HDDS-10658. audit log message having transaction id

[sumitagrawl]

What changes were proposed in this pull request?

For debuggability of om transaction and audit log mapping, audit log is included with transaction Id. And additionally unknown failure case, also having audit log with exception message and transaction id.

What is the link to the Apache JIRA

https://issues.apache.org/jira/browse/HDDS-10658

How was this patch tested?

• Unit test is added

[adoroszlai]

Thanks @sumitagrawl for the patch.

[adoroszlai]

This method is for building the audit message based on the input map. termIndex should not be added as a parameter. It should be added to the map by the caller.

[sumitagrawl]

Chances of missing adding termIndex at every caller is high, so its done at common place. This is to avoid coding mistake, as this should be mandatory parameter for logging, and required for logging. Other parameter can be optional and based on specific request type.

[adoroszlai]

This piece of code (passing -1):

ozone/hadoop-ozone/ozone-manager/src/main/java/org/apache/hadoop/ozone/om/request/security/OMGetDelegationTokenRequest.java

Lines 83 to 84 in 63011e1

	buildAuditMessage(OMAction.GET_DELEGATION_TOKEN,
	new LinkedHashMap<>(), ioe, request.getUserInfo(), TransactionInfo.getTermIndex(-1)));

suggests that it is not mandatory for all requests. On closer look this call is from preExecute, which doesn't have termIndex. Only requests related to delegation tokens do audit failures in preExecute. We may want to fix that.

I think we should change the place where audit logging happens for OM requests:

• start building AuditMessage.Builder in requests, but do not build() and log it yet
• add the in-progress AuditMessage.Builder to the OMClientResponse returned from validateAndUpdateCache
• central server-side request/response handler (OzoneManagerProtocolServerSideTranslatorPB OzoneManagerRequestHandler) can add common information like termIndex, finish building AuditMessage and log it
• OzoneManagerProtocolServerSideTranslatorPB can also log failures that occur in preExecute

[errose28]

Can you include some before and after log message examples in the PR description for others to quickly reference?

[sumitagrawl]

Changes are udpated in new pull request #6949