Zone's vm+rss usage
I have two limits set, one for swap and one for rss in a Solaris u4 zone.
How do the applications behave in this environment? If I have 128mb capped-memory (rss) set, and I set swap to 256mb. Do the applications start eating swap once the 128mb gets used up? (Is it a full contiguous 384mb of vm?)
I am trying to ascertain, how much total memory a zone is using - should I just look at rss or add rss+swap (from "prstat -s rss -Z output")?
Thanks
If you have the RSS capped at 128MB and an application crosses that threshold, the scanner will begin working on paging until it gets below the set cap. Note that it is perfectly reasonable that this may not be a successful attempt if the app's RSS is genuinely larger than 128MB, in which case you'll be glad you have more on tap.
In the prstat output, swap is the total swap space allocated to the zone, and RSS is the resident set. In other words, the actively consumed memory. So, if I had swap of 462M and RSS of 415M, I have 47M which has been successfully paged.
-Chris
Similar Messages
-
Rcapd to limit zone RSS usage - total confusion!
I'm trying to use rcapd to limit an entire zone to a certain RSS limit.
i do this by setting all processes in the zone to run in within the 'system' project, with the following /etc/project file set to cap RSS at 128MB:
system:0::*::process.max-file-descriptor=(basic,4096,deny);rcap.max-rss=134217728
user.root:1::::process.max-file-descriptor=(basic,4096,deny);rcap.max-rss=134217728
noproject:2::::rcap.max-rss=134217728
default:3::::process.max-file-descriptor=(basic,4096,deny);rcap.max-rss=134217728
group.staff:10::::process.maxand-file-descriptor=(basic,4096,deny);rcap.max-rss=134217728and the follwing /etc/user_attr :
adm::::profiles=Log Management;project=system
lp::::profiles=Printer Management;project=system
root::::auths=solaris.*,solaris.grant;profiles=All;lock_after_retries=no;project
=systemThis appears to properly put everything within the zone running within a single project :
zone11# ps -e -o pid,project,comm
PID PROJECT COMMAND
20772 system /usr/local/apache2/bin/httpd
2121 system /usr/lib/sendmail
2107 system /usr/lib/sendmail
20798 system /usr/local/apache2/bin/httpd
2662 system zsched
26755 system /usr/local/sbin/clamd
533 system /opt/local/pgsql/bin/postmaster
3489 system /sbin/init
27103 system /usr/lib/rcap/rcapd
26840 system /usr/bin/mimedefang
3548 system /lib/svc/bin/svc.startd
1300 system /usr/local/apache2/bin/httpd
1301 system /usr/local/bin/ruby
24109 system /usr/bin/perl
583 system /opt/local/pgsql/bin/postmaster
20773 system /usr/local/apache2/bin/httpd
1292 system /usr/local/apache2/bin/httpd
3639 system /bin/perl
3551 system /lib/svc/bin/svc.configd
4786 system /usr/lib/utmpd
4813 system /usr/lib/saf/ttymon
4791 system /usr/lib/inet/inetd
4790 system /usr/sbin/syslogd
4751 system /usr/sbin/cron
4306 system /usr/sbin/saslauthd
4304 system /usr/sbin/saslauthd
4305 system /usr/sbin/saslauthd
4307 system /usr/sbin/saslauthd
4308 system /usr/sbin/saslauthd
4316 system /usr/lib/crypto/kcfd
2880 system /opt/local/pgsql/bin/postmaster
27361 system /bin/ksh
4815 system /usr/lib/ssh/sshd
1302 system /usr/local/apache2/bin/httpd
1303 system /usr/local/apache2/bin/httpd
24116 system /usr/bin/perl
26824 system /usr/bin/mimedefang-multiplexor
25826 system /opt/local/pgsql/bin/postmaster
582 system /opt/local/pgsql/bin/postmaster
1316 system /usr/local/bin/ruby
17091 system /usr/local/apache2/bin/httpd
4946 system /usr/bin/perl
1305 system /usr/local/apache2/bin/httpd
24117 system /usr/bin/perl
1304 system /usr/local/apache2/bin/httpd
2879 system /usr/local/apache2/bin/httpd
584 system /opt/local/pgsql/bin/postmaster
20763 system /usr/local/apache2/bin/httpd
8913 system /usr/bin/perl
28246 system psSo - unless I've totally misunderstood things, this should cap RSS for the entire zone to 128MB......
