From: Etienne Cordonnier <ecordonnier@snap.com>
To: Martin Jansa <martin.jansa@gmail.com>
Cc: randy.macleod@windriver.com,
Richard Purdie <richard.purdie@linuxfoundation.org>,
bitbake-devel@lists.openembedded.org
Subject: Re: [bitbake-devel] [PATCH] runqueue: Add pressure change logging
Date: Wed, 10 Jan 2024 18:14:16 +0100 [thread overview]
Message-ID: <CAHUKmYbY9RjgQGOxQEZwMyykAGUmkamQxhY0P=aheP1u1J4o+Q@mail.gmail.com> (raw)
In-Reply-To: <CA+chaQfuyyOACJ1SL6gu=0ASnzqgD++Q73H2+ibbZRA4YsLBwg@mail.gmail.com>
[-- Attachment #1: Type: text/plain, Size: 7464 bytes --]
@Randy: "Also, despite the logs, do you find that pressure regulation is
useful for you?"
-> yes, testing a while back showed that setting this value to 500 reduced
the clean-build durations of several minutes, which is why I added this
configuration a year ago
@Martin: thanks for the info! I indeed configured the value to 500 after
doing some testing while I was on kirkstone 4.0.4 (long before
https://github.com/yoctoproject/poky/commit/a76bc698c43a3e9792ec0ff8c97c446a20c1439e
which was added in 4.0.15), and missed that I needed to change the value
after updating to 4.0.15. I'll test with higher values
On Wed, Jan 10, 2024 at 5:47 PM Martin Jansa <martin.jansa@gmail.com> wrote:
> I think your limit 500 is just too low after:
> https://lists.openembedded.org/g/bitbake-devel/message/15499
> <https://urldefense.proofpoint.com/v2/url?u=https-3A__lists.openembedded.org_g_bitbake-2Ddevel_message_15499&d=DwMFaQ&c=ncDTmphkJTvjIDPh0hpF_4vCHvabgGkICC2epckfdiw&r=AhkbNonVuMIGRfPx_Qj9TsRih1DULJTKUkSGa66m67E&m=XGfSv6zHSImmkkXtVGhYi8KvF7P-URlnWSS4yjgh00mSKcgkdUb_w74W6PwFLlOg&s=B4aqICG4GUIKccICnUztvgZdL24Ztt98IIZMsBNoUZk&e=>
> was backported (already included in kirkstone-4.0.15).
>
> FWIW: on 64 thread AMD Threadripper 3970X I'm using much higher CPU_MAX
> limit and much lower bitbake thread count and still trigger these messages
> (but I still find them useful).
>
> This was just "small incremental build" of "bitbake -k lib32-kmscube
> lib32-wavpack":
>
> Sstate summary: Wanted 62 Local 0 Mirrors 0 Missed 62 Current 431 (0%
> match, 87% complete)
> Removing 48 stale sstate objects for arch mach: 100%
> Removing 1 stale sstate objects for arch x86_64: 100%
> NOTE: Executing Tasks
> NOTE: Pressure status changed to CPU: True, IO: None, Mem: None (CPU:
> 533768.7/200000.0, IO: 3306.5/None, Mem: 2611.1/None) - using 2/8 bitbake
> threads
> NOTE: Pressure status changed to CPU: False, IO: None, Mem: None (CPU:
> 172224.8/200000.0, IO: 9830.9/None, Mem: 0.0/None) - using 2/8 bitbake
> threads
> NOTE: Pressure status changed to CPU: True, IO: None, Mem: None (CPU:
> 426774.0/200000.0, IO: 3229.3/None, Mem: 0.0/None) - using 2/8 bitbake
> threads
> NOTE: Pressure status changed to CPU: False, IO: None, Mem: None (CPU:
> 170159.7/200000.0, IO: 798.0/None, Mem: 0.0/None) - using 2/8 bitbake
> threads
> NOTE: Pressure status changed to CPU: True, IO: None, Mem: None (CPU:
> 201391.4/200000.0, IO: 132.8/None, Mem: 0.0/None) - using 2/8 bitbake
> threads
> NOTE: Pressure status changed to CPU: False, IO: None, Mem: None (CPU:
