P4
Currently, to get the description/name of the safepoint that is started, you need to enable -Xlog:safepoint=debug.
However, this enables thousands of additional log messages that are very detailed, like for every java thread that it hit the safepoint polling page.
That fills up the log just to get the
Entering safepoint region: <something>
Leaving safepoint region
messages. Log level "info" is too coarse, as it does not print the reason for the safepoint at all, like
[705.131s][info ][safepoint ] Total time for which application threads were stopped: 0.0559900 seconds, Stopping threads took: 0.0507676 seconds
[705.388s][info ][safepoint ] Application time: 0.2566720 seconds
It would be great to either move the very verbose messages to trace level, or add this information about the safepoint kind to the info level.
Attached a log that is mostly contains these "polling page" messages.
However, this enables thousands of additional log messages that are very detailed, like for every java thread that it hit the safepoint polling page.
That fills up the log just to get the
Entering safepoint region: <something>
Leaving safepoint region
messages. Log level "info" is too coarse, as it does not print the reason for the safepoint at all, like
[705.131s][info ][safepoint ] Total time for which application threads were stopped: 0.0559900 seconds, Stopping threads took: 0.0507676 seconds
[705.388s][info ][safepoint ] Application time: 0.2566720 seconds
It would be great to either move the very verbose messages to trace level, or add this information about the safepoint kind to the info level.
Attached a log that is mostly contains these "polling page" messages.