[
https://issues.apache.org/jira/browse/CAMEL-24592?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Federico Mariani updated CAMEL-24592:
-------------------------------------
Description:
{{CamelHealthHelper.applyHealthDetail}} adds the full stack trace of a failed
health check to the actuator output at every exposure level except {{oneline}}:
{code:java}
result.getError().ifPresent(error -> {
if (error.getMessage() != null) {
builder.withDetail("error.message", error.getMessage());
}
final StringWriter stackTraceWriter = new StringWriter();
try (final PrintWriter pw = new PrintWriter(stackTraceWriter, true)) {
error.printStackTrace(pw);
data.put("error.stacktrace", stackTraceWriter.toString());
}
});
{code}
(core/camel-spring-boot: CamelHealthHelper.java:62-71)
So with the default {{camel.health.exposure-level=default}}, any DOWN check
whose result carries an exception - a JMS consumer that lost its broker, a JDBC
pool that cannot connect - serialises the whole cause chain, including internal
host names and ports from the connection exception, into {{/actuator/health}}
for any caller allowed to see details.
This is stricter everywhere else:
* Spring Boot's own health indicators expose at most the exception class and
message, never a trace.
* camel-main's management endpoint ({{ManagementHttpServer}}) includes
{{error-stacktrace}} only when the caller explicitly asks with
{{?stackTrace=true}}; the {{full}} level alone adds only the message.
* camel-microprofile-health, through which Camel Quarkus builds its health
responses, emits the trace at the {{default}} level as well, so the same gating
is applied there in a companion change to keep the runtimes aligned.
The user manual describes {{full}} as "all details and status from all the
invoked health checks" and {{default}} as detailed information only for checks
that are DOWN. The stack trace is the most verbose detail there is and belongs
to {{full}}.
The trace was added together with the exposure-level handling in CAMEL-18832
(2022) to align the output with microprofile-health, and that module still
carries the trace at the {{default}} level; gating both on {{full}} keeps the
alignment while removing the trace from the default output.
*Proposal*
* Emit {{error.stacktrace}} only when the exposure level is {{full}}.
* Keep {{error.message}} at the {{default}} level, which mirrors what Spring
Boot's own indicators do.
* Keep the full trace in the application log, where it is today.
* Add a test with a DOWN check that carries an exception, asserting that
{{error.stacktrace}} is absent at {{default}} and present at {{full}}.
Note that this touches the same method as CAMEL-24512 ({{error.message}}
overwritten when several checks are DOWN); the two are best done in one pass.
Anyone relying on the trace at the default level can set
{{camel.health.exposure-level=full}}. Worth a line in the upgrade guide.
----
_This issue was drafted by Claude Code on behalf of Federico Mariani._
was:
{{CamelHealthHelper.applyHealthDetail}} adds the full stack trace of a failed
health check to the actuator output at every exposure level except {{oneline}}:
{code:java}
result.getError().ifPresent(error -> {
if (error.getMessage() != null) {
builder.withDetail("error.message", error.getMessage());
}
final StringWriter stackTraceWriter = new StringWriter();
try (final PrintWriter pw = new PrintWriter(stackTraceWriter, true)) {
error.printStackTrace(pw);
data.put("error.stacktrace", stackTraceWriter.toString());
}
});
{code}
(core/camel-spring-boot: CamelHealthHelper.java:62-71)
So with the default {{camel.health.exposure-level=default}}, any DOWN check
whose result carries an exception - a JMS consumer that lost its broker, a JDBC
pool that cannot connect - serialises the whole cause chain, including internal
host names and ports from the connection exception, into {{/actuator/health}}
for any caller allowed to see details.
This is stricter everywhere else:
* Spring Boot's own health indicators expose at most the exception class and
message, never a trace.
* camel-main's management endpoint ({{ManagementHttpServer}}) includes
{{error-stacktrace}} only when the caller explicitly asks with
{{?stackTrace=true}}; the {{full}} level alone adds only the message.
* Camel Quarkus (camel-quarkus-microprofile-health) never emits a stack trace.
The user manual describes {{full}} as "all details and status from all the
invoked health checks" and {{default}} as detailed information only for checks
that are DOWN. The stack trace is the most verbose detail there is and belongs
to {{full}}.
