[
https://issues.apache.org/jira/browse/HADOOP-14989?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16235883#comment-16235883
]
Erik Krogen edited comment on HADOOP-14989 at 11/2/17 3:14 PM:
---
Assuming nothing else was submitting block reports, then {{val}} with the
current code would be 10, but it should be 5005 (it is an average so {{=
10010/2}}). Since it is taking metrics from a minicluster there are also
some real block reports that skew things; that's why I used a big value and a
comparison rather than equal. Like I said, hacky. But the test will
definitively pass if you omit the JMX call and definitely fail if you include
it. I'll try to put together a real unit test for this.
I am not sure what you mean about JMX mbean calling reset internally. Are you
talking here about the metrics2 level reset
({{MetricsSourceAdapter#updateJmxCache()}}) or something at a JVM level? I
explained how the cache reset is managed at the metrics2 level; let me know if
there's something about my explanation that was not clear.
was (Author: xkrogen):
Assuming nothing else was submitting block reports, then {{val}} with the
current code would be 10, but it should be 5005 (it is an average so {{=
10010/2}}). Since it is taking metrics from a minicluster there are also
some real block reports that skew things; that's why I used a big value and a
comparison rather than equal. Like I said, hacky. But the test will
definitively pass if you omit the JMX call and definitely fail if you include
it. I'll try to put together a real unit test for this.
I am not sure what you mean about JMX mbean calling reset internally. Are you
talking here about the metrics2 level reset
({{MetricsSourceAdapter#updateJmxCache()}} or something at a JVM level? I
explained how the cache reset is managed at the metrics2 level; let me know if
there's something about my explanation that was not clear.
> Multiple metrics2 sinks (incl JMX) result in inconsistent Mutable(Stat|Rate)
> values
> ---
>
> Key: HADOOP-14989
> URL: https://issues.apache.org/jira/browse/HADOOP-14989
> Project: Hadoop Common
> Issue Type: Bug
> Components: metrics
>Affects Versions: 2.6.5
>Reporter: Erik Krogen
>Assignee: Erik Krogen
>Priority: Critical
> Attachments: HADOOP-14989.test.patch
>
>
> While doing some digging in the metrics2 system recently, we noticed that the
> way {{MutableStat}} values are collected (and thus {{MutableRate}}, since it
> is based off of {{MutableStat}}) mean that each sink configured (including
> JMX) only receives a portion of the average information.
> {{MutableStat}}, to compute its average value, maintains a total value since
> last snapshot, as well as operation count since last snapshot. Upon
> snapshotting, the average is calculated as (total / opCount) and placed into
> a gauge metric, and total / operation count are cleared. So the average value
> represents the average since the last snapshot. If only a single sink ever
> snapshots, this would result in the expected behavior that the value is the
> average over the reporting period. However, if multiple sinks are configured,
> or if the JMX cache is refreshed, this is another snapshot operation. So, for
> example, if you have a FileSink configured at a 60 second interval and your
> JMX cache refreshes itself 1 second before the FileSink period fires, the
> values emitted to your FileSink only represent averages _over the last one
> second_.
> A few ways to solve this issue:
> * From an operator perspective, ensure only one sink is configured. This is
> not realistic given that the JMX cache exhibits the same behavior.
> * Make {{MutableRate}} manage its own average refresh, similar to
> {{MutableQuantiles}}, which has a refresh thread and saves a snapshot of the
> last quantile values that it will serve up until the next refresh. Given how
> many {{MutableRate}} metrics there are, a thread per metric is not really
> feasible, but could be done on e.g. a per-source basis. This has some
> downsides: if multiple sinks are configured with different periods, what is
> the right refresh period for the {{MutableRate}}?
> * Make {{MutableRate}} emit two counters, one for total and one for operation
> count, rather than an average gauge and an operation count counter. The
> average could then be calculated downstream from this information. This is
> cumbersome for operators and not backwards compatible. To improve on both of
> those downsides, we could have it keep the current behavior but
> _additionally_ emit the total as a counter. The snapshotted average is
> probably sufficient in the common case (we've been using it for years), a