Quanta requests to create a repo in OpenBMC github

Quanta requests to create a repo in OpenBMC github

[Fran Hsu (徐誌謙)]

Hi Brad,
	This is Fran from Quanta Computer.
Quanta OpenBMC team would like to share some useful tools to the OpenBMC community.
Such as the read-margin-temp daemon (https://github.com/quanta-bmc/read-margin-temp), we are using this daemon to fill sensor margin value to PID to do fan control.

We would like to have a repo named "quanta-misc" under the OpenBMC github.

Here is the maintainer list:
M:  Benjamin Fair <benjaminfair@google.com> <benjaminfair!>
M:  Brandon Kim <brandonkim@google.com> <brandonk!>
M:  Fran Hsu <fran.hsu@quantatw.com> <franhsu!>
M:  George Hung <george.hung@quantatw.com> <ghung!>
M:  Buddy Huang <buddy.huang@quantatw.com> <buddyhunag!>

Thank you.
Fran Hsu
E-Mail : Fran.Hsu@QuantaTW.com
Tel: +886-3-327-2345 Ext: 16935

Re: Quanta requests to create a repo in OpenBMC github

[Benjamin Fair]

On Wed, 16 Sep 2020 at 06:31, Fran Hsu (徐誌謙) <Fran.Hsu@quantatw.com> wrote:
>
> Hi Brad,
>         This is Fran from Quanta Computer.
> Quanta OpenBMC team would like to share some useful tools to the OpenBMC community.
> Such as the read-margin-temp daemon (https://github.com/quanta-bmc/read-margin-temp), we are using this daemon to fill sensor margin value to PID to do fan control.

read-margin-temp seems useful and widely applicable enough that I
think it's worth having a dedicated repo for it.

The quanta-misc repo could possibly be used for smaller, more
Quanta-specific things.

>
> We would like to have a repo named "quanta-misc" under the OpenBMC github.
>
> Here is the maintainer list:
> M:  Benjamin Fair <benjaminfair@google.com> <benjaminfair!>
> M:  Brandon Kim <brandonkim@google.com> <brandonk!>
> M:  Fran Hsu <fran.hsu@quantatw.com> <franhsu!>
> M:  George Hung <george.hung@quantatw.com> <ghung!>
> M:  Buddy Huang <buddy.huang@quantatw.com> <buddyhunag!>
>
> Thank you.
> Fran Hsu
> E-Mail : Fran.Hsu@QuantaTW.com
> Tel: +886-3-327-2345 Ext: 16935
>
>

Re: Quanta requests to create a repo in OpenBMC github

[Josh Lehan]

[-- Attachment #1: Type: text/plain, Size: 1618 bytes --]

Yay, this is great to see. I have some bug fixes to read-margin-temp and
was just wondering how to contribute them. I'm really hoping this makes it
into OpenBMC. It's the perfect preprocessor for phosphor-pid-control, I
have found.

Josh

Josh Lehan | Software Engineer | krellan@google.com | +1 650-733-8941



On Wed, Sep 16, 2020 at 7:15 PM Benjamin Fair <benjaminfair@google.com>
wrote:

> On Wed, 16 Sep 2020 at 06:31, Fran Hsu (徐誌謙) <Fran.Hsu@quantatw.com>
> wrote:
> >
> > Hi Brad,
> >         This is Fran from Quanta Computer.
> > Quanta OpenBMC team would like to share some useful tools to the OpenBMC
> community.
> > Such as the read-margin-temp daemon (
> https://github.com/quanta-bmc/read-margin-temp), we are using this daemon
> to fill sensor margin value to PID to do fan control.
>
> read-margin-temp seems useful and widely applicable enough that I
> think it's worth having a dedicated repo for it.
>
> The quanta-misc repo could possibly be used for smaller, more
> Quanta-specific things.
>
> >
> > We would like to have a repo named "quanta-misc" under the OpenBMC
> github.
> >
> > Here is the maintainer list:
> > M:  Benjamin Fair <benjaminfair@google.com> <benjaminfair!>
> > M:  Brandon Kim <brandonkim@google.com> <brandonk!>
> > M:  Fran Hsu <fran.hsu@quantatw.com> <franhsu!>
> > M:  George Hung <george.hung@quantatw.com> <ghung!>
> > M:  Buddy Huang <buddy.huang@quantatw.com> <buddyhunag!>
> >
> > Thank you.
> > Fran Hsu
> > E-Mail : Fran.Hsu@QuantaTW.com
> > Tel: +886-3-327-2345 Ext: 16935 <+886%203%20327%202345>
> >
> >
>