> 159116.4/200000.0, IO: 0.0/None, Mem: 0.0/None) - using 2/8 bitbake threads
> NOTE: Pressure status changed to CPU: True, IO: None, Mem: None (CPU:
> 210248.5/200000.0, IO: 270.8/None, Mem: 0.0/None) - using 3/8 bitbake
> threads
> NOTE: Pressure status changed to CPU: False, IO: None, Mem: None (CPU:
> 198485.4/200000.0, IO: 1577.3/None, Mem: 0.0/None) - using 3/8 bitbake
> threads
> NOTE: Pressure status changed to CPU: True, IO: None, Mem: None (CPU:
> 219595.7/200000.0, IO: 3824.7/None, Mem: 0.0/None) - using 3/8 bitbake
> threads
> NOTE: Pressure status changed to CPU: False, IO: None, Mem: None (CPU:
> 199729.1/200000.0, IO: 1121.1/None, Mem: 0.0/None) - using 3/8 bitbake
> threads
> NOTE: Pressure status changed to CPU: True, IO: None, Mem: None (CPU:
> 304074.6/200000.0, IO: 5498.7/None, Mem: 44.2/None) - using 3/8 bitbake
> threads
> NOTE: Pressure status changed to CPU: False, IO: None, Mem: None (CPU:
> 194765.9/200000.0, IO: 2600.7/None, Mem: 0.0/None) - using 3/8 bitbake
> threads
> NOTE: Pressure status changed to CPU: True, IO: None, Mem: None (CPU:
> 201263.5/200000.0, IO: 2437.4/None, Mem: 0.0/None) - using 4/8 bitbake
> threads
> NOTE: Pressure status changed to CPU: False, IO: None, Mem: None (CPU:
> 195733.7/200000.0, IO: 2454.0/None, Mem: 0.0/None) - using 4/8 bitbake
> threads
> NOTE: Tasks Summary: Attempted 1473 tasks of which 1335 didn't need to be
> rerun and all succeeded.
>
> On Wed, Jan 10, 2024 at 4:28 PM Randy MacLeod via lists.openembedded.org
> <https://urldefense.proofpoint.com/v2/url?u=http-3A__lists.openembedded.org&d=DwMFaQ&c=ncDTmphkJTvjIDPh0hpF_4vCHvabgGkICC2epckfdiw&r=AhkbNonVuMIGRfPx_Qj9TsRih1DULJTKUkSGa66m67E&m=XGfSv6zHSImmkkXtVGhYi8KvF7P-URlnWSS4yjgh00mSKcgkdUb_w74W6PwFLlOg&s=1CqYM87mmPOKlTX29Ag8j0-3WXx1ZhPx0vB9ZjrdApo&e=>
> <randy.macleod=windriver.com@lists.openembedded.org> wrote:
>
>> Hi Etienne,
>>
>> On 2024-01-10 9:09 a.m., Etienne Cordonnier wrote:
>>
>> "How many physical/virtual cores are actually available?
>>
>> Is some other activity doing CPU intensive work on the system?
>> Is the one task running > 48 compile jobs and causing CPU contention in
>> the system?"
>>
>> -> This is a physical PC with 24 physical cores and 48 cores when you
>> take hyperthreading into account (Intel Xeon Gold 6240R)
>>
>> Thanks.
>>
>> -> there is no other activity doing CPU intensive work
>> -> some bitbake tasks can easily load 48 cores, yes (this is a custom
>> image with some extra layers added)
>>
>>
>> I see, then other than deferring new tasks, there's not much that
>> pressure regulation can do to help once one of these
>> CPU intensive tasks is started.
>>
>> Other than the log verbosity which we likely won't change for a while,
>> did you have any PSI-related concerns?
>>
>> I did notice that your last log was:
>> > NOTE: Pressure status changed to CPU: True, IO: None, Mem: None (CPU:
>> 21922.9/500.0, IO: 10137.8/None, Mem: 194.8/None) - using 1/48 bitbake
>> threads
>>
>> When there's memory pressure, that's usually an indication that the
>> system is paging to disk.
