wrapper.log.warning.threshold

Compatibility :3.5.8
Editions :Professional EditionStandard EditionCommunity Edition
Platforms :WindowsMac OSXLinuxIBM AIXFreeBSDHP-UXSolarisIBM z/OSIBM z/Linux

When set to a value larger than "0" (zero), the Wrapper will measure how long each individual logging log takes. It will then record a log entry if the time is ever more than the specified threshold.

NOTE

This is only designed for Wrapper testing and should not be enabled for most systems. Its purpose is to help identify when the Wrapper is encountering delays to IO backlogs at the system level.

The default value is "0" (zero) milliseconds. Possible values are then 1 to 3,600,000ms (1 hour). Values larger or smaller than this range will be silently adjusted.

Values less than 20 milliseconds will result in fairly regular warning on most systems. The measurements work off of the system time so adjustments to the system time can cause warnings if they occur between measurements. It is normal for 10 or 16ms times to be logged as part of normal operation due to the process switching of the OS even on unloaded systems.

Example:(10 milliseconds)
wrapper.log.warning.threshold=10

When a log warning message is generated, it will be logged as a queued message. This can cause the warning message to be logged several messages after the actual slow message. You may wish to also enable the 'G' log format as described below to narrow down the actual slow message.

In general slow log messages are caused by slow I/O access to the HD. This problem is more common in virtual servers, especially when a number of guest VMs are sharing the same host. Enabling warnings or the 'G' log format can help identify when I/O problems are being encountered as they will most likely be affecting other portions of your application as well.

'G' log format

By enabling the 'G' log format, it is also possible to see the number of milliseconds that the previous log message took even for those below this threshold.

Note that this time is logged for each message that goes through the logging system so a slow DEBUG message followed by an INFO message would show the DEBUG message's time. If the log level of the console is DEBUG while the log level of the log file is INFO, then the previous message time will be shown in both places as part of the INFO message. When only viewing the logfile, this can be a bit misleading as to which message was slow, however it is not possible to log a message's log duration within itself.

Console example with wrapper.console.loglevel=DEBUG and wrapper.console.format=TGM
INFO   | 2011/04/01 14:02:29 |        0 | Normal Message
DEBUG  | 2014/10/03 14:02:29 |        0 | Slow Message
INFO   | 2014/10/03 14:02:29 |      100 | Normal Message
Logfile example with wrapper.logfile.loglevel=INFO and wrapper.logfile.format=TGM
INFO   | 2011/04/01 14:02:29 |        0 | Normal Message
INFO   | 2014/10/03 14:02:29 |      100 | Normal Message

Reference: logfile