[
https://issues.apache.org/jira/browse/FLINK-14164?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16977102#comment-16977102
]
Zhu Zhu commented on FLINK-14164:
---------------------------------
Hi [~stevenz3wu], we have to inform you that `numberOfRestarts` is added as a
gauge rather than a meter.
This is because we found the meter can be inaccurate if the measured events
happen in a very low frequency (see
[discussion|https://github.com/apache/flink/pull/10082#discussion_r343562150]).
This meter can be only used to build alerts "restarts > 0" and would not be
able to accurately trigger other alerts, like "restarts > 10 in the past hour".
So it's not good to add it as a meter.
The current idea is, it would be better to have users to use time-series
databases that can derive the rate in whatever granularity they desire, thus to
build flexible and accurate monitoring/alerting for low frequency events.
Would that work for you? Feel free to share your concerns.
> Add a metric to show failover count regarding fine grained recovery
> -------------------------------------------------------------------
>
> Key: FLINK-14164
> URL: https://issues.apache.org/jira/browse/FLINK-14164
> Project: Flink
> Issue Type: Sub-task
> Components: Runtime / Coordination, Runtime / Metrics
> Affects Versions: 1.10.0
> Reporter: Zhu Zhu
> Assignee: Zhu Zhu
> Priority: Major
> Labels: pull-request-available
> Fix For: 1.10.0
>
> Time Spent: 40m
> Remaining Estimate: 0h
>
> Previously Flink uses restart all strategy to recover jobs from failures. And
> the metric "fullRestart" is used to show the count of failovers.
> However, with fine grained recovery introduced in 1.9.0, the "fullRestart"
> metric only reveals how many times the entire graph has been restarted, not
> including the number of fine grained failure recoveries.
> As many users want to build their job alerting based on failovers, I'd
> propose to add such a new metric {{numberOfRestarts}} which also respects
> fine grained recoveries. The metric should be a Gauge.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
