[WFCORE-6476-fix]: Reducing logging level for failed internal read-on…

[ehsavoie]

…ly operation steps.

Jira issue: https://issues.redhat.com/browse/WFCORE-6476

• Fixing missing case

[yersan]

I am not sure why this case was left from the previous initial work. However, it makes sense to me.

I can only see two consequences:

1. Errors arriving at this point when the stage is Stage.DOMAIN, will no longer logged, but @bstansberry added this comment on his previous change

   wildfly-core/controller/src/main/java/org/jboss/as/controller/AbstractOperationContext.java

   Lines 1132 to 1133 in 035d412

   	// Currently Stage.DOMAIN failure handling involves message manipulation before sending the
   	// failure data to the client; logging stuff before that is done is liable to just produce a log mess.

   , so probably ignoring them would be fine.

If even with that we are not sure and want to be conservative, we could ask for the stage here and log them in such a case. For example

if (currentStage != Stage.DOMAIN) {
   logStepFailure(step, false);
} else {
MGMT_OP_LOGGER.operationFailed(t, step.operation.get(OP), 
step.operation.get(OP_ADDR));
}

2. In the worst case, this change will change some specific logs from ERROR to DEBUG and will not change specific server behavior.

If there are no other thoughts, I would approve it as it is.

@pferraro , do you have any thoughts about this?

[yersan]

Another possible more conservative alternative is to use the new method only for the internal callers and left anything else as it was before:

if (!isExternalClient()) {
    logStepFailure(step, false);
} else {
    MGMT_OP_LOGGER.operationFailed(t, step.operation.get(OP), step.operation.get(OP_ADDR));
}

@ehsavoie Would the above resolve your issue too? It seems it could be a more secure option to avoid miss or change any log-level

[ehsavoie]

I would also ensure that the op is read-only

[yersan]

Thanks @ehsavoie