|
[
Permalink
| « Hide
]
Dennis Reedy added a comment - 16/Feb/10 08:59 PM
Can you include a snippet of the logger messages?
The scheme below is repeated originating from the different sigar methods.
WARNING: Failed invoking getProcCpu method Instead of printing the stacktrace by default, Rio will log the cause and the message. So the above would look like:
WARNING: Failed invoking getProcCpu method. Caused by: org.hyperic.sigar.SigarException: java.lang.UnsatisfiedLinkError: org.hyperic.sigar.ProcCpu.gather(Lorg/hyperic/sigar/Sigar;J)V: -1 If however the org.rioproject.system.measurable logger is set to Level.FINE, the verbosity will be turned up to include the complete stacktrace. Is this sufficient? Reduce verbosity and also put a test check even if the Sigar class is loaded that it can be used
The other problem beyond verbosity is the constant production of such failed invocations. Even single lines can still amount to large numbers with time.
As indicated, a test has been put in to verify that not only the sigar class can be reflectively loaded, but also to check if the loaded sigar runtime can be used. This seems to address your unique condition for a sigar installation that is not valid.
OK, it seems I did not understand what that means. It's fine then. Thanks!
|
|||||||||||||||||||||||||||||||||||||||||||||||