The trace was added together with the exposure-level handling in CAMEL-18832
(2022) to align the output with microprofile-health, but the MicroProfile
implementation does not carry a trace, so the alignment argument no longer
applies.
*Proposal*
* Emit {{error.stacktrace}} only when the exposure level is {{full}}.
* Keep {{error.message}} at the {{default}} level, which mirrors what Spring
Boot's own indicators do.
* Keep the full trace in the application log, where it is today.
* Add a test with a DOWN check that carries an exception, asserting that
{{error.stacktrace}} is absent at {{default}} and present at {{full}}.
Note that this touches the same method as CAMEL-24512 ({{error.message}}
overwritten when several checks are DOWN); the two are best done in one pass.
Anyone relying on the trace at the default level can set
{{camel.health.exposure-level=full}}. Worth a line in the upgrade guide.
----
_This issue was drafted by Claude Code on behalf of Federico Mariani._
> camel-spring-boot - health check details include the full error stack trace
> at the default exposure level
> ---------------------------------------------------------------------------------------------------------
>
> Key: CAMEL-24592
> URL: https://issues.apache.org/jira/browse/CAMEL-24592
> Project: Camel
> Issue Type: Bug
> Components: camel-spring-boot
> Reporter: Federico Mariani
> Assignee: Federico Mariani
> Priority: Major
> Fix For: 4.23.0
>
>
> {{CamelHealthHelper.applyHealthDetail}} adds the full stack trace of a failed
> health check to the actuator output at every exposure level except
> {{oneline}}:
> {code:java}
> result.getError().ifPresent(error -> {
> if (error.getMessage() != null) {
> builder.withDetail("error.message", error.getMessage());
> }
> final StringWriter stackTraceWriter = new StringWriter();
> try (final PrintWriter pw = new PrintWriter(stackTraceWriter, true)) {
> error.printStackTrace(pw);
> data.put("error.stacktrace", stackTraceWriter.toString());
> }
> });
> {code}
> (core/camel-spring-boot: CamelHealthHelper.java:62-71)
> So with the default {{camel.health.exposure-level=default}}, any DOWN check
> whose result carries an exception - a JMS consumer that lost its broker, a
> JDBC pool that cannot connect - serialises the whole cause chain, including
> internal host names and ports from the connection exception, into
> {{/actuator/health}} for any caller allowed to see details.
> This is stricter everywhere else:
> * Spring Boot's own health indicators expose at most the exception class and
> message, never a trace.
> * camel-main's management endpoint ({{ManagementHttpServer}}) includes
> {{error-stacktrace}} only when the caller explicitly asks with
> {{?stackTrace=true}}; the {{full}} level alone adds only the message.
> * camel-microprofile-health, through which Camel Quarkus builds its health
> responses, emits the trace at the {{default}} level as well, so the same
> gating is applied there in a companion change to keep the runtimes aligned.
> The user manual describes {{full}} as "all details and status from all the
> invoked health checks" and {{default}} as detailed information only for
> checks that are DOWN. The stack trace is the most verbose detail there is and
> belongs to {{full}}.
> The trace was added together with the exposure-level handling in CAMEL-18832
> (2022) to align the output with microprofile-health, and that module still
> carries the trace at the {{default}} level; gating both on {{full}} keeps the
> alignment while removing the trace from the default output.
> *Proposal*
> * Emit {{error.stacktrace}} only when the exposure level is {{full}}.
> * Keep {{error.message}} at the {{default}} level, which mirrors what Spring
> Boot's own indicators do.
> * Keep the full trace in the application log, where it is today.
> * Add a test with a DOWN check that carries an exception, asserting that
> {{error.stacktrace}} is absent at {{default}} and present at {{full}}.
> Note that this touches the same method as CAMEL-24512 ({{error.message}}
> overwritten when several checks are DOWN); the two are best done in one pass.
> Anyone relying on the trace at the default level can set
> {{camel.health.exposure-level=full}}. Worth a line in the upgrade guide.
> ----
> _This issue was drafted by Claude Code on behalf of Federico Mariani._
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
