaded view
|

Re: GetJMS Queue Processor not releasing connections

Matt Gilman
Rob,

The fix for the issue you were seeing has been committed to the develop
branch. I'm pretty sure based off that behavior you were already running
0.1.0-incubating-SNAPSHOT. The release should be put together today and
then it has to go through the normal voting process making it available end
of this week/early next week. If you want to address the issues your seeing
before then, you could build another SNAPSHOT locally. Thanks.

Matt

On Mon, May 4, 2015 at 2:05 PM, Rob Weiss <[hidden email]> wrote:

> Matt,
> Thanks! Glad I am not crazy!
>
> On Mon, May 4, 2015 at 1:06 PM, Matt Gilman <[hidden email]>
> wrote:
>
> > Rob,
> >
> > I was able to replicate the behavior your seeing locally. We are in the
> > process of cutting a release but I think we can get this addressed and
> > included.
> >
> > Matt
> >
> > On Mon, May 4, 2015 at 12:16 PM, Rob Weiss <[hidden email]> wrote:
> >
> > > Matt,
> > > I am not seeing any error messages at all related to the JMS server
> > > connection. I am also not seeing and connection tear down messages.
> This
> > is
> > > really odd.
> > >
> > > It seems fairly easy to replicate, just add a GetJMS processor, then
> > remove
> > > it then add another one. I am using ActiveMQ as the broker.
> > >
> > > On Mon, May 4, 2015 at 12:05 PM, Matt Gilman <[hidden email]>
> > > wrote:
> > >
> > > > Rob,
> > > >
> > > > It looks as though the connection is closed when the processor is
> > > stopped.
> > > > If that is not the behavior you're seeing can you check the logs for
> > any
> > > > warnings that contain the message
> > > >
> > > > "unable to close connection to JMS Server due to "
> > > >
> > > > followed by a more detailed explanation. Thanks.
> > > >
> > > > Matt
> > > >
> > > > On Mon, May 4, 2015 at 11:59 AM, Rob Weiss <[hidden email]>
> > wrote:
> > > >
> > > > > All,
> > > > > What is the expectation of the connection state when the processor
> is
> > > not
> > > > > in a running state? We have seen that the connection does not
> release
> > > and
> > > > > has to be forced to release by shutting the NiFi process down. This
> > is
> > > > > causing a major headache when trying to consume the queue via
> another
> > > > > process (either within NiFi or outside of NiFi).
> > > > >
> > > > > Thoughts?
> > > > >
> > > > > Thanks,
> > > > > Rob.
> > > > >
> > > >
> > >
> >
>
Reply | Threaded
Open this post in threaded view
|

Re: GetJMS Queue Processor not releasing connections

Rob Weiss
Matt, Thanks.

On Tue, May 5, 2015 at 8:33 AM, Matt Gilman <[hidden email]> wrote:

> Rob,
>
> The fix for the issue you were seeing has been committed to the develop
> branch. I'm pretty sure based off that behavior you were already running
> 0.1.0-incubating-SNAPSHOT. The release should be put together today and
> then it has to go through the normal voting process making it available end
> of this week/early next week. If you want to address the issues your seeing
> before then, you could build another SNAPSHOT locally. Thanks.
>
> Matt
>
> On Mon, May 4, 2015 at 2:05 PM, Rob Weiss <[hidden email]> wrote:
>
> > Matt,
> > Thanks! Glad I am not crazy!
> >
> > On Mon, May 4, 2015 at 1:06 PM, Matt Gilman <[hidden email]>
> > wrote:
> >
> > > Rob,
> > >
> > > I was able to replicate the behavior your seeing locally. We are in the
> > > process of cutting a release but I think we can get this addressed and
> > > included.
> > >
> > > Matt
> > >
> > > On Mon, May 4, 2015 at 12:16 PM, Rob Weiss <[hidden email]>
> wrote:
> > >
> > > > Matt,
> > > > I am not seeing any error messages at all related to the JMS server
> > > > connection. I am also not seeing and connection tear down messages.
> > This
> > > is
> > > > really odd.
> > > >
> > > > It seems fairly easy to replicate, just add a GetJMS processor, then
> > > remove
> > > > it then add another one. I am using ActiveMQ as the broker.
> > > >
> > > > On Mon, May 4, 2015 at 12:05 PM, Matt Gilman <
> [hidden email]>
> > > > wrote:
> > > >
> > > > > Rob,
> > > > >
> > > > > It looks as though the connection is closed when the processor is
> > > > stopped.
> > > > > If that is not the behavior you're seeing can you check the logs
> for
> > > any
> > > > > warnings that contain the message
> > > > >
> > > > > "unable to close connection to JMS Server due to "
> > > > >
> > > > > followed by a more detailed explanation. Thanks.
> > > > >
> > > > > Matt
> > > > >
> > > > > On Mon, May 4, 2015 at 11:59 AM, Rob Weiss <[hidden email]>
> > > wrote:
> > > > >
> > > > > > All,
> > > > > > What is the expectation of the connection state when the
> processor
> > is
> > > > not
> > > > > > in a running state? We have seen that the connection does not
> > release
> > > > and
> > > > > > has to be forced to release by shutting the NiFi process down.
> This
> > > is
> > > > > > causing a major headache when trying to consume the queue via
> > another
> > > > > > process (either within NiFi or outside of NiFi).
> > > > > >
> > > > > > Thoughts?
> > > > > >
> > > > > > Thanks,
> > > > > > Rob.
> > > > > >
> > > > >
> > > >
> > >
> >
>