[-- Attachment #2: Type: text/html, Size: 3468 bytes --]

Re: Quanta requests to create a repo in OpenBMC github

[Patrick Williams]

[-- Attachment #1: Type: text/plain, Size: 2211 bytes --]

Would someone mind giving a 1-2 sentence explaination of what "sensor
margin values" are?  I'm certainly not a thermal control expert.

I don't care if there are multiple implementations of similar
functionality, but there is also work going on for a repository called
'phosphor-virtual-sensors'.  What little I could understand of
read-margin-temp's role seems to somewhat overlap with that effort.

On Wed, Sep 16, 2020 at 10:37:14PM -0700, Josh Lehan wrote:
> Yay, this is great to see. I have some bug fixes to read-margin-temp and
> was just wondering how to contribute them. I'm really hoping this makes it
> into OpenBMC. It's the perfect preprocessor for phosphor-pid-control, I
> have found.
> 
> Josh
> 
> Josh Lehan | Software Engineer | krellan@google.com | +1 650-733-8941
> 
> 
> 
> On Wed, Sep 16, 2020 at 7:15 PM Benjamin Fair <benjaminfair@google.com>
> wrote:
> 
> > On Wed, 16 Sep 2020 at 06:31, Fran Hsu (徐誌謙) <Fran.Hsu@quantatw.com>
> > wrote:
> > >
> > > Hi Brad,
> > >         This is Fran from Quanta Computer.
> > > Quanta OpenBMC team would like to share some useful tools to the OpenBMC
> > community.
> > > Such as the read-margin-temp daemon (
> > https://github.com/quanta-bmc/read-margin-temp), we are using this daemon
> > to fill sensor margin value to PID to do fan control.
> >
> > read-margin-temp seems useful and widely applicable enough that I
> > think it's worth having a dedicated repo for it.
> >
> > The quanta-misc repo could possibly be used for smaller, more
> > Quanta-specific things.
> >
> > >
> > > We would like to have a repo named "quanta-misc" under the OpenBMC
> > github.
> > >
> > > Here is the maintainer list:
> > > M:  Benjamin Fair <benjaminfair@google.com> <benjaminfair!>
> > > M:  Brandon Kim <brandonkim@google.com> <brandonk!>
> > > M:  Fran Hsu <fran.hsu@quantatw.com> <franhsu!>
> > > M:  George Hung <george.hung@quantatw.com> <ghung!>
> > > M:  Buddy Huang <buddy.huang@quantatw.com> <buddyhunag!>
> > >
> > > Thank you.
> > > Fran Hsu
> > > E-Mail : Fran.Hsu@QuantaTW.com
> > > Tel: +886-3-327-2345 Ext: 16935 <+886%203%20327%202345>
> > >
> > >
> >

-- 
Patrick Williams

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: Quanta requests to create a repo in OpenBMC github

[Benjamin Fair]

On Thu, 17 Sep 2020 at 08:29, Patrick Williams <patrick@stwcx.xyz> wrote:
>
> Would someone mind giving a 1-2 sentence explaination of what "sensor
> margin values" are?  I'm certainly not a thermal control expert.

This daemon is intended to be used in combination with
phosphor-pid-control. You define a configuration file with two
sections: sensors and zones.

Sensors have a D-Bus path to read from along with a target temperature
and some other parameters. The daemon subtracks the current
temperature from the target temperature to get the "margin". A smaller
margin means the temperature is closer to the limit.

Zones group sensors together and look for the worst (lowest) margin
value within that group. This value is then exported for use in
phosphor-pid-control.

