[
https://issues.apache.org/jira/browse/SLING-6698?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Georg Henzler updated SLING-6698:
---------------------------------
Description:
This health check should use java.lang.management.ThreadMXBean to provide the
following:
* A quick overview what an instance is doing via http (there are many other
means of monitoring, but there is nothing easier than using the HC servlet).
This works well as all relevant tasks in sling (requests, jobs, schedulers) set
the thread name to a meaningful name. The list of threads shall be sorted from
busiest to idlest and by default only the top 3 are shown (top 10 if CPU is >
15%). Using hcDebug=true will show all threads.
* It should WARN if all threads are highly utilised (but never send CRITICAL as
a fully utilised instance is not necessarily a problem)
* It should use the deadlock detection of ThreadMXBean to send CRITICAL
was:
This health check should use java.lang.management.ThreadMXBean to provide the
following:
* A quick overview what an instance is doing via http (there are many other
means of monitoring, but there is nothing easier than using the the HC
servlet). This works well as all relevant tasks in sling (requests, jobs,
schedulers) set the thread name to a meaningful name. The list of threads shall
be sorted from busiest to idlest and by default only the top 3 are shown (top
10 if CPU is > 15%). Using hcDebug=true will show all threads.
* It should WARN if all threads are highly utilised (but never send CRITICAL as
a fully utilised instance is not necessarily a problem)
* It should use the deadlock detection of ThreadMXBean of send CRITICAL
> Thread Usage Health Check
> -------------------------
>
> Key: SLING-6698
> URL: https://issues.apache.org/jira/browse/SLING-6698
> Project: Sling
> Issue Type: New Feature
> Components: Health Check
> Reporter: Georg Henzler
> Assignee: Georg Henzler
> Fix For: Health Check Support 1.0.6
>
> Attachments: threads-hc-idle.png, threads-hc-warn.png
>
>
> This health check should use java.lang.management.ThreadMXBean to provide the
> following:
> * A quick overview what an instance is doing via http (there are many other
> means of monitoring, but there is nothing easier than using the HC servlet).
> This works well as all relevant tasks in sling (requests, jobs, schedulers)
> set the thread name to a meaningful name. The list of threads shall be sorted
> from busiest to idlest and by default only the top 3 are shown (top 10 if CPU
> is > 15%). Using hcDebug=true will show all threads.
> * It should WARN if all threads are highly utilised (but never send CRITICAL
> as a fully utilised instance is not necessarily a problem)
> * It should use the deadlock detection of ThreadMXBean to send CRITICAL
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
