omrsysinfo_get_CPU_load() return a new error for insufficient data #7189

thallium · 2023-11-22T15:36:10Z

This is to determine if a call to omrsysinfo_get_CPU_load has only
one data point recorded.

A new error, OMRPORT_ERROR_INSUFFICIENT_DATA, is returned
in this scenario.

Also fixed a mistake in the description saying that the first two
invocations return OMRPORT_ERROR_OPFAILED and fix the time unit in the
description to ms.

Related: eclipse-openj9/openj9#18451

thallium · 2023-11-22T15:36:24Z

@tajila FYI

port/unix/omrsysinfo.c

github-actions

Thank you for supporting the project, and congratulations on your first contribution! A project committer will shortly review your contribution. In the mean time, if you haven't had a chance please skim over the contribution guidelines which all pull requests must adhere to. If the ECA pull request check fails, have a look at the instructions for signing the ECA in the legal considerations section.

If you run into any problems our community will be happy to assist you in any way we can. There are a number of recommended ways to interact with the community. We encourage you to ask questions, or drop by to say hello.

tajila · 2023-11-22T18:55:51Z

@babsingh Please review these changes

babsingh · 2023-11-22T20:13:39Z

jenkins build all

babsingh · 2023-11-22T21:52:52Z

Currently, this PR only updates the unix impl. The win32 impl also needs to be updated:
https://github.com/eclipse/omr/blob/8b19b8082de56b067c23bafd59dd767a58d8c02f/port/win32/omrsysinfo.c#L1271-L1275

babsingh · 2023-11-22T21:55:23Z

Also, the function description needs to be updated to reflect the new error handling:
https://github.com/eclipse/omr/blob/8b19b8082de56b067c23bafd59dd767a58d8c02f/port/common/omrsysinfo.c#L643-L663

babsingh · 2023-11-22T22:16:58Z

There is also the zos impl, which needs to be updated:
https://github.com/eclipse/omr/blob/8b19b8082de56b067c23bafd59dd767a58d8c02f/port/unix/omrsysinfo.c#L4671-L4678

thallium · 2023-11-23T03:54:20Z

Also, the function description needs to be updated to reflect the new error handling:

https://github.com/eclipse/omr/blob/8b19b8082de56b067c23bafd59dd767a58d8c02f/port/common/omrsysinfo.c#L643-L663

Why does it says "on the first two invocations"? Is that a typo?

babsingh · 2023-11-23T15:38:20Z

Why does it says "on the first two invocations"? Is that a typo?

I am unable to confirm by just reading the below design and implementation; it doesn't seem straightforward.

Design/requirements: WIP: Enable System CPU info for z/OS #3152 (comment)
Implementation (4 commits): Add new omrsysinfo_get_CPU_load API #4702

@thallium Can you verify by adding print statements for the return value in the existing test (see below)? I can run PR builds on all platforms once you have added debug code in the test. on the first two invocations: might be platform specific.

https://github.com/eclipse/omr/blob/8b19b8082de56b067c23bafd59dd767a58d8c02f/fvtest/porttest/si.cpp#L1689-L1716

babsingh · 2023-11-23T15:50:47Z

@fjeremic For #7189 (comment), do you remember why the second invocation of omrsysinfo_get_CPU_load also returns OMRPORT_ERROR_OPFAILED?

thallium · 2023-11-23T21:29:51Z

@babsingh I've added the debug output, can you run the test?

babsingh · 2023-11-23T22:14:38Z

jenkins build all(test:'-R porttest',env:'GTEST_FILTER=PortSysinfoTest.sysinfo_test_get_CPU_load',compile:'-j8')

If the printf output is not seen, then the log methods in the test suite will need to be used.

babsingh · 2023-11-24T05:20:17Z

jenkins build all(test:'-R porttest',env:'GTEST_FILTER=PortSysinfoTest.sysinfo_test_get_CPU_load')

babsingh · 2023-11-24T14:21:19Z

Windows compilation failed:

00:25:44      53>c:\omr\workspace\pullrequest-win_x86-64\build\fvtest\porttest\si.cpp(1705): error C2220: warning treated as error - no 'object' file generated [C:\omr\workspace\PullRequest-win_x86-64\Build\build\fvtest\porttest\omrporttest.vcxproj]
00:25:44      53>c:\omr\workspace\pullrequest-win_x86-64\build\fvtest\porttest\si.cpp(1705): warning C4244: '=': conversion from 'intptr_t' to 'int', possible loss of data [C:\omr\workspace\PullRequest-win_x86-64\Build\build\fvtest\porttest\omrporttest.vcxproj]
00:25:44      53>c:\omr\workspace\pullrequest-win_x86-64\build\fvtest\porttest\si.cpp(1707): warning C4244: '=': conversion from 'intptr_t' to 'int', possible loss of data [C:\omr\workspace\PullRequest-win_x86-64\Build\build\fvtest\porttest\omrporttest.vcxproj]