>
> I don't care if there are multiple implementations of similar
> functionality, but there is also work going on for a repository called
> 'phosphor-virtual-sensors'.  What little I could understand of
> read-margin-temp's role seems to somewhat overlap with that effort.

It may be possible to implement this using the functionality provided
by phosphor-virtual-sensors, but the resulting configuration would
likely be complicated and difficult to maintain. Margin calculation is
a common enough use case that I think it's worth having a dedicated
service.

>
> On Wed, Sep 16, 2020 at 10:37:14PM -0700, Josh Lehan wrote:
> > Yay, this is great to see. I have some bug fixes to read-margin-temp and
> > was just wondering how to contribute them. I'm really hoping this makes it
> > into OpenBMC. It's the perfect preprocessor for phosphor-pid-control, I
> > have found.
> >
> > Josh
> >
> > Josh Lehan | Software Engineer | krellan@google.com | +1 650-733-8941
> >
> >
> >
> > On Wed, Sep 16, 2020 at 7:15 PM Benjamin Fair <benjaminfair@google.com>
> > wrote:
> >
> > > On Wed, 16 Sep 2020 at 06:31, Fran Hsu (徐誌謙) <Fran.Hsu@quantatw.com>
> > > wrote:
> > > >
> > > > Hi Brad,
> > > >         This is Fran from Quanta Computer.
> > > > Quanta OpenBMC team would like to share some useful tools to the OpenBMC
> > > community.
> > > > Such as the read-margin-temp daemon (
> > > https://github.com/quanta-bmc/read-margin-temp), we are using this daemon
> > > to fill sensor margin value to PID to do fan control.
> > >
> > > read-margin-temp seems useful and widely applicable enough that I
> > > think it's worth having a dedicated repo for it.
> > >
> > > The quanta-misc repo could possibly be used for smaller, more
> > > Quanta-specific things.
> > >
> > > >
> > > > We would like to have a repo named "quanta-misc" under the OpenBMC
> > > github.
> > > >
> > > > Here is the maintainer list:
> > > > M:  Benjamin Fair <benjaminfair@google.com> <benjaminfair!>
> > > > M:  Brandon Kim <brandonkim@google.com> <brandonk!>
> > > > M:  Fran Hsu <fran.hsu@quantatw.com> <franhsu!>
> > > > M:  George Hung <george.hung@quantatw.com> <ghung!>
> > > > M:  Buddy Huang <buddy.huang@quantatw.com> <buddyhunag!>
> > > >
> > > > Thank you.
> > > > Fran Hsu
> > > > E-Mail : Fran.Hsu@QuantaTW.com
> > > > Tel: +886-3-327-2345 Ext: 16935 <+886%203%20327%202345>
> > > >
> > > >
> > >
>
> --
> Patrick Williams

Re: Quanta requests to create a repo in OpenBMC github

[Andrei Kartashev]

> This daemon is intended to be used in combination with
> phosphor-pid-control. You define a configuration file with two
> sections: sensors and zones.
> 
> Sensors have a D-Bus path to read from along with a target
> temperature
> and some other parameters. The daemon subtracks the current
> temperature from the target temperature to get the "margin". A
> smaller
> margin means the temperature is closer to the limit.
> 
> Zones group sensors together and look for the worst (lowest) margin
> value within that group. This value is then exported for use in
> phosphor-pid-control.


I wonder, isn't phosphor-pid-control do the same by its own? How does
it works inside pid-control daemon, when you set several thermal
sensors for one Zone an set setpoint for each of them?


-- 
Best regards,
Andrei Kartashev

Re: Quanta requests to create a repo in OpenBMC github

[Patrick Venture]

On Fri, Sep 18, 2020 at 1:06 AM Andrei Kartashev <a.kartashev@yadro.com> wrote:
>
>
> > This daemon is intended to be used in combination with
> > phosphor-pid-control. You define a configuration file with two
> > sections: sensors and zones.
> >
> > Sensors have a D-Bus path to read from along with a target
> > temperature
> > and some other parameters. The daemon subtracks the current
> > temperature from the target temperature to get the "margin". A
> > smaller
> > margin means the temperature is closer to the limit.
> >
> > Zones group sensors together and look for the worst (lowest) margin
> > value within that group. This value is then exported for use in
> > phosphor-pid-control.
>
>
> I wonder, isn't phosphor-pid-control do the same by its own? How does
> it works inside pid-control daemon, when you set several thermal
> sensors for one Zone an set setpoint for each of them?

