I've been attempting to monitor the WSO2 MB 2.1.1 via JMX which appears to be a valid option for other WSO2 tools such as the ESB however the MB does not appear to be updating any of the MBeans for the Queue attributes.
Ex:
org.wso2.andes:type=VirtualHost.Queue,VirtualHost="carbon",name="testQueue"
This has a number of potentially useful attributes like the following:
ConsumerCount
ActiveConsumerCount
ReceivedMessageCount
MessageCount
However the counters always read "0" even when there are messages in the queue dropped in via the sample sender or manually.
Creating a new queue will create the following, but the issue is seen on the new queue as well.
org.wso2.andes:type=VirtualHost.Queue,VirtualHost="carbon",name="testQueue2"
Am I simply looking in the incorrect location or any other variety of user error?
Is this the intentional behavior of the application?
Do you have any suggestions which may assist in getting this data reported via JMX?
Any help would be appreciated.
Thanks
This does not appear to be functional in MB 2.1.1, it looks as if they may be moving to using the CEP or the BAM to provide metrics on some products.
Related
I've just successfully implemented a JMS-message-processor in my WSO2-process. However, when logged in as an admin on the ActiveMQ console, I can view the stats of the queue but I cannot access the contents of the pending message. Instead, I see this error:
Cannot display ObjectMessage body. Reason: Failed to build body from content. Serializable class not available to broker. Reason: java.lang.ClassNotFoundException: org.apache.synapse.message.store.impl.commons.StorableMessage
Accessing the contents of a pending message is possible when using the RabbitMQ, so I figured it should also be possible with this message broker.
I've found solutions online, but they don't seem to be working, or perhaps they're out of date. Can somebody point me to the correct answer to solve this? I'm on WSO2 Integration Studio 8.1.0 and am using ActiveMQ 5.17.1.
It appears that WSO2 is using javax.jms.ObjectMessage despite the fact that there's a long list of reasons why doing so is a bad idea, including the issue you're currently facing, but I digress.
The problem here is that the implementation of the javax.jms.ObjectMessage (i.e. org.apache.synapse.message.store.impl.commons.StorableMessage in this case) must be on the classpath of any application which wants to deserialize that message. This class is not, in fact, on the classpath of ActiveMQ therefore the ActiveMQ web console cannot deserialize the message and display its contents (assuming those contents are human-readable in the first place). That's why a ClassNotFoundException is thrown.
You may be able to resolve the issue by putting org.apache.synapse.message.store.impl.commons.StorableMessage on ActiveMQ's classpath. Aside from that there's really nothing to be done.
I assume this is different for RabbitMQ because in that case WSO2 doesn't use the JMS API and uses the AMQP protocol which is not Java-centric like JMS.
Justin has explained the cause of the issue and as suggested you can try adding the class to ActiveMQ runtime and see whether it resolves the issue. This class is located in the Synapse-Core(synapse-core_2.x.x.wso2vXXX.jar). But remember although this may resolve the ActiveMQ UI issue, this message is not consumable by other systems unless they know how to deserialize it.
Let me add more details on why the message is serialized. Message serialization happens when you use the Store mediator, and when you store a message with Store Mediator it is intended to be only read by a Message Processor. Simply the serialized message can only be consumed by WSO2 Message Processors. In other words, Store Mediator and the Message processors are tightly coupled.
If message serialization is an issue for you. For example, if other systems are consuming the messages that WSO2 publishes you can try using the JMS transport to produce and to consume messages from ActiveMQ. Other than that you can also consider using JMS Inbound Endpoint to consume messages, which all use standard media types when storing the message.
We have deployed API-M 2.1 in a distributed way (each component, GW, TM, KM are running in their own Docker image) on top on DC/OS 1.9 ( Mesos ).
We have issues to get the gateway to enforce throttling policies (should it be subscription tiers or app-level policies). Here is what we have managed to define so far:
The Traffic Manager itself does it job : it receives the event streams, analyzes them on the fly and pushes an event onto the JMS topic throttledata
The Gateway reads the message properly.
