<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
</head>
<body>
<p><br>
</p>
<div class="moz-cite-prefix">Le 08/04/2021 à 17:27, Mantas Mikulėnas
a écrit :<br>
</div>
<blockquote type="cite"
cite="mid:CAPWNY8XBA4fmkktKdsMvN4scyYo_yswLT4LaSsUQu43U94v12Q@mail.gmail.com">
<meta http-equiv="content-type" content="text/html; charset=UTF-8">
<div dir="ltr">
<div class="gmail_quote">
<div dir="ltr" class="gmail_attr">On Thu, Apr 8, 2021 at 6:19
PM Hadrien Grasland <<a
href="mailto:hadrien.grasland@ijclab.in2p3.fr"
moz-do-not-send="true">hadrien.grasland@ijclab.in2p3.fr</a>>
wrote:<br>
</div>
<blockquote class="gmail_quote" style="margin:0px 0px 0px
0.8ex;border-left:1px solid
rgb(204,204,204);padding-left:1ex">Le 08/04/2021 à 16:11,
Lennart Poettering a écrit :<br>
> On Do, 08.04.21 12:24, Hadrien Grasland (<a
href="mailto:hadrien.grasland@ijclab.in2p3.fr"
target="_blank" moz-do-not-send="true">hadrien.grasland@ijclab.in2p3.fr</a>)
wrote:<br>
><br>
>> Hi everyone,<br>
>><br>
>> In a scenario where running benchmarks on dedicated
hardware is not<br>
>> possible, I'm trying to momentarily cap the I/O
bandwidth used by<br>
>> interactive user sessions while benchmarks are
running, in order to improve<br>
>> the stability of said benchmark's I/O performance.<br>
> Is this on cgroupsv1 or cgroupsv2?<br>
><br>
> IIRC there was some issue that the block io controller
wasn't fully<br>
> recursive on cgroupsv1. It should work on cgroupsv2.<br>
<br>
This is on a hybrid cgroup configuration. I (perhaps
mistakenly) assumed <br>
that modern systemd (v246) will use the cgroups v2 hierarchy
in that <br>
case, even though cgroups v1 is still exposed for
compatibility with <br>
older apps.<br>
</blockquote>
<div><br>
</div>
<div>If e.g. the io controller is exposed through cgroups v1,
as far as I know it cannot be simultaneously used through
cgroups v2, and vice versa.</div>
<div><br>
</div>
<div>(Hmm, wasn't there an option to choose which controllers
to assign to v1 and which ones to v2?)<br clear="all">
</div>
</div>
</div>
</blockquote>
Ah yes indeed, I missed this important bit when I last read
<a class="moz-txt-link-freetext" href="https://systemd.io/CGROUP_DELEGATION/">https://systemd.io/CGROUP_DELEGATION/</a> :
<p>"Note that in
[hybrid] mode the unified hierarchy won’t have controllers
attached, the
controllers are all mounted as separate hierarchies as in legacy
mode,
i.e. <code class="language-plaintext highlighter-rouge">/sys/fs/cgroup/unified/</code>
is purely and exclusively about core cgroup v2
functionality and not about resource management."</p>
<p>Then I guess unless your parenthesized comment is correct (which
means that the above doc is out of date), I must use pure
unified/cgroups2 mode in order to do what I want with IO... except
I can't switch to it because this will break slurm's cgroups
support, which I need for a couple of other things, as it seems
Slurm only supports v1 at the moment :
<a class="moz-txt-link-freetext" href="https://groups.google.com/g/slurm-users/c/z57-Z3Tz0Oc?pli=1">https://groups.google.com/g/slurm-users/c/z57-Z3Tz0Oc?pli=1</a> .
Hmmm...</p>
<p>I'd appreciate any confirmation/refuttal on the ability to do
adjust controller asignment in a fine-grained way, otherwise it
seems I'll need to live without I/O caps until slurm implements
cgroup v2 support.</p>
<p>Hadrien<br>
</p>
</body>
</html>