Not sure what your question is, but yes, the pid-control daemon can
host its own margin sensor values that are populated externally,
however this is only supported via the json configuration at present.

>
>
> --
> Best regards,
> Andrei Kartashev
>
>

Re: Quanta requests to create a repo in OpenBMC github

[Benjamin Fair]

On Fri, 18 Sep 2020 at 08:50, Patrick Venture <venture@google.com> wrote:
>
> On Fri, Sep 18, 2020 at 1:06 AM Andrei Kartashev <a.kartashev@yadro.com> wrote:
> >
> >
> > > This daemon is intended to be used in combination with
> > > phosphor-pid-control. You define a configuration file with two
> > > sections: sensors and zones.
> > >
> > > Sensors have a D-Bus path to read from along with a target
> > > temperature
> > > and some other parameters. The daemon subtracks the current
> > > temperature from the target temperature to get the "margin". A
> > > smaller
> > > margin means the temperature is closer to the limit.
> > >
> > > Zones group sensors together and look for the worst (lowest) margin
> > > value within that group. This value is then exported for use in
> > > phosphor-pid-control.
> >
> >
> > I wonder, isn't phosphor-pid-control do the same by its own? How does
> > it works inside pid-control daemon, when you set several thermal
> > sensors for one Zone an set setpoint for each of them?
>
> Not sure what your question is, but yes, the pid-control daemon can
> host its own margin sensor values that are populated externally,
> however this is only supported via the json configuration at present.

It also doesn't support mixing margin sensors and absolute sensors.

Doing this margin calculation seems separate enough from the core
functions of phosphor-pid-control that I think splitting it out is the
best approach, but I do also see the argument for merging the
functionality into phosphor-pid-control.

>
> >
> >
> > --
> > Best regards,
> > Andrei Kartashev
> >
> >

Re: Quanta requests to create a repo in OpenBMC github

[Andrei Kartashev]

I'm just curious, because I don't really know how does pid-control
actually works. Let's say we have following zone configuration (EM
configuration file, unrelated PID parameters are hidden):
{
    "FailSafePercent": 100,
    "MinThermalOutput": 30,
    "Name": "Main",
    "Type": "Pid.Zone"
},
{
    "Class": "temp",
    "Inputs": [
        "DIMM [ABCDEF]\\d+ CPU0",
        "DIMM [ABCDEF]\\d+ CPU1"
    ],
    "Name": "DIMM ALL",
    "Outputs": [],
    "SetPoint": 80.0,
    "Type": "Pid",
    "Zones": [
        "Main"
    ],
    ...
},
{
    "Class": "temp",
    "Inputs": [
        "EXIT_AIR"
    ],
    "Name": "EXIT_AIR",
    "Outputs": [],
    "SetPoint": 50.0,
    "Type": "Pid",
    "Zones": [
        "Main"
    ],
    ...
},
{
    "Class": "fan",
    "Inputs": [
        "SYS_FAN"
    ],
    "Name": "SYS_FAN",
    "Outputs": [
        "SYS_FAN_PWM"
    ],
    "Type": "Pid",
    "Zones": [
        "Main"
    ],
    ...
},


How would it work? I was sure pid-control would examine all the sensors
and try to keep temperature below setpoints: 80 degrees for any of DIMM
memory sensors and 50 degree for exit air. If so, what is the benefit
to use this virtual margin sensor?

