[prev in list] [next in list] [prev in thread] [next in thread]
List: systemd-devel
Subject: [systemd-devel] Am I getting/understanding certain bits on systemd/cgroups -- low-prio query
From: jsl6uy js16uy <js16uy () gmail ! com>
Date: 2016-05-27 17:07:33
Message-ID: CAAGD8_RxqOdxkdzKLxLxQZwFfZbZzsr4N1mHMMYs9P8hoFxRrg () mail ! gmail ! com
[Download RAW message or body]
[Attachment #2 (multipart/alternative)]
I dig transient services/scopes
I also like interacting with resource limits via systemctl set-property
interface
--easy to list resources that are available and set sans finding and
catting on /sys/fs/cgroup/foo
So...
Basically I have a system up and running. See some processes I want to
corral
So I would like to create a transient bucket, a scope seems best over a
service with the mentality that these process were not started by a systemd
unit file.
so right now I run
systemd-run --unit=foo --remain-after-exit cat /dev/null
The transient service stays up
I set an arbitrary resources limit to create files under /sys/fs/cgroup
Now I move processes into the tasks file within this cgroup path created
via systemctl/systemd
And I now can setup limits on a group, whatever I move into the tasks file,
on a set arbitrary run processes.
I could just set the limits on the individual processes but would have to
hit each and work with /proc and /sys/fs/cgroup to get them Corralled
I really would prefer to use the structure systemd creates/leverages and
use systemctl to interact with processes as a manually assembled group
So does this seem correct or should I be thinking of this differently?
--Also I question if I'm thinking of this right b/c of the "hack" to get an
"empty" bucket that I can dump running processes into for quick resource
control
Apologies if this is not the right forum for a concepts question, I can try
to find another list if needs be
As always thanks for any/all help
[Attachment #5 (text/html)]
<div dir="ltr"><div><div><br>I dig transient services/scopes<br></div><div>I also \
like interacting with resource limits via systemctl set-property \
interface<br></div><div>--easy to list resources that are available and set sans \
finding and catting on \
/sys/fs/cgroup/foo<br></div><div></div><div><br></div><div>So...<br></div>Basically I \
have a system up and running. See some processes I want to corral <br>So I would like \
to create a transient bucket, a scope seems best over a service with the mentality \
that these process were not started by a systemd unit file.<br></div><div>so right \
now I run<br></div><div>systemd-run --unit=foo --remain-after-exit cat \
/dev/null<br></div><div>The transient service stays up<br></div><div>I set an \
arbitrary resources limit to create files under /sys/fs/cgroup<br></div><div>Now I \
move processes into the tasks file within this cgroup path created via \
systemctl/systemd<br></div><div>And I now can setup limits on a group, whatever I \
move into the tasks file, on a set arbitrary run processes. <br><br></div><div>I \
could just set the limits on the individual processes but would have to hit each and \
work with /proc and /sys/fs/cgroup to get them Corralled<br><br></div><div>I really \
would prefer to use the structure systemd creates/leverages and use systemctl to \
interact with processes as a manually assembled group<br><br></div><div>So does this \
seem correct or should I be thinking of this differently?<br></div><div>--Also I \
question if I'm thinking of this right b/c of the "hack" to get an \
"empty" bucket that I can dump running processes into for quick resource \
control<br></div><div><br></div><div>Apologies if this is not the right forum for a \
concepts question, I can try to find another list if needs be<br><br></div><div>As \
always thanks for any/all help<br></div></div>
[Attachment #6 (text/plain)]
_______________________________________________
systemd-devel mailing list
systemd-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/systemd-devel
[prev in list] [next in list] [prev in thread] [next in thread]
Configure |
About |
News |
Add a list |
Sponsored by KoreLogic