>>
>>
>> Also, despite the logs, do you find that pressure regulation is useful
>> for you?
>> You can always leave it turned off and only rely on the task and job
>> limits.
>>
>> Thanks,
>>
>> ../Randy
>>
>>
>>
>> Etienne
>>
>>
>> --
>> # Randy MacLeod
>> # Wind River Linux
>>
>>
>> -=-=-=-=-=-=-=-=-=-=-=-
>> Links: You receive all messages sent to this group.
>> View/Reply Online (#15779):
>> https://lists.openembedded.org/g/bitbake-devel/message/15779
>> <https://urldefense.proofpoint.com/v2/url?u=https-3A__lists.openembedded.org_g_bitbake-2Ddevel_message_15779&d=DwMFaQ&c=ncDTmphkJTvjIDPh0hpF_4vCHvabgGkICC2epckfdiw&r=AhkbNonVuMIGRfPx_Qj9TsRih1DULJTKUkSGa66m67E&m=XGfSv6zHSImmkkXtVGhYi8KvF7P-URlnWSS4yjgh00mSKcgkdUb_w74W6PwFLlOg&s=o0eGuVu9EDjTnjX6J0BuJ1YzBW3dlc3UVSYYslDY0gs&e=>
>> Mute This Topic: https://lists.openembedded.org/mt/100117828/3617156
>> <https://urldefense.proofpoint.com/v2/url?u=https-3A__lists.openembedded.org_mt_100117828_3617156&d=DwMFaQ&c=ncDTmphkJTvjIDPh0hpF_4vCHvabgGkICC2epckfdiw&r=AhkbNonVuMIGRfPx_Qj9TsRih1DULJTKUkSGa66m67E&m=XGfSv6zHSImmkkXtVGhYi8KvF7P-URlnWSS4yjgh00mSKcgkdUb_w74W6PwFLlOg&s=RWBsbi9oTfLo48SrGXNbjtF13PxjrJX1y-PoEimQFS4&e=>
>> Group Owner: bitbake-devel+owner@lists.openembedded.org
>> Unsubscribe: https://lists.openembedded.org/g/bitbake-devel/unsub
>> <https://urldefense.proofpoint.com/v2/url?u=https-3A__lists.openembedded.org_g_bitbake-2Ddevel_unsub&d=DwMFaQ&c=ncDTmphkJTvjIDPh0hpF_4vCHvabgGkICC2epckfdiw&r=AhkbNonVuMIGRfPx_Qj9TsRih1DULJTKUkSGa66m67E&m=XGfSv6zHSImmkkXtVGhYi8KvF7P-URlnWSS4yjgh00mSKcgkdUb_w74W6PwFLlOg&s=bpBRZ_4shRibijymE6MJZi-SHIkHCZMsixgigmP-rJ0&e=>
>> [martin.jansa@gmail.com]
>> -=-=-=-=-=-=-=-=-=-=-=-
>>
>>
[-- Attachment #2: Type: text/html, Size: 9442 bytes --]
prev parent reply other threads:[~2024-01-10 17:14 UTC|newest]
Thread overview: 9+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-07-13 10:53 [PATCH] runqueue: Add pressure change logging Richard Purdie
2024-01-09 14:09 ` [bitbake-devel] " Etienne Cordonnier
2024-01-09 14:12 ` Richard Purdie
2024-01-09 21:49 ` Randy MacLeod
2024-01-10 2:20 ` Randy MacLeod
2024-01-10 14:09 ` Etienne Cordonnier
2024-01-10 15:27 ` Randy MacLeod
2024-01-10 16:46 ` Martin Jansa
2024-01-10 17:14 ` Etienne Cordonnier [this message]
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to='CAHUKmYbY9RjgQGOxQEZwMyykAGUmkamQxhY0P=aheP1u1J4o+Q@mail.gmail.com' \
--to=ecordonnier@snap.com \
--cc=bitbake-devel@lists.openembedded.org \
--cc=martin.jansa@gmail.com \
--cc=randy.macleod@windriver.com \
--cc=richard.purdie@linuxfoundation.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).