Gentoo Archives: gentoo-user

From: thegeezer <thegeezer@×××××××××.net>
To: gentoo-user@l.g.o
Subject: Re: [gentoo-user] Re: gigabyte mobo latency
Date: Sat, 18 Oct 2014 23:02:35
Message-Id: 5442F17C.7040904@thegeezer.net
In Reply to: [gentoo-user] Re: gigabyte mobo latency by James
1 On 18/10/14 22:51, James wrote:
2 > thegeezer <thegeezer <at> thegeezer.net> writes:
3 >
4 >
5 >>> So. Is there a make.conf setting or elsewhere to make the
6 >>> terminal session response times, in the browsers (seamonkey, firefox)
7 >>> faster?
8 >>> the typing latency in the browser windows).....
9 >>> ideas?
10 >> two things you might like to look into: 1. cgroups (including freezer)
11 >> to help isolate your browsers and also 2. look at atop instead of htop
12 >> as this includes disk io
13 >
14 > 2. The system rarely uses 8 G of the 32 G available, so disk IO is
15 > not the problem. No heavy writes. It was the java scripts....
16 >
17 > 1. Ahhhhhhhhhhhhhhhhhhh! tell me more. I found these links quickly:
18 >
19 > https://www.kernel.org/doc/Documentation/cgroups/freezer-subsystem.txt
20 >
21 > http://wiki.gentoo.org/wiki/LXC#Freezer_Support
22 >
23 > I'm not sure if you've read any of my clustering_frustration posts
24 > over the last month or so, but cgroups is at the heart of clustering now.
25 > It seems many of the systemd based cluster solutions are having all
26 > sorts of OOM, OOM-killer etc etc issues. So any and all good information,
27 > examples and docs related to cgroups is of keen interests to me. My efforts
28 > to build up a mesos/spark cluster, center around openrc and therefore
29 > direct management of resources via cgroups.
30 >
31 > The freezer is exactly what I'm looking for. Maybe I also need to read up
32 > on lxc? What are the best ways to dynamically manage via cgroups? A gui?
33 > A static config file? a CLI tool?
34 >
35 >
36 > curiously,
37 > James
38 >
39 >
40 >
41 >
42 >
43 the thing with cgroups is that you can choose to create a hierarchy of
44 what _you_ want to have as your priority
45 unfortunately you need to tell the machine what it is you want, it can't
46 really guess granularly iuc
47 e.g. your favourite terminal / ide etc you want high prio and your
48 favourite file mangler to be low prio ( assuming you want compiling to
49 take precedence over bit munging to usb etc)
50
51 there is however cgroup support in htop, and i thought that you could
52 adjust cgroup in stead of nice but a quick google is showing that i
53 dreamed it as a nice feature; that would be super slick as you can
54 easily adjust all parts of program demand -- io / memory / cpu
55
56 using openRC you can start services i.e. apache to have a certain
57 priority, and ssh to have another
58 http://wiki.gentoo.org/wiki/OpenRC/CGroups
59 http://qnikst.github.io/posts/2013-02-20-openrc-cgroup.html
60
61
62 the reason i suggest freezer is that you can more easily "pause" or
63 CTRL-Z something that would otherwise be in a GUI and maybe not respond
64 to SIGSTP
65 on my laptop :-
66
67 # mount | grep freez
68 freezer on /sys/fs/cgroup/freezer type cgroup
69 (rw,nosuid,nodev,noexec,relatime,freezer)
70 # cd /sys/fs/cgroup/freezer/
71
72 call the folder something meaningful
73 # mkdir investigate
74 # cd investigate
75
76 you can use the following, i just did echo $$ for local bash pid... make
77 sure to get all threads especially something like chrome spawns many
78 children
79 # ps -eLf | grep mybadapp
80
81 note the single > actually concatenates
82 # echo $PID > tasks
83
84 to remove you actually have to "move" the process into a different
85 cgroup i.e.
86 # echo $PID > /sys/fs/cgroup/freezer/tasks
87
88 ok so once you have all your tasks in there just make sure you are in
89 /sys/fs/cgroup/freezer/investigate
90 # echo FROZEN > freezer.state
91
92 to thaw
93 # echo THAWED > freezer.state
94
95
96
97 there is a little more here
98 http://gentoo-en.vfose.ru/wiki/Improve_responsiveness_with_cgroups
99 which will allow you to script creating a cgroup with the processID of
100 an interactive shell, that you can start from to help save hunting down
101 all the threads spawned by chrome.
102 you can then do fun stuff with echo $$ >
103 /sys/fs/cgroup/cpu/high_priority/tasks
104
105 but for your original point of maybe it's not an issue with something
106 like IO it could still be a very high number disk reads -- low actual
107 thoughput but the demand on the io system was high, i.e. 6zillion reads
108 hopefully this will give you a bit more control over all of that though

Replies

Subject Author
[gentoo-user] Re: gigabyte mobo latency James <wireless@×××××××××××.com>