Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:126239
Precedence: bulk
MIME-Version: 1.0
References: <CAGVaEcmDxxbFzNxXnF=jmdUDnQTk55zgorbsqQqyP8w2Z=-iaw@mail.gmail.com>
In-Reply-To: <CAGVaEcmDxxbFzNxXnF=jmdUDnQTk55zgorbsqQqyP8w2Z=-iaw@mail.gmail.com>
Date: Thu, 30 Jan 2025 13:42:47 +0100
Message-ID: <CAEKnhAFeiJESkqJ0o9PuU59VFK1hdfk2xeEDJn_BbsaF0sBy7Q@mail.gmail.com>
Subject: Re: [PHP-DEV] [RFC] Introducing pm.max_memory for PHP-FPM
To: Arkadiy Kulev <eth@ethaniel.com>
Cc: internals@lists.php.net
Content-Type: multipart/alternative; boundary="00000000000012086e062cebc4af"
From: bukka@php.net (Jakub Zelenka)

--00000000000012086e062cebc4af
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

Hi,

On Tue, Jan 28, 2025 at 2:17=E2=80=AFAM Arkadiy Kulev <eth@ethaniel.com> wr=
ote:

> Hello everyone,
>
> I=E2=80=99d like to propose adding a new configuration directive, tentati=
vely
> called *pm.max_memory*, to PHP-FPM. This directive would allow
> administrators to specify a per-child memory usage limit, after which the
> PHP-FPM child process would gracefully exit or restart.
>

I think it's reasonable as an additional protection and a bit nicer
than pm.max_requests. However, implementation might be tricky.


> Implementation Details and Challenges
>
> I am not proposing to implement this feature myself=E2=80=94I=E2=80=99m m=
ore of a sysadmin
> and don=E2=80=99t have the necessary C knowledge to patch php-src. Howeve=
r, here
> are some thoughts on what the implementation might involve:
>
>    1. *Measuring memory usage:* Likely via getrusage(), mallinfo() (on
>    some platforms), or reading from /proc/self/statm on Linux.
>
>
This wouldn't really work because FPM does not control the script during
execution and would have to check it out after each allocation which is not
really viable.


>
>    - *System-level OOM or cgroups:* This approach can kill the entire
>    pool or container, which is often more disruptive than recycling a sin=
gle
>    worker.
>
>
I think this is really the only way that I can see to make it work
reliably. We have got already an old feature request for that:
https://bugs.php.net/bug.php?id=3D70605 . There was even a PR
https://github.com/php/php-src/pull/2440 but it was using unmaintained lib
and cgroup v1. We should use something cgroup v2 based but it is not a
small job to add it.

Anyway you can create a GitHub feature request for this so it doesn't get
forgotten and maybe one day there will a better PR that can be accepted.

Cheers

Jakub

--00000000000012086e062cebc4af
Content-Type: text/html; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><div>Hi,</div><div><br></div><div dir=3D"ltr">On Tue, Jan =
28, 2025 at 2:17=E2=80=AFAM Arkadiy Kulev &lt;<a href=3D"mailto:eth@ethanie=
l.com" target=3D"_blank">eth@ethaniel.com</a>&gt; wrote:</div><div class=3D=
"gmail_quote"><blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px=
 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div dir=3D=
"ltr"><div><p>Hello everyone,</p><p>I=E2=80=99d like to propose adding a ne=
w configuration directive, tentatively called <strong><code>pm.max_memory</=
code></strong>, to PHP-FPM. This directive would allow administrators to sp=
ecify a per-child memory usage limit, after which the PHP-FPM child process=
 would gracefully exit or restart.</p><h3></h3></div></div></blockquote><di=
v><br></div><div>I think it&#39;s reasonable as an additional protection an=
d a bit nicer than=C2=A0pm.max_requests. However, implementation might be t=
ricky.</div><div>=C2=A0</div><blockquote class=3D"gmail_quote" style=3D"mar=
gin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1=
ex"><div dir=3D"ltr"><h3><span style=3D"font-size:small;font-weight:normal"=
>Implementation Details and Challenges</span></h3><p>I am not proposing to =
implement this feature myself=E2=80=94I=E2=80=99m more of a sysadmin and do=
n=E2=80=99t have the necessary C knowledge to patch <code>php-src</code>. H=
owever, here are some thoughts on what the implementation might involve:</p=
><ol><li><strong>Measuring memory usage:</strong> Likely via <code>getrusag=
e()</code>, <code>mallinfo()</code> (on some platforms), or reading from <c=
ode>/proc/self/statm</code> on Linux.<br></li></ol></div></blockquote><div>=
<br></div><div>This wouldn&#39;t really work because FPM does not control t=
he script during execution and would have to check it out after each alloca=
tion which is not really viable.</div><div>=C2=A0</div><blockquote class=3D=
"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(2=
04,204,204);padding-left:1ex"><div dir=3D"ltr"><div><ul><li><strong>System-=
level OOM or cgroups:</strong> This approach can kill the entire pool or co=
ntainer, which is often more disruptive than recycling a single worker.</li=
></ul><h3></h3></div></div></blockquote><div><br></div><div>I think this is=
 really the only way that I can see to make it work reliably. We have got a=
lready an old feature request for that: <a href=3D"https://bugs.php.net/bug=
.php?id=3D70605">https://bugs.php.net/bug.php?id=3D70605</a> . There was ev=
en a PR=C2=A0<a href=3D"https://github.com/php/php-src/pull/2440">https://g=
ithub.com/php/php-src/pull/2440</a> but it was using unmaintained lib and c=
group v1. We should use something cgroup v2 based but it is not a small job=
 to add it.</div><div><br></div><div>Anyway you can create a GitHub feature=
 request for this so it doesn&#39;t get forgotten and maybe one day there w=
ill a better PR that can be accepted.</div><div><br></div><div>Cheers</div>=
<div><br></div><div>Jakub</div></div>
</div>

--00000000000012086e062cebc4af--