A. BEST PRACTICES:
- Ensure that NO other conflicting program is being bootstrapped into the JVM with Diags/TV; especially Wily Introscope, ITCAM
- Ensure that ALL probes have an explicit and unique probe.id.
- Ensure that ALL probes start, and that you validate the port numbers upon start-up.
- ALWAYS install a Diagnostics server for a TV installation, and use it as the sanity console : validate probe communications, application call flow, instrumentation, probe counts, probe status
B. POSSIBLE SIDE-EFFECTS : Failure to follow all BEST PRACTICES may lead to these conditions
- Failure to remove -javaagent of -Xbootclasspath JVM arguments that load conflicting products. Our tool may start, the app may start, but there are countless unexpected side-effects of these conflicts.
- Cutting-and-pasting JVM arguments, and starting more than one probe in the enterprise with non-unique probe.id's
- Unstable app - unexplained crashes, excessive exceptions
C. OBSERVATIONS : Watch for these conditions and act accordingly
- Probe starts with an unexpectedly high port number, such as 35002 when you only expect ONE or TWO probes on the server.
- Probe count is higher than you expect in the Enterprise view.
- Unexpected classes instrumented.
- Unexpected server requests may be a result of:
- either our server is receiving extraneous requests that are not part of our app/traffic,
- another probe with a duplicate probe.id is running on another app
Have more questions? Submit a request