Skip to content

CPU and process util stats - #7420

Merged
babsingh merged 1 commit into
eclipse-omr:masterfrom
amicic:cpu_and_process_stats
Jul 24, 2024
Merged

babsingh merged 1 commit into
eclipse-omr:masterfrom
amicic:cpu_and_process_stats

Conversation

@amicic

@amicic amicic commented Jul 22, 2024 •

Copy link
Copy Markdown
Contributor

They are measured between GC STW points, excluding GC STW points
themselves. It is reported at GC start point in Verbose GC, but there
are also plans to use it for some future heuristics.

A sample report line, showing how and where it's reported:

<gc-start id="190" type="scavenge" contextid="189" timestamp="2024-07-22T09:17:33.286">
  <cpu-util id="191" total="80.44" process="68.95" />
  <mem-info id="192" free="1438506168" total="10737418240" percent="13">

@amicic amicic added the comp:gc label Jul 22, 2024
@amicic
amicic force-pushed the cpu_and_process_stats branch 8 times, most recently from bf49609 to 9824b51 Compare July 22, 2024 18:00
@amicic amicic changed the title WIP: CPU and process util stats CPU and process util stats Jul 22, 2024
@amicic
amicic force-pushed the cpu_and_process_stats branch 8 times, most recently from 22a67ae to 275db7b Compare July 22, 2024 19:23
They are measured between GC STW points, excluding GC STW points
themselves.

They are reported at GC start point in Verbose GC as percentages, but
there are also plans to use them for some future heuristics.

A sample report line, showing how and where it's reported:

```
<gc-start id="190" type="scavenge" contextid="189"
timestamp="2024-07-22T09:17:33.286">
  <cpu-util id="191" total="80.44" process="68.95" />
  <mem-info id="192" free="1438506168" total="10737418240" percent="13">
```

Signed-off-by: Aleksandar Micic <Aleksandar_Micic@ca.ibm.com>
@amicic
amicic force-pushed the cpu_and_process_stats branch from 275db7b to 21103f6 Compare July 22, 2024 19:26
@dmitripivkine

Copy link
Copy Markdown
Contributor

@babsingh Would you please proceed?

@babsingh

Copy link
Copy Markdown
Contributor

jenkins build all

@babsingh babsingh left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Only known failures are seen.

@babsingh
babsingh merged commit 6002b4a into eclipse-omr:master Jul 24, 2024
Comment thread gc/verbose/schema.xsd
Comment on lines +367 to 372
<sequence maxOccurs="1" minOccurs="0">
<element ref="vgc:cpu-util" maxOccurs="1" minOccurs="0" />
</sequence>
<sequence maxOccurs="1" minOccurs="1">
<element ref="vgc:mem-info" maxOccurs="1" minOccurs="0" />
</sequence>

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

An OpenJ9 internal test (starting with build 75102) doesn't appear to like two <sequence> elements here. Perhaps this should be:

		<sequence maxOccurs="1" minOccurs="0">
			<element ref="vgc:cpu-util" maxOccurs="1" minOccurs="0" />
			<element ref="vgc:mem-info" maxOccurs="1" minOccurs="0" />
		</sequence>

amicic pushed a commit to amicic/omr that referenced this pull request Aug 1, 2024
Seems to cause regression, and since investigation is not trivial and
revert of the original change eclipse-omr#7420
would drag 3 more PRs to be reverted this is just a partial revert, that
basically gets the same run-time behavior, although it does not revert
some other code this will not be used/invoked.

Signed-off-by: Aleksandar Micic <Aleksandar_Micic@ca.ibm.com>
amicic pushed a commit to amicic/omr that referenced this pull request Aug 1, 2024
eclipse-omr#7420 seems to cause perf regression,
and since investigation is not trivial and revert of the original change
would drag 3 more PRs to be reverted this is just a partial revert, that
basically gets the same run-time behavior, although it does not revert
some other code this will not be used/invoked.

Signed-off-by: Aleksandar Micic <Aleksandar_Micic@ca.ibm.com>
amicic pushed a commit to amicic/omr that referenced this pull request Aug 1, 2024
eclipse-omr#7420 seems to cause perf regression,
and since investigation is not trivial and revert of the original change
would drag 3 more PRs to be reverted this is just a partial revert, that
basically gets the same run-time behavior, although it does not revert
some other code this will not be used/invoked.

Signed-off-by: Aleksandar Micic <Aleksandar_Micic@ca.ibm.com>
amicic pushed a commit to amicic/omr that referenced this pull request Aug 1, 2024
eclipse-omr#7420 seems to cause perf regression,
and since investigation is not trivial and revert of the original change
would drag 3 more PRs to be reverted this is just a partial revert, that
basically gets the same run-time behavior, although it does not revert
some other code this will not be used/invoked.

Signed-off-by: Aleksandar Micic <Aleksandar_Micic@ca.ibm.com>
pshipton pushed a commit to pshipton/openj9-omr that referenced this pull request Aug 1, 2024
eclipse-omr/omr#7420 seems to cause perf regression,
and since investigation is not trivial and revert of the original change
would drag 3 more PRs to be reverted this is just a partial revert, that
basically gets the same run-time behavior, although it does not revert
some other code this will not be used/invoked.

Signed-off-by: Aleksandar Micic <Aleksandar_Micic@ca.ibm.com>
amicic pushed a commit to amicic/omr that referenced this pull request Oct 16, 2024
eclipse-omr#7420 seems to cause perf regression,
and since investigation is not trivial and revert of the original change
would drag 3 more PRs to be reverted this is just a partial revert, that
basically gets the same run-time behavior, although it does not revert
some other code this will not be used/invoked.

Signed-off-by: Aleksandar Micic <Aleksandar_Micic@ca.ibm.com>
amicic pushed a commit to amicic/omr that referenced this pull request Oct 16, 2024
eclipse-omr#7420 seems to cause perf regression,
and since investigation is not trivial and revert of the original change
would drag 3 more PRs to be reverted this is just a partial revert, that
basically gets the same run-time behavior, although it does not revert
some other code this will not be used/invoked.

Signed-off-by: Aleksandar Micic <Aleksandar_Micic@ca.ibm.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants