[
https://issues.apache.org/jira/browse/LOG4J2-975?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14360575#comment-14360575
]
Ronald C Albury edited comment on LOG4J2-975 at 3/13/15 7:58 PM:
-----------------------------------------------------------------
I have slapped together a temporary fix for our site ... I am not suggesting
this is the final solution. There is more cleaning-up and elimination of dead
code that can be done, but my goal was to change as few lines of code as
possible.
The biggest drawback is that any appender which failed at start-up appears to
be kicked out of the failover list - which means if our network share was down
at start-up we never log to it without a restart. I think this is where we need
to use the timer and occasionally try to pick-up the 'early' appenders that
failed at start-up ... that is time consumptive enough that I wouldn't want to
do it too often.
<note: We are currently looking at 'touching' the configuration file once we
get the share working and letting the MonitorInterval reload reconnect to the
share>
Essentially I totally ignore the primary appender and just run with a list of
failover appenders.
{code}
@Override
public void start() {
final Map<String, Appender> map = config.getAppenders();
int errors = 0;
/**********
if (map.containsKey(primaryRef)) {
primary = new AppenderControl(map.get(primaryRef), null, null);
} else {
LOGGER.error("Unable to locate primary Appender " + primaryRef);
++errors;
}
**********/
for (final String name : failovers) {
if (map.containsKey(name)) {
failoverAppenders.add(new AppenderControl(map.get(name), null,
null));
} else {
LOGGER.error("Failover appender " + name + " is not
configured");
}
}
if (failoverAppenders.isEmpty()) {
LOGGER.error("No failover appenders are available");
++errors;
}
if (errors == 0) {
super.start();
}
}
/**
* Handle the Log event.
* @param event The LogEvent.
*/
@Override
public void append(final LogEvent event) {
if (!isStarted()) {
error("FailoverAppender " + getName() + " did not start
successfully");
return;
}
/**********
final long localCheckMillis = nextCheckMillis;
if (localCheckMillis == 0 || System.currentTimeMillis() >
localCheckMillis) {
callAppender(event);
} else {
failover(event, null);
}
**********/
failover(event, null);
}
/**********
private void callAppender(final LogEvent event) {
try {
primary.callAppender(event);
nextCheckMillis = 0;
} catch (final Exception ex) {
nextCheckMillis = System.currentTimeMillis() + intervalMillis;
failover(event, ex);
}
}
**********/
{code}
was (Author: eialbur):
I have slapped together a temporary fix for our site ... I am not suggesting
this is the final solution. There is more cleaning-up and elimination of dead
code that can be done, but my goal was to change as few lines of code as
possible.
The biggest drawback is that any appender which failed at start-up appears to
be kicked out of the failover list - which means if our network share was down
at start-up we never log to it without a restart. I think this is where we need
to use the timer and occasionally try to pick-up the 'early' appenders that
failed at start-up ... that is time consumptive enough that I wouldn't want to
do it too often.
Essentially I totally ignore the primary appender and just run with a list of
failover appenders.
{code}
@Override
public void start() {
final Map<String, Appender> map = config.getAppenders();
int errors = 0;
/**********
if (map.containsKey(primaryRef)) {
primary = new AppenderControl(map.get(primaryRef), null, null);
} else {
LOGGER.error("Unable to locate primary Appender " + primaryRef);
++errors;
}
**********/
for (final String name : failovers) {
if (map.containsKey(name)) {
failoverAppenders.add(new AppenderControl(map.get(name), null,
null));
} else {
LOGGER.error("Failover appender " + name + " is not
configured");
}
}
if (failoverAppenders.isEmpty()) {
LOGGER.error("No failover appenders are available");
++errors;
}
if (errors == 0) {
super.start();
}
}
/**
* Handle the Log event.
* @param event The LogEvent.
*/
@Override
public void append(final LogEvent event) {
if (!isStarted()) {
error("FailoverAppender " + getName() + " did not start
successfully");
return;
}
/**********
final long localCheckMillis = nextCheckMillis;
if (localCheckMillis == 0 || System.currentTimeMillis() >
localCheckMillis) {
callAppender(event);
} else {
failover(event, null);
}
**********/
failover(event, null);
}
/**********
private void callAppender(final LogEvent event) {
try {
primary.callAppender(event);
nextCheckMillis = 0;
} catch (final Exception ex) {
nextCheckMillis = System.currentTimeMillis() + intervalMillis;
failover(event, ex);
}
}
**********/
{code}
> Failover Appender does not start if primary appender fails
> ----------------------------------------------------------
>
> Key: LOG4J2-975
> URL: https://issues.apache.org/jira/browse/LOG4J2-975
> Project: Log4j 2
> Issue Type: Bug
> Components: Appenders
> Affects Versions: 2.2
> Environment: Windows-7
> Reporter: Ronald C Albury
>
> We are supposed to log to a network share - but sometimes it is not
> available. I created a Failover appender with a RollingFile appender to the
> network share as the primary, and a RollingFile appender to a local drive as
> a failover. If the network share is not available when my application starts,
> the Failover appender fails in its entirety and nothing is logged anywhere.
> <Failover name="MyFile" primary="ShareFile">
> <Failovers>
> <AppenderRef ref="LocalFile" />
> <AppenderRef ref="MyConsole" />
> </Failovers>
> </Failover>
> 2015-03-12 12:06:49,717 ERROR Attempted to append to non-started appender
> MyFile
> 2015-03-12 12:06:49,717 ERROR FailoverAppender MyFile did not start
> successfully
> 2015-03-12 12:06:49,717 ERROR Attempted to append to non-started appender
> MyFile
> 2015-03-12 12:06:49,717 ERROR FailoverAppender MyFile did not start
> successfully
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