So basically we have discarded a communication issue.
However we found two potential issues:
In the event which is pushed to the TM component, the value of the appTenant is null (instead of carbon.super)- We have a single tenant defined.
When the gateway receives the throttling message, it decides to let the message go thinking the "stopOnQuotaReach" is set to false, when it is set to true (we checked the value in the database).
Digging into the source code, we related those two issues to a single source: the value for both values above are read from the authContext and apparently incorrectly set. We are stuck and running out of ideas of things to try and would need some pointers to what could be a potential source of the problem and things to check.
Can somebody help please ?
Thanks- Isabelle.
Is there two TM with HA enabled available in the system?
If the TM is HA enabled, how gateways publish data to TM. Is it load balanced data publishing or failover data publishing to the TMs?
Did you follow below articles to configure the environment with respect to your deployment?
http://wso2.com/library/articles/2016/10/article-scalable-traffic-manager-deployment-patterns-for-wso2-api-manager-part-1/
http://wso2.com/library/articles/2016/10/article-scalable-traffic-manager-deployment-patterns-for-wso2-api-manager-part-2/
Is throttling completely not working in your environment?
Have you noticed any JMS connection related logs in gateways nodes?
In these tests, we have disabled HA to avoid possible complications. Neither subscription nor app throttling policies are working, both because parameters that should have values have not the adequate value (appTenant, stopOnQuotaReach).
Our scenario is far more basic. If we go with one instance of each component, it fails as Isabelle described. And the only thing we know is that both parameters come from the Authentication Context.
Thank you!
I have an WSO2 BPS environment, over default H2 database; since some days ago, my DB is growing without control, until be > 20 Gb. Is any way to avoid this behavior or some commands to compact or clean this file?.
Any assistance would be greatly appreciated.
Execute process cleanup task by configuring scheduler at bps.xml.
Documentation Link1: Process Instance clean Up
When executing BPS, lot of events are generated by default and hence database is size is increased. You can limit the event generation by configuring “Event filtering”. This can be done on “Process Level” and “Scope Level”
Documentation Link2: Refer Filtering Events
While trying to follow the instructions from the wso2am (1.10.0) manual, regarding working with statistics with the wso2das (3.0.1) server i have encountered a problem.
If i choose to let the wso2am server define the stream while making the first call of the api, the wso2das server refuses to post statistics to the WSO2_STATS_DB.
If on the other hand i choose to import the analytics.car file in wso2das (as stated here ) i get an exception (AsyncDataPublisher Stream definition already exist) because the org.wso2.apimgt.statistics.request defined in the latest Analytics.car is different to the one being send from wso2am.
I pinpointed the problem in the definition of the Eventstream_request_1.0 in files
org.wso2.apimgt.statistics.request_1.0.0.json ,
throttledOutORG_WSO2_APIMGT_STATISTICS_REQUEST.xml
where the definition of the throttledOut option is missing
Is there a way to solve this issue?
Thank you.
I think your DAS is in some kind of a corrupted state. Can you first delete the car application (/repository/deployment/server/carbonapps) and then log in to DAS and go to Manage > Event > Streams and delete any existing streams. Then try again to deploy the car app in the /repository/deployment/server/carbonapps location.
If everything goes well you would see two scripts in Manage > Batch Analytics > Scripts section. Try to execute each script and see if there is any error. If not then you can point the API manager to DAS
We want to push legency system's log to BAM, and to send email to administrator if an exception occurs.
Data publisher helps a lot, but I don't know about CEP. How can I manage the email sent real time?
If you are analyzing the logs real-time, you can use WSO2 CEP with an email output event adaptor to generate notifications as emails. For a sample on generating email notifications, you can refer to sample no. 0106 here.
You will have to write a query to detect exceptions in the logs here. Writing a filter query with 'contains' clause may be enough if you are just detecting the occurrences of word 'exception'. You can find documentation on writing filters here.