Test output:

00:43:49  15: Return value of first invocation: -21
00:43:49  15: Return value of second invocation: -1

I only checked a few build jobs. I consistently saw the above test output. @thallium Verify the test output in the PR builds as well. Fix the Windows compilation issue and verify the test output on Windows.

babsingh · 2023-11-24T16:23:35Z

The test output for Windows is also the same:

00:43:49  15: Return value of first invocation: -21
00:43:49  15: Return value of second invocation: -1

The second OMRPORT_ERROR_OPFAILED is probably returned from the end of the function. @tajila I don't believe the current changes are sufficient to support eclipse-openj9/openj9#18451. Based upon the above behaviour, there is still ambiguity between the insufficient data and insufficient time cases.

tajila · 2023-12-04T22:46:46Z

Intent of eclipse-openj9/openj9#18451: Make get[Process|System]CpuLoad() return 0 on the first call.

I don't think this is fully accurate. The real intent is to return zero if we have one data point instead of -1 as the RI does. This is usually the case on the first call.

can the RI still throw an error in the first call?

Im not certain, ubt I would expect that to be the case. We can try to verify this. For example, if /proc/stat was not reachable then I would expect RI to fail as well and not return zero.

If RI doesn't do this, then they are losing information, because returning 0 when there is a hard failure means there is no way to tell that there is a functional issue. In which case, I would still be okay with the behaviour being proposed here.

babsingh · 2023-12-05T03:53:45Z

@thallium Please post an update in the PR once the below tasks are completed.

Update the documentation to reflect .... The real intent is to return zero if we have one data point instead of -1 as the RI does.
Verify RI's behaviour if /proc/stat is not reachable.

thallium · 2023-12-05T15:28:56Z

Verify RI's behaviour if /proc/stat is not reachable.

RI returns -1 in this situation.

babsingh · 2023-12-06T15:25:57Z

Thanks @thallium for verifying RI's behaviour. Can you also update the commit message to reflect the new documentation?

Return another error if only one data point has been recorded

This is to determine if a call to omrsysinfo_get_CPU_load has only
one data point recorded.

...

babsingh · 2023-12-06T15:47:20Z

new line (return) is missing from the first line.

babsingh · 2023-12-06T15:54:49Z

still a space is missing onlyone -> only one. i have updated the PR description for you.

This is to determine if a call to omrsysinfo_get_CPU_load has only one data point recorded. Also fixed a mistake in the description saying that the first two invocations return OMRPORT_ERROR_OPFAILED and fix the time unit in the description to "ms". Related: eclipse-openj9/openj9#18451 Signed-off-by: Gengchen Tuo <gengchen.tuo@ibm.com>

babsingh · 2023-12-06T16:01:24Z

jenkins build all

babsingh · 2023-12-06T16:26:18Z

jenkins build aix

babsingh · 2023-12-06T19:27:17Z

Only a known issue is observed:

OSX PR builds: OMR Socket Test Failures #6516

Currently, getProcessCpuLoad() and getSystemCpuLoad() return -1 if only one data point has been recorded. A compatibility flag -XX:[+/-]CpuLoadCompatibility is added to match the behaviour of the RI, which is to return 0. Fixes: eclipse-openj9#13389 Related: eclipse-omr/omr#7189 Signed-off-by: Gengchen Tuo <gengchen.tuo@ibm.com>

thallium requested review from charliegracie, youngar and babsingh as code owners November 22, 2023 15:36

thallium mentioned this pull request Nov 22, 2023

Add a flag to make get[Process|System]CpuLoad() match RI behaviour eclipse-openj9/openj9#18451

