VMware vFabricTM AppInsight 5.0 Release Notes
vFabric AppInsight 5.0 | 19 November 2012 | Build 898678
Last Document Update: 27 November 2012
These release notes cover the following topics:
This AppInsight 5.0 version includes the following new features.
- Substantial increase in the scale of components that can be monitored.
- Transaction tracing based on code agents.
- Mapping of transactions between application servers, queues, and so on.
- Code monitoring support for WebLogic.
- Code monitoring support for Java 7.
- Postgres databases can be network-monitored components.
- Database components that are code monitored have database query breakdown.
- Enhanced AppInsight application topology synchronization with Application Director deployments.
- New look dashboard and topology map with enhanced data.
- Documentation is provided online and in downloadable PDF and mobile formats.
VMware vFabric AppInsight 5.0 Documentation
Additional documentation is available through the AppInsight Documentation Center.
System Requirements for VMware vFabric AppInsight 5.0 Installation
Before you begin the installation, make sure you have these system requirements available. You also need to
have specific ports open for the AppInsight server virtual machine and network probe.
You require the following disk space and memory for the AppInsight server virtual machine and the network probe:
- AppInsight Server virtual machine: 100 GB disk space; 8 GB memory, 4 vCPU
- Network probe: 6 GB disk space; 1 GB memory, 2 vCPU
You require the following open ports for the AppInsight server virtual machine and the network probe:
- AppInsight Server virtual machine
- TCP Port 80, 8443 - For the AppInsight user interface connection from the external client
- TCP Port 1194 - For the network probe connection to this AppInsight server virtual machine
- TCP Port 21234 - (Optional) Code agent connection with the AppInsight server
- TCP Port 5671 - For RabbitMQ using SSL on the AppInsight server virtual machine
- Network Probe
- TCP Port 1194 - For the connection to the AppInsight server virtual machine
- UDP Port 123 - For the NTP on the AppInsight server virtual machine
- TCP Port 5671 - (When multiple collectors are installed) For the connection to the AppInsight message queue
The AppInsight server must be time-synchronized with the code agent, network probes, and the servers of any other VMware products that are integrated with AppInsight, for example vFabric Hyperic and vFabric Application Director.
Supported Screen Resolution
AppInsight is supported on screen resolutions of 1024 x 768, and higher.
VMware recommends that you view vFabric AppInsight in full screen mode.
Web Interface Support
Supported Web browsers:
- Microsoft Internet Explorer 8, and later
- Google Chrome 12, and later
- Mozilla FireFox 3.6, and later
- Safari 4.0, and later
System Limitations for Code Agents
If you are using Ubuntu 10.04 and 11.04 running the Ubuntu Java package, upgrade the JRE version to 1.7, or change the Ubuntu JRE to an Oracle JRE on the code agent machine.
Refer to the VMware vFabric AppInsight Installation Guide for installation instructions.
After you have installed AppInsight, you can deploy it in large scale environments. See, KB article 2039352.
Resolved Issues in the VMware vFabric AppInsight 5.0 Release
The following issues have been resolved since the last AppInsight release:
When you install multiple code agents, the temp folder becomes full, causing the AppInsight server to stop working.
Delete the content in the /opt/vmware/apm/insight-dashboard/temp directory after each code agent installation.
vFabric Hyperic Integration
During startup, if an adapter connection fails, AppInsight will not start.
- Stop the Hyperic server.
- Start AppInsight.
- Restart the Hyperic server.
- Reestablish the adapter connection.
vCenter Chargeback Integration
In the Threshold Setting window for the Cost KPI, you cannot enable monitoring.
Cost monitoring is not available in the 5.0 version of AppInsight.
- Following an AppInsight Server virtual machine restart, KPI metrics might not be displayed.
It might take up to two minutes before the KPI colors appear in the dashboard.
- Run service rabbitmq-server status and check that RabbitMQ is running.
- Run service metric-store restart.
- Check metric-store.log to verify that no errors have occurred since the server virtual machine startup.
- The Cost KPI is measured per hour, not per day as stated in the signpost help popup.
- Occasionally, after a snapshot has been taken of a virtual machine, the adapters do not connect and KPIs are not calculated.
Restart the virtual machine.
When more than one component type exists on the same virtual machine, AppInsight shows a single component type for all elements.
To delete and reregister an adapter follow this process.
- Delete the adapter.
- From conmanage.py, run restart_appinsight.
- Reregister the adapter.
In the event that an adapter fails to collect data or fails to display metrics follow this process.
- Delete the adapter.
- From conmanage.py, run stop_appinsight.
- Run psql -U postgres appinsight to access the database.
- Run the SQL truncate table am.objectequivalent; against the AppInsight database.
- From conmanage.py, run restart_appinsight.
- Reregister the adapter.
Known Issues with VMware vFabric AppInsight 5.0
The following known issues affect the AppInsight 5.0 release:
- If there is a mismatch between the name of a vSwitch and the name of the AppInsight port group on the vSwitch, network probe deployment fails.
- You cannot install a code agent on a Windows 2008 Server 64-bit running Tomcat 7.0.25 as a service.
Upgrade to the next Tomcat version.
- When you add or remove a metric to a configurable chart, the colors of all existing lines change.
- In Internet Explorer, a Certificate Error message appears.
