[
https://issues.apache.org/jira/browse/MESOS-7819?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16127362#comment-16127362
]
Benno Evers commented on MESOS-7819:
------------------------------------
For the metrics where we think they might be occasionally useful for debugging
but are worried about exposing too much internal state (points 1,2,5), maybe
another idea would be to introduce something like private metrics, which would
essentially be something like a {{volatile static int64_t}} (so all
modifications are preserved even at high optimization levels, but the only way
to actually see the value would be through a debugger)
Some thoughts about the individual proposed metrics, it seems to me like any
single one wouldn't be very useful because it's hard to say in isolation how
many actors/connections/messages are "normal" for the different parts of mesos,
but having multiple of them it would become possible to compare their ratios to
known "normal" ranges and maybe pinpoint the fault location more precisely.
In particular, average number of pending messages might be useful not only for
debugging but also for performance regression tests in the future.
> Libprocess internal state is not monitored by metrics.
> ------------------------------------------------------
>
> Key: MESOS-7819
> URL: https://issues.apache.org/jira/browse/MESOS-7819
> Project: Mesos
> Issue Type: Improvement
> Components: libprocess
> Reporter: Alexander Rukletsov
> Labels: metrics, newbie++
>
> Libprocess does not expose its internal state via metrics. Active sockets,
> number of HTTP proxies, number of running actors, number of pending messages
> for all active sockets, etc — may be of interest when monitoring and
> debugging Mesos clusters.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
