Mailing-List: contact users-help@activemq.apache.org; run by ezmlm
Precedence: bulk
Reply-To: users@activemq.apache.org
From: Clebert <clebert.suconic@gmail.com>
Content-Type: text/plain;
	charset=us-ascii
Content-Transfer-Encoding: quoted-printable
Mime-Version: 1.0 (1.0)
Subject: Re: [Artemis] org.apache.activemq.transport.InactivityIOException:
 Channel was inactive for too (>30000) long: tcp://127.0.0.1:61616
Message-Id: <F2CE7210-21DB-4108-A4A8-7915A295EDEA@gmail.com>
Date: Wed, 21 Oct 2015 00:13:28 -0400
References: <1444664740445-4702833.post@n4.nabble.com>
 <1445354611304-4703169.post@n4.nabble.com>
 <CAPVVMa_ReEJJ1D3nE5Ud6dpKhXB_H1D1orhCms12_Qq2qW+NdA@mail.gmail.com>
In-Reply-To: 
 <CAPVVMa_ReEJJ1D3nE5Ud6dpKhXB_H1D1orhCms12_Qq2qW+NdA@mail.gmail.com>
To: users@activemq.apache.org

I am teaming with Howard gao around failover on open wire protocol manager f=
or Artemis.  So I don't know how well it would work now with open wire and f=
ailover.=20

You should see fixes around this over the next week and this week.=20

-- Clebert Suconic typing on the iPhone.=20

> On Oct 21, 2015, at 00:06, Tim Bain <tbain@alumni.duke.edu> wrote:
>=20
> I believe the problem with your consumers is that you haven't set
> maxReconnectAttempts=3D0 on the failover transport, which is necessary whe=
n
> using failover within static in a networkConnector because it allows the
> network bridge to detect the connection failure and restore properly.
> Without it, the failover protocol eats the failure and the necessary
> handshaking isn't done.
>=20
> None of that relates to your observations that this seemed to work in 5.11=

> and seems not to in 5.12, though.  Can you isolate the behavior into a
> small, repeatable scenario that you could attach to a JIRA bug report?
> (Better still, can you demonstrate it in a unit test?)
>=20
> Tim
>> On Oct 20, 2015 12:45 PM, "jochenw" <jochen.walz.mail@googlemail.com> wro=
te:
>>=20
>> Hello,
>>=20
>> I've got a very similar problem with ActiveMQ 5.12.0 when building a brok=
er
>> network via a SSL connector. I have set
>>=20
>> networkConnector
>>=20
>> uri=3D"static:(failover:(ssl://${offboardBroker1}:10017?wireFormat.maxIna=
ctivityDuration=3D0,ssl://${offboardBroker2}:10017?wireFormat.maxInactivityD=
uration=3D0)?randomize=3Dtrue)"
>> ...
>>=20
>> in the network connector definition. However, after startup an exception
>> (java.io.EOFexception at
>> java.io.DataInputStream.readInt(DataInputStream.java:392) is thrown, and
>> the
>> broker on the other side logs that the network connection has been
>> interrupted due to inactivity although the maxInactivityDuration was set t=
o
>> zero, and the monitor should therefore be activated.
>>=20
>> The connection is re-established, but after that the consumers of my brok=
er
>> cannot be seen on the other side any longer.
>>=20
>> With ActiveMQ 5.11.1, the setting maxInactivityDuration=3D0 worked, and t=
he
>> connection is never timed out. Seems to be different in 5.12.0.
>>=20
>> Of course, I would also be interested if anybody has an idea why my
>> consumers are no longer seen on the broker connected via the network
>> connector after a reconnection. With "normal" network disruption, this
>> works, but after the exception, it doesn't.
>>=20
>> Best Regards,
>> Jochen
>>=20
>>=20
>>=20
>> --
>> View this message in context:
>> http://activemq.2283324.n4.nabble.com/Artemis-org-apache-activemq-transpo=
rt-InactivityIOException-Channel-was-inactive-for-too-30000-long-6-tp4702833=
p4703169.html
>> Sent from the ActiveMQ - User mailing list archive at Nabble.com.
>>=20