EdColeman commented on code in PR #3167:
URL: https://github.com/apache/accumulo/pull/3167#discussion_r1136053141
##########
server/manager/src/main/java/org/apache/accumulo/manager/Manager.java:
##########
@@ -1020,17 +1022,29 @@ private SortedMap<TServerInstance,TabletServerStatus>
gatherTableInformation(
badServers.remove(server);
}
}
- });
- }
- tp.shutdown();
- try {
- tp.awaitTermination(Math.max(10000, rpcTimeout / 3), MILLISECONDS);
- } catch (InterruptedException e) {
- log.debug("Interrupted while fetching status");
+ }));
+ }
+ final long timeToWaitForCompletion = Math.max(10000, rpcTimeout / 3);
+ long currTime = NANOSECONDS.toMillis(System.nanoTime());
+ final long timeToCancelTasks = currTime + timeToWaitForCompletion;
+ // Wait for all tasks to complete
+ while (!tasks.isEmpty()) {
+ boolean cancel = (currTime > timeToCancelTasks);
Review Comment:
The time returned by `System.nanoTime()` is only valid when compared
relative to another nano time. The value is relative to some arbitrary counter
in the JVM and can be negative.
This should be sometime like ```
boolean cancel = ((currTime - System.nanoTime()) >
MILLISECONDS.toNanos(timeToCancelTasks));
```
If may also help if all times were kept as nanos - or converted to nanos for
the calculations and only converting back to millis (or seconds) for display to
the user. Using one set of units consistently may help limit conversion errors
if a conversion was missed.
--
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]
