Re: RSBAC performance in Linux 2.4.10


From: Pontus Lidman <pontus@lysator.liu.se>
Subject: Re: RSBAC performance in Linux 2.4.10
Date: 16 Jan 2002 20:18:04 +0100

Next Article (by Subject): Re: RSBAC performance in Linux 2.4.10 Amon Ott
Previous Article (by Subject): Re: RSBAC performance in Linux 2.4.10 Amon Ott
Top of Thread: RSBAC performance in Linux 2.4.10 Pontus Lidman
Next in Thread: Re: RSBAC performance in Linux 2.4.10 Amon Ott
Articles sorted by: [Date] [Author] [Subject]


Amon Ott <ao@rsbac.org> writes:

[...]

> > Of interest: Did you look into your /proc/rsbac-info/xstats, what calls are
> > made too often? How many attribute objects do you have
> > (/proc/rsbac-info/stats)?
> 
> Just rethinking: what updatedb does is an awful lot of readdir() system 
> calls, with one READ request each, going through all dir objects in the 
> system. This is a kind of worst case szenario, which should certainly be 
> optimized.

I don't have the 'intercept read/write', and neither do I use the MS
module.  I guess this performance loss is the expected behaviour for
this case then. Is there a similiar explanation to some NFS slowness
I've also experienced? Unfortunately I couldn't think of a good test
to compare NFS performance.

Here's my /proc/rsbac-info/stats, before and after the test, in case
it gives you some useful information:

--- stats_1     Tue Jan 15 18:24:12 2002
+++ stats_2     Tue Jan 15 19:59:33 2002
@@ -4,27 +4,27 @@
 Compiled Modules: MAC RC AUTH ACL
 All modules active (no switching)
 
-Device 08:01 has 303 fd-items and 1 dirty lists
+Device 08:01 has 305 fd-items and 1 dirty lists
 Device 00:02 has 0 fd-items and 0 dirty lists
 Device 08:05 has 0 fd-items and 0 dirty lists
 Device 08:06 has 28389 fd-items and 0 dirty lists
 Device 08:07 has 3757 fd-items and 0 dirty lists
 Device 03:01 has 0 fd-items and 0 dirty lists
 Device 03:05 has 0 fd-items and 0 dirty lists
-Device 03:04 has 1227 fd-items and 0 dirty lists
+Device 03:04 has 1220 fd-items and 0 dirty lists
 Device 08:17 has 1027 fd-items and 0 dirty lists
 Device 00:06 has 0 fd-items and 0 dirty lists
 Device 00:07 has 0 fd-items and 0 dirty lists
 Device 22:00 has 0 fd-items and 0 dirty lists
 Device 00:08 has 0 fd-items and 0 dirty lists
-Sum of 13 Devices with 34703 fd-items
+Sum of 13 Devices with 34698 fd-items
 0 dev-items
 23 ipc-items
 4 user-items
 41 process-items
-Total of 34771 registered rsbac-items, 1 lists dirty
+Total of 34766 registered rsbac-items, 1 lists dirty
 
 adf_request calls:
-file: 22415, dir: 47910, fifo: 3348, symlink: 2710, dev: 1920, ipc: 691, scd: 3
5, user: 0, process: 1766
+file: 344403, dir: 1186637, fifo: 9630, symlink: 19112, dev: 7991, ipc: 1065, s
cd: 57, user: 0, process: 2932
 adf_set_attr calls:
-file: 11010, dir: 663, fifo: 2989, symlink: 0, dev: 1908, ipc: 690, scd: 0, use
r: 0, process: 880
+file: 17208, dir: 98355, fifo: 5578, symlink: 0, dev: 7979, ipc: 1064, scd: 0, 
user: 0, process: 1400

-- 
Pontus Lidman, pontus@lysator.liu.se, Software Engineer
No matter how cynical you get, it's impossible to keep up.
Scene: www.dc-s.com | MUD: tyme.envy.com 6969 | irc: irc.quakenet.eu.org
-
To unsubscribe from the rsbac list, send a mail to
majordomo@rsbac.org with
unsubscribe rsbac
as single line in the body.

Next Article (by Subject): Re: RSBAC performance in Linux 2.4.10 Amon Ott
Previous Article (by Subject): Re: RSBAC performance in Linux 2.4.10 Amon Ott
Top of Thread: RSBAC performance in Linux 2.4.10 Pontus Lidman
Next in Thread: Re: RSBAC performance in Linux 2.4.10 Amon Ott
Articles sorted by: [Date] [Author] [Subject]


Go to Compuniverse LWGate Home Page.