On Fri, 2020-09-18 at 08:49 -0700, Patrick Venture wrote:
> On Fri, Sep 18, 2020 at 1:06 AM Andrei Kartashev <
> a.kartashev@yadro.com> wrote:
> > 
> > > This daemon is intended to be used in combination with
> > > phosphor-pid-control. You define a configuration file with two
> > > sections: sensors and zones.
> > > 
> > > Sensors have a D-Bus path to read from along with a target
> > > temperature
> > > and some other parameters. The daemon subtracks the current
> > > temperature from the target temperature to get the "margin". A
> > > smaller
> > > margin means the temperature is closer to the limit.
> > > 
> > > Zones group sensors together and look for the worst (lowest)
> > > margin
> > > value within that group. This value is then exported for use in
> > > phosphor-pid-control.
> > 
> > I wonder, isn't phosphor-pid-control do the same by its own? How
> > does
> > it works inside pid-control daemon, when you set several thermal
> > sensors for one Zone an set setpoint for each of them?
> 
> Not sure what your question is, but yes, the pid-control daemon can
> host its own margin sensor values that are populated externally,
> however this is only supported via the json configuration at present.
> 
> > 
> > --
> > Best regards,
> > Andrei Kartashev
> > 
> > 
-- 
Best regards,
Andrei Kartashev

Re: Quanta requests to create a repo in OpenBMC github

[Benjamin Fair]

On Fri, 18 Sep 2020 at 11:23, Andrei Kartashev <a.kartashev@yadro.com> wrote:
>
> I'm just curious, because I don't really know how does pid-control
> actually works. Let's say we have following zone configuration (EM
> configuration file, unrelated PID parameters are hidden):
> {
>     "FailSafePercent": 100,
>     "MinThermalOutput": 30,
>     "Name": "Main",
>     "Type": "Pid.Zone"
> },
> {
>     "Class": "temp",
>     "Inputs": [
>         "DIMM [ABCDEF]\\d+ CPU0",
>         "DIMM [ABCDEF]\\d+ CPU1"
>     ],
>     "Name": "DIMM ALL",
>     "Outputs": [],
>     "SetPoint": 80.0,
>     "Type": "Pid",
>     "Zones": [
>         "Main"
>     ],
>     ...
> },
> {
>     "Class": "temp",
>     "Inputs": [
>         "EXIT_AIR"
>     ],
>     "Name": "EXIT_AIR",
>     "Outputs": [],
>     "SetPoint": 50.0,
>     "Type": "Pid",
>     "Zones": [
>         "Main"
>     ],
>     ...
> },
> {
>     "Class": "fan",
>     "Inputs": [
>         "SYS_FAN"
>     ],
>     "Name": "SYS_FAN",
>     "Outputs": [
>         "SYS_FAN_PWM"
>     ],
>     "Type": "Pid",
>     "Zones": [
>         "Main"
>     ],
>     ...
> },
>
>
> How would it work? I was sure pid-control would examine all the sensors
> and try to keep temperature below setpoints: 80 degrees for any of DIMM
> memory sensors and 50 degree for exit air. If so, what is the benefit
> to use this virtual margin sensor?

The most important use case is mixing absolute and margin temperatures.

Some sensors in a zone are only readable by the host, so there's a
service that collects that data and calculates a worst margin. This
margin is sent over IPMI to the BMC, but then needs to be combined
with the other, absolute temperatures that the BMC can read directly.

>
> On Fri, 2020-09-18 at 08:49 -0700, Patrick Venture wrote:
> > On Fri, Sep 18, 2020 at 1:06 AM Andrei Kartashev <
> > a.kartashev@yadro.com> wrote:
> > >
> > > > This daemon is intended to be used in combination with
> > > > phosphor-pid-control. You define a configuration file with two
> > > > sections: sensors and zones.
> > > >
> > > > Sensors have a D-Bus path to read from along with a target
> > > > temperature
> > > > and some other parameters. The daemon subtracks the current
> > > > temperature from the target temperature to get the "margin". A
> > > > smaller
> > > > margin means the temperature is closer to the limit.
> > > >
> > > > Zones group sensors together and look for the worst (lowest)
> > > > margin
> > > > value within that group. This value is then exported for use in
> > > > phosphor-pid-control.
> > >
> > > I wonder, isn't phosphor-pid-control do the same by its own? How
> > > does
> > > it works inside pid-control daemon, when you set several thermal
> > > sensors for one Zone an set setpoint for each of them?
> >
> > Not sure what your question is, but yes, the pid-control daemon can
> > host its own margin sensor values that are populated externally,
> > however this is only supported via the json configuration at present.
> >
> > >
> > > --
> > > Best regards,
> > > Andrei Kartashev
> > >
> > >
> --
> Best regards,
> Andrei Kartashev
>
>

RE: Quanta requests to create a repo in OpenBMC github

[George Hung (洪忠敬)]

Hi Brad, Patrick,

May we have quanta-misc repository first ? we still have other packages could be pushed.