- The name of a virtual machine that is powered off does not appear in the user interface. A different IP-based platform name appears.
- AppInsight only supports languages that read left-to-right.
- When you try to log into AppInsight on Internet Explorer 9 on Windows 2008 R2, you get a blank page.
Either disable Internet Explorer Enhanced Security Configuration, or configure it to allow access to the AppInsight server by changing the security level to medium.
- When you load a topology map the first time, it might take a few minutes to load.
There is no indication that the map is loading.
- If you use an unsupported SSL key format, the following message appears Some components were not added. Try adding these components individually.
If you make configuration changes in servers that are monitored by Hyperic, you must delete the Hyperic adapter in AppInsight and reregister it so that the changes are reflected in the Middleware KPI.
- In the case in which duplicate IP addresses exist in vCenter Server, which generally only occurs with internal IP addresses, AppInsight might display an incorrect virtual machine name.
- vCenter virtual disk and datastore metrics are not available the 5.0 release of AppInsight.
- If you are using tcServer, verify that there there is only one JVM_OPTS line in the setenv.sh file before deploying the code agent.
- Database transactions, other than JDBC, are not recognized by the code agent.
- You cannot completely shut down a Tomcat server while a code agent is installed. Run kill -9 to manually stop Tomcat.
vFabric Hyperic Integration
- Middleware KPI data that is retrieved from Hyperic might be inaccurate or unavailable if the Hyperic time is not synchronized with APM Server time.
- A scaled-out virtual machine that was not detected as a potential component is not deleted in Hyperic during a teardown.
You must manually delete the virtual machine in Hyperic.
- If a root directory that is being monitored for changes is deleted, changes are no longer tracked, even if the root directory is re-created.
Reconfigure the tracking or restart the agent.
vCenter Orchestrator Integration
- If vCenter Orchestrator is configured with multiple vCenter Servers and an action is performed on a virtual machine that uses the same name in multiple vCenter Servers, the action is not performed, to prevent errors. A message appears in the user interface log.
- You must use an HTTP connection to manage vCenter Orchestrator workflows in AppInsight.
When you add a vCenter Orchestrator adapter in AppInsight, ensure that the "SSL Connection" check box is not selected.
vFabric Application Director Integration
- Before using the AppInsight code agent installer service, verify that cURL exists in the virtual machine template /usr/bin/curl.
- Avoid tearing down an application that contains transaction elements with the same name.
In the case in which a teardown has already been performed, the teardown will be successful in Application Director but will not succeed in AppInsight. Contact AppInsight technical support for assistance.
- In the case in which you have scaled out or torn down an application and the process has not completed successfully, and you deploy a new application, the virtual machines might be assigned the IP addresses of the original application, causing a conflict.
Contact AppInsight technical support for assistance.
- When a server has two IP addresses, one for incoming traffic and another for outgoing traffic, hints related to outgoing traffic are not available.
- If there is a correlation between components in the left-most tier of an application's topology, redundant transactions are created.
- Rarely, when metrics are added or removed from vCenter or Hyperic, the change does not appear in the topology in AppInsight.
Try restarting AppInsight to resolve the issue.
- If you attempt to add a cloned virtual machine with a static IP address to the topology of an AppInsight application, the virtual machine will not be added.
- Servers that use a range of ports are not supported. In AppInsight, the server is identified as a single application component.
- When you add a new tier to the topology as the left-most tier, all existing transaction data is deleted.
New transaction data is displayed on the new left-most tier, as it is detected by AppInsight.
- In the potential component table, you cannot switch a network component in the topoology with more than one code-monitored component at a time.
Switch the network component with a single code-monitored component, then add other code-monitored components as required.
- AppInsight supports PEM- and PKCS-format SSL keys.
In the event that you are unable to manually add WebSphere SSL certificates, verify that your keys are of a supported format.
- When the host clock for a probe is not synchronized with the clock on the server host, probe deployment will fail until the time on the probe host matches the time when the server was first installed on the server host.
- In applications that are built from the same page template, for example Wiki, transactions are not learned by AppInsight.
- AppInsight does not recognize Oracle errors.
In such cases, 0% errors appears instead of the correct error rate.
- When you create a new application component group, it might take up to 25 minutes before monitored data appears in the application's topology.
- When you create a new application component group, the monitored data is not correct until each of the components in group have existed for the entire time resolution.
For example, if an application group has three components that are being monitored in a time resolution of one hour, but one of the components was only created 45 minutes earlier, the displayed application component group data is only accurate after one hour and 15 minutes (after the most recently created component is in existence for one hour.)
- The latency breakdown chart in the Metrics view is not available for code-monitored components.
When you create an alert on a specific application using the Every Component application object, notifications might not be sent.
Either select Alert on Every application and Every Component, or a specific application and specific application object.
- When you are automatically logged out of AppInsight after a period of inactivity, relogin might not display the AppInsight user interface.
Close the browser, then reopen and restart AppInsight.
- If the the hostname used by an application server to connect to an external resource, for example a database or LDAP, is defined in the /etc/hosts file, you must also define it in the AppInsight server /etc/hosts file in order to retrieve middleware and infrastructure data from the external device.
- If the AppInsight process stops or is restarted, or the vCenter adapter is disconnected, when AppInsight is running again or the adapter is reconnected, verify that the AppInsight license is still applied.