Github user noedetore commented on a diff in the pull request:
https://github.com/apache/accumulo/pull/220#discussion_r102815549
--- Diff:
server/master/src/main/java/org/apache/accumulo/master/metrics/Metrics2ReplicationMetrics.java
---
@@ -58,6 +59,7 @@ protected void snapshot() {
registry.add(PENDING_FILES, getNumFilesPendingReplication());
registry.add(NUM_PEERS, getNumConfiguredPeers());
registry.add(MAX_REPLICATION_THREADS, getMaxReplicationThreads());
+ registry.add(LATENCY, getLatencyInSeconds());
--- End diff --
I agree there are better ways of getting a better metric. This approach is
a 90%(maybe to generous) solution using the information that already exists to
give an end user some bearing of confidence as to current replication latency.
From a users perspective, they want a historical way to know what is the
greatest latency they can expect their data to be replicated at a given point
in time. Spikes help explain the possible -1% unhealthy replicated files. As in
the picture above there are couple spikes, but they don't seem to paint a bad
picture. I will look into the histogram class. Are there any points of
reference in the code you can point me to?
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
