But you go throw the HTTP connector right ?
Not tcp.

So, do you have maxFrameSize on the http connector ? And also, do you use a 
jetty.xml dedicated for the http connector (it’s not the same as conf/jetty.xml 
used by the admin web console).

Regards
JB

> Le 15 juin 2021 à 14:58, Andrew Pomponio <apompo...@perforce.com> a écrit :
> 
> Yes, the message size that seems to start the issue is 65,536 b or 65k in 
> size. Smaller than that and the issue does not occur, we tried with a 32k 
> message and did not have issues.
> 
> On 6/15/21, 6:37 AM, "Jean-Baptiste Onofre" <j...@nanthrax.net> wrote:
> 
>    OK, and the message you sent is smaller than 104857600 b ?
> 
>    Anyway, if the message is greater than the maxFrameSize, you have a 
> message in the activemq log and an exception on the client side 
> (JMSException).
> 
>    Regards
>    JB
> 
>> Le 15 juin 2021 à 13:53, Andrew Pomponio <apompo...@perforce.com> a écrit :
>> 
>> Hello Jean-Baptiste, thank you so much for your reply,
>> 
>> For testing, we've been using stock defaults, for example, in our test for 
>> 5.15.14 we used maxFrameSize=104857600.
>> 
>> On 6/15/21, 5:48 AM, "Jean-Baptiste Onofre" <j...@nanthrax.net> wrote:
>> 
>>   Hi,
>> 
>>   Sorry I missed the message.
>> 
>>   Maybe you have reach the maxFrameSize on the transport connector. Did you 
>> try to increase it ?
>> 
>>   What’s your transport connector configuration in activemq.xml ?
>> 
>>   Regards
>>   JB
>> 
>>> Le 15 juin 2021 à 13:35, Andrew Pomponio <apompo...@perforce.com> a écrit :
>>> 
>>> Dear ActiveMQ Community,
>>> 
>>> An update on the issue, we've done continued testing and found that this 
>>> issue starts as far back as 5.15.14 and is also present in 5.16.2.
>>> 
>>> On 6/10/21, 8:57 AM, "Andrew Pomponio" <apompo...@perforce.com> wrote:
>>> 
>>>  Dear ActiveMQ Community,
>>> 
>>>  I am reaching out on behalf of a customer of ours experiencing what 
>>> appears to be a bug (or new default limitation) in an upgraded version of 
>>> AMQ from the version they are currently using. We have tested this behavior 
>>> in our lab environment and are able to replicate what they are seeing using 
>>> the following environmental components;
>>> 
>>>  ActiveMQ 5.15.11 (works) and 5.16.1 (does not work)
>>>  Postman 8.6.1
>>>  72k sized payload (the exact limitations appears to be 65536 bytes)
>>> 
>>>  The following test is to be run once on 5.15 and then on 5.16.
>>> 
>>>  First, create a payload of random gibberish by executing the following;
>>> 
>>>  dd bs=1024 count=71 </dev/random >payload-test72k.txt
>>> 
>>>  cat the file and copy the contents (gibberish) into the <ns2:Data> tags 
>>> like so;
>>> 
>>>  <ns2:NormalizedMessage xmlns:ns2="http://private"; 
>>> xmlns:ns4="http://differentprivate";><ns2:BaseDescriptor><ns2:MsgID>AVSN#I64534#ajz123403252016#2016-03-25T13:15:55</ns2:MsgID><ns2:MsgType>INTF</ns2:MsgType><ns2:MsgMode>INBOUND</ns2:MsgMode><ns2:SourceTimeStamp>2021-04-28T15:28:57.294Z</ns2:SourceTimeStamp><ns2:SourceApplicationID>over
>>>  9000</ns2:SourceApplicationID><ns2:SourceUserID>over 
>>> 9000</ns2:SourceUserID><ns2:SequenceTimeStamp>2021-04-28T15:28:57.294Z</ns2:SequenceTimeStamp>
>>>  
>>> </ns2:BaseDescriptor><ns2:RoutingDescriptor><ns4:RoutingInfo><ns4:ApplicationName>over
>>>  9000</ns4:ApplicationName><ns4:ApplicationCode>over 
>>> 9000</ns4:ApplicationCode><ns4:ApplicationVersion>1.0</ns4:ApplicationVersion><ns4:PublishTimeStamp>2021-04-28T15:28:57.294Z</ns4:PublishTimeStamp><ns4:Status>true</ns4:Status><ns4:StatusDescription>DocumentUploadRequest</ns4:StatusDescription>
>>>  </ns4:RoutingInfo> 
>>> </ns2:RoutingDescriptor><ns2:lolDescriptor><ns2:lolReferenceID>MEDDOCINT042803</ns2:lolReferenceID><ns2:AliaslolReferenceIDs><ns2:lolReferenceID>MEDDOCINT042803</ns2:lolReferenceID>
>>>  
>>> </ns2:AliaslolReferenceIDs><ns2:lolFolderID>0</ns2:lolFolderID><ns2:ClientFileID></ns2:ClientFileID><ns2:nopeCompany><ns2:VantiveCode>QO</ns2:VantiveCode><ns2:nahbroCustID>0</ns2:nahbroCustID>
