[
https://issues.apache.org/jira/browse/LOG4J2-3154?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17407241#comment-17407241
]
Michael Cramer commented on LOG4J2-3154:
----------------------------------------
we considered and tried both:
* maxStringLength: what we have seen is that mostly we "lost" the interesting
part of the entire stack trace as the cause is at the end and truncation
removes that part
* stack trace truncation: the challenge or problem here is when we want to
exclude all org.spring.. related parts which appear multiple times in a stack
trace also in the middle of our code (caused by aspects and proxies) we "lose"
everything afterwards
Â
maybe as an idea (we are doing smth. similar for a different use-case)
{code:java}
private static final NameAbbreviator nameAbbreviator =
NameAbbreviator.getAbbreviator("1.");
public static String abbreviateStackTrace(@Nonnull final Throwable throwable) {
final StringBuilder sb = new StringBuilder();
Arrays.stream(ExceptionUtils.getStackFrames(throwable)).forEach(line -> {
// ExceptionUtils from commons-lang3
if (line.startsWith("\tat ")) {
final String classNameDotMethod = line.substring("\tat ".length(),
line.indexOf('('));
final int methodNameIndex = classNameDotMethod.lastIndexOf('.');
final String className = classNameDotMethod.substring(0,
methodNameIndex);
sb.append(line, 0, "\tat ".length());
nameAbbreviator.abbreviate(className, sb);
sb.append(line.substring("\tat ".length() + methodNameIndex));
} else {
sb.append(line);
}
sb.append(System.lineSeparator());
});
return sb.toString();
} {code}
> Support NameAbbreviator in ExceptionResolver when using JsonTemplateLayout
> --------------------------------------------------------------------------
>
> Key: LOG4J2-3154
> URL: https://issues.apache.org/jira/browse/LOG4J2-3154
> Project: Log4j 2
> Issue Type: Improvement
> Components: JsonTemplateLayout, Layouts
> Affects Versions: 2.14.1
> Reporter: Michael Cramer
> Priority: Major
> Labels: abbreviation, exception, json, layout, resolver, template
>
> when stack traces are written to the JSON that is later consumed by Graylog
> for example sometimes the processing failed because the field containing the
> stack trace is larger than 32k which is a limit of the in ElasticSearch
> ([https://github.com/Graylog2/graylog2-server/issues/873)]
> I propose to add support for using the already existing NameAbbreviator to
> shorten stack traces. it might not completely solve all issues that the field
> might have content that is too large but tries to mitigate it.
> {noformat}
> {
> "$resolver": "exception",
> "field": "stackTrace",
> "stackTrace": {
> "stringified": true
> }
> } {noformat}
> can be smth. like
> {noformat}
> {
> "$resolver": "exception",
> "field": "stackTrace",
> "stackTrace": {
> "stringified": true,
> "abbreviatePattern": "1."
> }
> } {noformat}
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