Merged

github-actions bot added the comp:port label Nov 22, 2023

tajila reviewed Nov 22, 2023

View reviewed changes

port/unix/omrsysinfo.c Outdated Show resolved Hide resolved

github-actions bot reviewed Nov 22, 2023

View reviewed changes

github-actions bot added the first contribution label Nov 22, 2023

thallium force-pushed the master branch from 5058194 to a8d230b Compare November 22, 2023 16:21

thallium requested review from mstoodle and 0xdaryl as code owners November 22, 2023 16:21

tajila approved these changes Nov 22, 2023

View reviewed changes

thallium force-pushed the master branch from a8d230b to bcf6dfe Compare November 23, 2023 21:28

thallium force-pushed the master branch from bcf6dfe to 4e03fdf Compare November 24, 2023 15:27

thallium force-pushed the master branch from a18b161 to 48c46df Compare December 5, 2023 15:28

thallium force-pushed the master branch 2 times, most recently from 4897e4a to 5c4e691 Compare December 6, 2023 15:32

babsingh changed the title ~~Make omrsysinfo_get_CPU_load() return another error code on first call~~ omrsysinfo_get_CPU_load() return a new error for insufficient data Dec 6, 2023

thallium force-pushed the master branch from 5c4e691 to f5607db Compare December 6, 2023 15:50

thallium force-pushed the master branch from f5607db to 6071008 Compare December 6, 2023 15:56

babsingh approved these changes Dec 6, 2023

View reviewed changes

babsingh merged commit e5e7d55 into eclipse-omr:master Dec 6, 2023
16 of 18 checks passed

thallium mentioned this pull request Dec 15, 2023

Add a flag to return 0 if only one data point has been recorded eclipse-openj9/openj9#18636

Merged

thallium mentioned this pull request Dec 17, 2023

(v0.43) Add a flag to return 0 if only one data point has been recorded eclipse-openj9/openj9#18639

Merged

pshipton mentioned this pull request Dec 17, 2023

(v0.43) Return another error if only one data point has been recorded eclipse-openj9/openj9-omr#189

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

omrsysinfo_get_CPU_load() return a new error for insufficient data #7189

omrsysinfo_get_CPU_load() return a new error for insufficient data #7189

thallium commented Nov 22, 2023 •

edited by babsingh

Loading

thallium commented Nov 22, 2023

github-actions bot left a comment

tajila commented Nov 22, 2023

babsingh commented Nov 22, 2023

babsingh commented Nov 22, 2023

babsingh commented Nov 22, 2023

babsingh commented Nov 22, 2023

thallium commented Nov 23, 2023

babsingh commented Nov 23, 2023 •

edited

Loading

babsingh commented Nov 23, 2023 •

edited

Loading

thallium commented Nov 23, 2023

babsingh commented Nov 23, 2023

babsingh commented Nov 24, 2023

babsingh commented Nov 24, 2023

babsingh commented Nov 24, 2023 •

edited

Loading

tajila commented Dec 4, 2023

babsingh commented Dec 5, 2023 •

edited

Loading

thallium commented Dec 5, 2023

babsingh commented Dec 6, 2023

babsingh commented Dec 6, 2023

babsingh commented Dec 6, 2023 •

edited

Loading

babsingh commented Dec 6, 2023

babsingh commented Dec 6, 2023

babsingh commented Dec 6, 2023

omrsysinfo_get_CPU_load() return a new error for insufficient data #7189

omrsysinfo_get_CPU_load() return a new error for insufficient data #7189

Conversation

thallium commented Nov 22, 2023 • edited by babsingh Loading

thallium commented Nov 22, 2023

github-actions bot left a comment

Choose a reason for hiding this comment

tajila commented Nov 22, 2023

babsingh commented Nov 22, 2023

babsingh commented Nov 22, 2023

babsingh commented Nov 22, 2023

babsingh commented Nov 22, 2023

thallium commented Nov 23, 2023

babsingh commented Nov 23, 2023 • edited Loading

babsingh commented Nov 23, 2023 • edited Loading

thallium commented Nov 23, 2023

babsingh commented Nov 23, 2023

babsingh commented Nov 24, 2023

babsingh commented Nov 24, 2023

babsingh commented Nov 24, 2023 • edited Loading

tajila commented Dec 4, 2023

babsingh commented Dec 5, 2023 • edited Loading

thallium commented Dec 5, 2023

babsingh commented Dec 6, 2023

babsingh commented Dec 6, 2023

babsingh commented Dec 6, 2023 • edited Loading

babsingh commented Dec 6, 2023

babsingh commented Dec 6, 2023

babsingh commented Dec 6, 2023

thallium commented Nov 22, 2023 •

edited by babsingh

Loading

babsingh commented Nov 23, 2023 •

edited

Loading

babsingh commented Nov 23, 2023 •

edited

Loading

babsingh commented Nov 24, 2023 •

edited

Loading

babsingh commented Dec 5, 2023 •

edited

Loading

babsingh commented Dec 6, 2023 •

edited

Loading