tkhurana commented on code in PR #5081:
URL: https://github.com/apache/hbase/pull/5081#discussion_r1125125840
##########
hbase-client/src/main/java/org/apache/hadoop/hbase/client/ConnectionImplementation.java:
##########
@@ -1123,12 +1127,26 @@ rpcControllerFactory, getMetaLookupPool(),
connectionConfig.getMetaReadRpcTimeou
}
}
- void takeUserRegionLock() throws IOException {
+ private long heldStartTime;
+
+ void takeUserRegionLock(int tries) throws IOException {
try {
long waitTime = connectionConfig.getMetaOperationTimeout();
+ long waitStartTime = 0;
+ if (metrics != null) {
+ metrics.updateUserRegionLockQueue(userRegionLock.getQueueLength());
+ waitStartTime = EnvironmentEdgeManager.currentTime();
+ }
if (!userRegionLock.tryLock(waitTime, TimeUnit.MILLISECONDS)) {
+ if (metrics != null) {
+ metrics.incrUserRegionLockTimeout();
+ }
throw new LockTimeoutException("Failed to get user region lock in" +
waitTime + " ms. "
+ " for accessing meta region server.");
+ } else if (metrics != null) {
+ // successfully grabbed the lock, start timer of holding the lock
+ heldStartTime = EnvironmentEdgeManager.currentTime();
+ metrics.updateUserRegionLockWaiting(tries * waitTime + heldStartTime -
waitStartTime);
Review Comment:
This is not correct. In every iteration we take the user lock and then
release the lock. So the wait time cannot be aggregated across tries. Also, in
every attempt if we get an exception while acquiring the lock we throw and do
not make further retries.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