As for the read-margin-temp, we can discuss further.

Thanks.

On Thu, 17 Sep 2020 at 08:29, Patrick Williams <patrick@stwcx.xyz> wrote:
>
> Would someone mind giving a 1-2 sentence explaination of what "sensor 
> margin values" are?  I'm certainly not a thermal control expert.

This daemon is intended to be used in combination with phosphor-pid-control. You define a configuration file with two
sections: sensors and zones.

Sensors have a D-Bus path to read from along with a target temperature and some other parameters. The daemon subtracks the current temperature from the target temperature to get the "margin". A smaller margin means the temperature is closer to the limit.

Zones group sensors together and look for the worst (lowest) margin value within that group. This value is then exported for use in phosphor-pid-control.

>
> I don't care if there are multiple implementations of similar 
> functionality, but there is also work going on for a repository called 
> 'phosphor-virtual-sensors'.  What little I could understand of 
> read-margin-temp's role seems to somewhat overlap with that effort.

It may be possible to implement this using the functionality provided by phosphor-virtual-sensors, but the resulting configuration would likely be complicated and difficult to maintain. Margin calculation is a common enough use case that I think it's worth having a dedicated service.

>
> On Wed, Sep 16, 2020 at 10:37:14PM -0700, Josh Lehan wrote:
> > Yay, this is great to see. I have some bug fixes to read-margin-temp 
> > and was just wondering how to contribute them. I'm really hoping 
> > this makes it into OpenBMC. It's the perfect preprocessor for 
> > phosphor-pid-control, I have found.
> >
> > Josh
> >
> > Josh Lehan | Software Engineer | krellan@google.com | +1 
> > 650-733-8941
> >
> >
> >
> > On Wed, Sep 16, 2020 at 7:15 PM Benjamin Fair 
> > <benjaminfair@google.com>
> > wrote:
> >
> > > On Wed, 16 Sep 2020 at 06:31, Fran Hsu (徐誌謙) 
> > > <Fran.Hsu@quantatw.com>
> > > wrote:
> > > >
> > > > Hi Brad,
> > > >         This is Fran from Quanta Computer.
> > > > Quanta OpenBMC team would like to share some useful tools to the 
> > > > OpenBMC
> > > community.
> > > > Such as the read-margin-temp daemon (
> > > https://github.com/quanta-bmc/read-margin-temp), we are using this 
> > > daemon to fill sensor margin value to PID to do fan control.
> > >
> > > read-margin-temp seems useful and widely applicable enough that I 
> > > think it's worth having a dedicated repo for it.
> > >
> > > The quanta-misc repo could possibly be used for smaller, more 
> > > Quanta-specific things.
> > >
> > > >
> > > > We would like to have a repo named "quanta-misc" under the 
> > > > OpenBMC
> > > github.
> > > >
> > > > Here is the maintainer list:
> > > > M:  Benjamin Fair <benjaminfair@google.com> <benjaminfair!>
> > > > M:  Brandon Kim <brandonkim@google.com> <brandonk!>
> > > > M:  Fran Hsu <fran.hsu@quantatw.com> <franhsu!>
> > > > M:  George Hung <george.hung@quantatw.com> <ghung!>
> > > > M:  Buddy Huang <buddy.huang@quantatw.com> <buddyhunag!>
> > > >
> > > > Thank you.
> > > > Fran Hsu
> > > > E-Mail : Fran.Hsu@QuantaTW.com
> > > > Tel: +886-3-327-2345 Ext: 16935 <+886%203%20327%202345>
> > > >
> > > >
> > >
>
> --
> Patrick Williams