HDDS-15287. Remove unused deferred/async RPC

[adoroszlai]

What changes were proposed in this pull request?

Same as #10280, plus:

• removed async call from forked hadoop.io.retry package (aad36a6)
• merged master to resolve conflicts

https://issues.apache.org/jira/browse/HDDS-15287

How was this patch tested?

https://github.com/adoroszlai/ozone/actions/runs/26091212865

[ivandika3]

LGTM +1.

[ivandika3]

Nit: Remove asynchronous calls, I'm OK to leave it be.

[adoroszlai]

Thanks. I'll leave it for now, but while checking it, I spotted a potential bug (both here and in Hadoop):

1. RetryInfo.retryTime is calculated when it is created:
   

   ozone/hadoop-hdds/common/src/main/java/org/apache/hadoop/io_/retry/RetryInvocationHandler.java

   Lines 253 to 254 in c917a6e

   	this.delay = delay;
   	this.retryTime = Time.monotonicNow() + delay;

2. waitTime is calculated before sleep as:
   

   ozone/hadoop-hdds/common/src/main/java/org/apache/hadoop/io_/retry/RetryInvocationHandler.java

   Lines 85 to 86 in c917a6e

   	synchronized Long getWaitTime(final long now) {
   	return retryInfo == null? null: retryInfo.retryTime - now;

3. but then original RetryInfo.delay is used for actual sleep time:
   

   ozone/hadoop-hdds/common/src/main/java/org/apache/hadoop/io_/retry/RetryInvocationHandler.java

   Lines 131 to 136 in c917a6e

   	final Long waitTime = getWaitTime(Time.monotonicNow());
   	LOG.trace("#{} processRetryInfo: retryInfo={}, waitTime={}",
   	callId, retryInfo, waitTime);
   	if (waitTime != null && waitTime > 0) {
   	try {
   	Thread.sleep(retryInfo.delay);

What do you think?

[szetszwo]

You are right that it is a bug -- it should use waitTime to sleep.

[szetszwo]

+1 the change looks good.

[adoroszlai]

Thanks @ivandika3, @szetszwo for the repeated review.