>>>  </ns2:nopeCompany> </ns2:lolDescriptor><ns2:Payload><ns2:Data>[…] 
>>> �M��pt=��/�s
>>>  .7u�a��F�{�S���Zt��R�^&�(/k�*!y���\��}I�i
>>>                    ��
>>>  �؍I�����ՙ�y1
>>>  ��2��˧v�i�
>>>            ���-F�ʅ�"��5�ԍ;C�+V[…]</ns2:Data> </ns2:Payload> 
>>> </ns2:NormalizedMessage>
>>> 
>>>  Copy the above XML into the raw body of Postman. The XML and payload you 
>>> copy and pasted from payload-test72k.txt is going to 
>>> http://localhost:8161/api/message/orders.input/. With ActiveMQ running in 
>>> debug mode, using Postman, with basic authorization login, click POST. You 
>>> should receive an output of message sent, though the message will not show 
>>> up in the UI of AMQ. We do not need it to for the purposes of this 
>>> experiment.
>>> 
>>>  Back in a terminal from the bin directory in AMQ;
>>> 
>>>  tail -200 ../data/activemq.log
>>> 
>>>  You then can see in the log entries two different examples of the message, 
>>> one that shows the full body of text being sent to ActiveMQ, and another 
>>> that shows a condensed summary of the message being sent.
>>> 
>>>  2021-06-10 07:50:44,510 | DEBUG | Sent! to destination: 
>>> topic://orders.input. message: ActiveMQTextMessage {commandId = 0, 
>>> responseRequired = false, messageId = 
>>> ID:apomponio-56497-1623331633553-4:1:1:1:6, originalDestination = null, 
>>> originalTransactionId = null, producerId = null, destination = 
>>> topic://orders.input., transactionId = null, expiration = 0, timestamp = 
>>> 1623333044509, arrival = 0, brokerInTime = 0, brokerOutTime = 0, 
>>> correlationId = null, replyTo = null, persistent = true, type = null, 
>>> priority = 5, groupID = null, groupSequence = 0, targetConsumerId = null, 
>>> compressed = false, userID = null, content = null, marshalledProperties = 
>>> null, dataStructure = null, redeliveryCounter = 0, size = 0, properties = 
>>> null, readOnlyProperties = false, readOnlyBody = false, droppable = false, 
>>> jmsXGroupFirstForConsumer = false, text = <ns2:NormalizedMessage 
>>> xmlns:ns2="http://priv...�T�T�Z5� �]
>>>  } | org.apache.activemq.web.WebClient | qtp632475595-35
>>> 
>>>  In the example above from the 5.16 server, we see in the text = field the 
>>> start of our message, and right after http://priv we see … followed by the 
>>> portion of gibberish where the message ends in the log. In our 5.15 logs we 
>>> see http://priv...zedMessage> showing us the partial ending XML tag of 
>>> </ns2:NormalizedMessage> which is the expected behavior.
>>> 
>>>  In 5.15.11, the full length of the body message shows up in the 
>>> activemq.log file, but in 5.16.1 that same message sent exactly the same 
>>> way is truncated around the 64k size.
>>> 
>>>  Is this a known default limitation introduced in 5.16? If so, how do we 
>>> increase it?
>>> 
>>>  If this is anticipated to be a bug, what additional information, if any, 
>>> do you need from us? Also, in the future, would something like this be more 
>>> appropriate for the users mailing list instead?
>>> 
>>> 
>>> 
>>> This e-mail may contain information that is privileged or confidential. If 
>>> you are not the intended recipient, please delete the e-mail and any 
>>> attachments and notify us immediately.
>>> 
>> 
>> 
>> 
>>   CAUTION: This email originated from outside of the organization. Do not 
>> click on links or open attachments unless you recognize the sender and know 
>> the content is safe.
>> 
>> 
>> This e-mail may contain information that is privileged or confidential. If 
>> you are not the intended recipient, please delete the e-mail and any 
>> attachments and notify us immediately.
>> 
> 
> 
> 
>    CAUTION: This email originated from outside of the organization. Do not 
> click on links or open attachments unless you recognize the sender and know 
> the content is safe.
> 
> 
> This e-mail may contain information that is privileged or confidential. If 
> you are not the intended recipient, please delete the e-mail and any 
> attachments and notify us immediately.
> 

Reply via email to