Ed, I would replace only the latter. I may be able to test it out tonight and give you some feedback.
John On Thu, Jul 21, 2016 at 3:30 PM Ed Kaltenbach <ekaltenb...@ara.com> wrote: > Does anybody know the steps to replace 1.1 with 1.3 in Wildfly 10? > > I see that Wildfly has a directory > "wildfly-10.0.0.Final\modules\system\layers\base\org\wildfly\extension\messaging-activemq\main" > that contains > - artemis-wildfly-integration-1.0.2.jar > - module.xml > - wildfly-messaging-activemq-10.0.0.Final.jar > > Wildfly also has the directory > "wildfly-10.0.0.Final\modules\system\layers\base\org\apache\activemq\artemis" > that contains subdirectories: > - "main", > - "protocol", > - "ra". > I see a lot of artemis-xxx.jar files in all of these subdirectories. > > Does anybody have any suggestions or inputs? > > Thanks, > Ed > > -----Original Message----- > From: John D. Ament [mailto:johndam...@apache.org] > Sent: Thursday, July 21, 2016 12:51 PM > To: dev@activemq.apache.org > Subject: Re: STOMP server quits sending to all subscribers when one client > disconnects > > I see that wf10 ships a custom build for 1.1, was there anything special > about it? > > On Jul 21, 2016 14:09, "Justin Bertram" <jbert...@apache.com> wrote: > > > I'm pretty confident it would work, but you never know until you try > > (which I haven't). > > > > > > Justin > > > > ----- Original Message ----- > > From: "Ed Kaltenbach" <ekaltenb...@ara.com> > > To: dev@activemq.apache.org > > Sent: Thursday, July 21, 2016 12:56:12 PM > > Subject: RE: FW: STOMP server quits sending to all subscribers when > > one client disconnects > > > > John, > > I had some time this morning to try against the 1.3 standalone broker. > > The problem seems to be fixed in 1.3. I first tried two clients, each > > with a unique subscription ID, and could not replicate the error. > > When one client ended, the other client still posted and received > > messages from the topic. > > I also tried it with both clients using the same subscription ID. I > > could not replicate the error here either. When one client ended, the > > other continued to send and receive messages. > > I even killed one of the clients abruptly and the other one continued > > to send and receive messages. > > > > So, in summary, the problem seems to be fixed in 1.3. > > > > How confident are people that Artemis 1.3 will work in Wildfly 10 > > seamlessly? > > > > Ed > > > > -----Original Message----- > > From: John D. Ament [mailto:johndam...@apache.org] > > Sent: Tuesday, July 19, 2016 5:22 PM > > To: dev@activemq.apache.org > > Subject: Re: FW: STOMP server quits sending to all subscribers when > > one client disconnects > > > > Ed, > > > > Sorry one more thing to try. Can you try against the 1.3 standalone > > broker instead of the 1.1 that ships w/ Wildfly? I'm wondering if the > > auto creation feature fixes this error. > > > > John > > > > On Tue, Jul 19, 2016 at 11:35 AM Ed Kaltenbach <ekaltenb...@ara.com> > > wrote: > > > > > > > > > > > I just modified my test client application so that each client has a > > > unique subscription id. Here is the new code: > > > > > > > > > > > > String destID = String.format("%d", System.currentTimeMillis()); > > > > > > msg = "SUBSCRIBE\n"; > > > > > > msg = msg + "destination:" + topicName + "\n"; > > > > > > msg = msg + "id:" + destID + "\n"; > > > > > > msg = msg + "ack:auto\n"; > > > > > > msg = msg + "\n"; > > > > > > msg = msg + '\0'; > > > > > > > > > > > > I still see the same error. When one of the clients ends, the other > > > clients start getting the “AMQ339001\c Destination does not exist\c > > > jms.topic.ACRS_Exit” error when they try to SEND a message to the > > > JMS > > topic. > > > > > > > > > > > > It all seems to work fine until one of the clients UNSUBSCRIBES, > > > DISCONNECTS, and shutdowns the socket. All of the clients were > > > receiving all of the messages. > > > > > > > > > > > > Here is some new information. If I run multiple instances of my > > > test client application (the new one that has unique subscription > > > IDs for each > > > client) and then I kill one using “ctrl-c” then I see the same error. > > > The other client instance starts getting the “AMQ339001\c > > > Destination does not exist\c jms.topic.ACRS_Exit” error when it > > > tries to SEND a message to the JMS topic. Therefore, I don’t think > > > the problem is related to the “UNSUBSCRIBE” or “DISCONNECT” messages > > > because they were never sent when the problem started. > > > > > > > > > > > > Ed > > > > > > > > > > > > *From:* Martyn Taylor [mailto:mtay...@redhat.com > > > <mtay...@redhat.com>] > > > *Sent:* Tuesday, July 19, 2016 9:00 AM > > > *To:* dev@activemq.apache.org > > > *Cc:* Ed Kaltenbach <ekaltenb...@ara.com> > > > > > > > > > *Subject:* Re: STOMP server quits sending to all subscribers when > > > one client disconnects > > > > > > > > > > > > Hi Ed, > > > > > > > > > > > > You mentioned that using a unique subscription ID does not resolve > > > this issue. Can you confirm that using different subscription IDs > > > across all your clients is the same? Were you seeing subscription > > > semantics before the error (i.e. was every subscription seeing every > > message?). > > > > > > > > > > > > I've taken a quick look and there may be an issue in the STOMP > > > protocol handler. The subscription ID is used to identify consumer > > > queues, this could cause some issues with multiple clients using the > > > same subscription ID during unsubscribe. The subscription ID only > > > needs to be unique within connections > > > https://stomp.github.io/stomp-specification-1.2.html#SUBSCRIBE > > > > > > > > > > > > Thanks > > > Martyn > > > > > >
