[
https://issues.apache.org/jira/browse/GEODE-9152?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Kirk Lund updated GEODE-9152:
-----------------------------
Description:
Logging statements using log markers should be INFO level.
Having log markers at TRACE or DEBUG is worthless because you have to set the
log-level to TRACE and also enable that log marker which will fill the disk
with logging.
All of the logging statements with log markers should be changed to INFO level.
Then having the log-level set to INFO and enabling the log marker will enable
that log statement which is the expectation that most users and developers have.
In Geode, all log markers are defined in
geode-core/src/main/java/org/apache/internal/logging/log4j/LogMarker. The ideal
test would simply be a LogMarker test which enables one LogMarker at a time and
verifies that logging it at INFO level results in it being logged as expected.
Then all uses of that LogMarker should be changed to INFO level. I don't think
every class using that LogMarker needs to have tests for enabling or disabling
it.
FastLogger in geode-logging will require the addition of one method to avoid
any performance degradation:
{noformat}
@Override
public boolean isInfoEnabled(Marker marker) {
return delegating && super.isInfoEnabled(marker);
}
{noformat}
was:
Logging statements using log markers should be INFO level.
Having log markers at TRACE or DEBUG is worthless because you have to set the
log-level to TRACE and also enable that log marker which will fill the disk
with logging.
All of the logging statements with log markers should be changed to INFO level.
Then having the log-level set to INFO and enabling the log marker will enable
that log statement which is the expectation that most users and developers have.
In Geode, all log markers are defined in
geode-core/src/main/java/org/apache/internal/logging/log4j/LogMarker. The ideal
test would simply be a LogMarker test which enables one LogMarker at a time and
verifies that logging it at INFO level results in it being logged as expected.
Then all uses of that LogMarker should be changed to INFO level. I don't think
every class using that LogMarker needs to have tests for enabling or disabling
it.
> Logging statements using log markers should be INFO level
> ---------------------------------------------------------
>
> Key: GEODE-9152
> URL: https://issues.apache.org/jira/browse/GEODE-9152
> Project: Geode
> Issue Type: Bug
> Components: logging
> Reporter: Kirk Lund
> Priority: Major
>
> Logging statements using log markers should be INFO level.
> Having log markers at TRACE or DEBUG is worthless because you have to set the
> log-level to TRACE and also enable that log marker which will fill the disk
> with logging.
> All of the logging statements with log markers should be changed to INFO
> level. Then having the log-level set to INFO and enabling the log marker will
> enable that log statement which is the expectation that most users and
> developers have.
> In Geode, all log markers are defined in
> geode-core/src/main/java/org/apache/internal/logging/log4j/LogMarker. The
> ideal test would simply be a LogMarker test which enables one LogMarker at a
> time and verifies that logging it at INFO level results in it being logged as
> expected. Then all uses of that LogMarker should be changed to INFO level. I
> don't think every class using that LogMarker needs to have tests for enabling
> or disabling it.
> FastLogger in geode-logging will require the addition of one method to avoid
> any performance degradation:
> {noformat}
> @Override
> public boolean isInfoEnabled(Marker marker) {
> return delegating && super.isInfoEnabled(marker);
> }
> {noformat}
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