But - 'prstat' from the global zone shows:
PID USERNAME SIZE RSS STATE PRI NICE TIME CPU PROCESS/NLWP
27103 daemon 155M 135M sleep 1 0 0:13:33 0.0% rcapd/1
1316 nobody 34M 22M sleep 59 0 0:00:27 0.0% dispatch.fcgi/1
25826 203 16M 13M sleep 1 0 0:00:01 0.0% postgres/1
2880 203 16M 12M sleep 59 0 0:00:01 0.0% postgres/1
533 203 15M 11M sleep 59 0 0:00:00 0.0% postgres/1
582 203 15M 11M sleep 59 0 0:00:19 0.0% postgres/1
1301 nobody 33M 6952K sleep 59 0 0:00:27 0.0% dispatch.fcgi/1
3551 root 8768K 4072K sleep 59 0 0:12:36 0.0% svc.configd/22
24109 root 26M 2408K sleep 58 0 0:03:50 0.0% spamd/1
1302 nobody 18M 2336K sleep 1 0 0:00:00 0.0% httpd/1
3639 root 10M 2280K sleep 59 0 0:00:47 0.0% miniserv.pl/1
4316 daemon 4292K 1912K sleep 59 0 0:03:54 0.0% kcfd/4
1303 nobody 18M 1884K sleep 1 0 0:00:00 0.0% httpd/1
4791 root 3280K 1608K sleep 58 0 0:00:36 0.0% inetd/4
24116 root 26M 1600K sleep 57 0 0:00:03 0.0% spamd/1
ZONEID NPROC SIZE RSS MEMORY TIME CPU ZONE
15 48 755M 262M 4.3% 0:40:57 0.1% zone11
Total: 48 processes, 106 lwps, load averages: 0.26, 0.30, 0.30So the zone is using 262MB of RSS, with the largest abuser being - rcapd .... ??
What am I missing here??
Oh, this is x86 S10u1 :
# uname -a
SunOS srv1 5.10 Generic_Patch_118844-30 i86pc i386 i86pcIf you have the RSS capped at 128MB and an application crosses that threshold, the scanner will begin working on paging until it gets below the set cap. Note that it is perfectly reasonable that this may not be a successful attempt if the app's RSS is genuinely larger than 128MB, in which case you'll be glad you have more on tap.
In the prstat output, swap is the total swap space allocated to the zone, and RSS is the resident set. In other words, the actively consumed memory. So, if I had swap of 462M and RSS of 415M, I have 47M which has been successfully paged.
-Chris -
Why the "prstat -a" showed "incorrect" rss result?
Hi, folks
I was a little bit confused recently when I was investigating a memory issue as per the data collected by "prstat".
The output of the prstat looked like,
27465 nicgrp01 258M 175M cpu15 10 0 535:30:36 3.1% bsmciapp/9
7211 nicgrp01 258M 206M cpu9 10 0 14:20:25 3.1% bsmciapp/9
29482 nicgrp01 258M 197M cpu3 10 0 225:08:29 3.1% bsmciapp/9
but the summary looked like,
NPROC USERNAME SWAP RSS MEMORY TIME CPU
27 nicgrp01 204M 427M 2.6% 775:10:35 9.4%
Obviously, the sum of rss of the username "nicgrp01" won't be 427M only...(note, we have total 27 processes of user nicgrp01, the top 3 of them have been using more than 427M rss already...)
As I know that the final result of the "prstat -a" will show the sum of CPU, Memory of the top N users (+# of processes of that user).
In which scenario, the sum will display the rss usage like above?
Appreciated if anyone can shed me a light here.
Thanks.looks that they are sharing some memories.
-
Hi,
I have zones that have both physical and locked limits:
capped-memory:
physical: 600M
[locked: 512M]
rctl:
name: zone.max-locked-memory
value: (priv=privileged,limit=536870912,action=deny)Inside the zone I can display the locked memory with this command:
[root@zone1] $ prctl -i zone -n zone.max-locked-memory -t privileged zone1
zone: 2: zone1
NAME PRIVILEGE VALUE FLAG ACTION RECIPIENT
zone.max-locked-memory
privileged 512MB - deny -Is there a way to find the physical property of the capped-memory from inside the zone?
Thanks.[root@zone1] $ rcapstat -z 1 1
id zone nproc vm rss cap at avgat pg avgpg
8 zone1 - 0K 0K 200M 0K 0K 0K 0K -
Memory between global and non-global
Hi,
I've a Solaris 10 with a Solaris 8 brand zone.
from global zone I run prstat -Z and see:
ZONEID NPROC SWAP RSS MEMORY TIME CPU ZONE
+11 463 2639M 1908M *23%* 317:19:31 27% MyZone+
+0 55 109M 112M 1.4% 289:12:46 2.8% global+
from non-global zone the prstat -J command show me:
PROJID NPROC SIZE RSS MEMORY TIME CPU PROJECT
+3 114 19G 18G *27%* 229:18.14 3.1% default+
+1 265 52G 48G *73%* 54:31.44 3.1% user.root+
+0 78 308M 181M *0.3%* 3:39.41 0.0% system+
Why I have got this difference in memory percentage?from global
-bash-3.2# rcapstat -z 1 1
rcapstat: rcapd is not active
start svc:/system/rcap:default then
-bash-3.2# rcapstat -z 1 1
id zone nproc vm rss cap at avgat pg avgpg
Edited by: 853883 on Oct 10, 2012 2:13 AM -
Changing MIDI Velocity over time to affect samples in the ES2
Hello all!
I'm using ES2 horn samples in a track. The ES2 features crossfades between different velocity layers which is great! But I don't know how I would utilise this feature on a single sustained note.
I'm trying to create a Horn fade in on a long note across the course of a bar and it sounds a bit rubbish! I'm using automation currently and what I suspect is happening is that it is simply changing the volume of the ES2's initial sample and not changing the velocity. I want to work up from a quiet sample to a loud sample so that I get the harmonic changes that occur when a horn player does the same. Is it possible to assign velocity to a modulation control that will be variable on one note?Ok - firstly, you have to realise that you are working within the limitations of samples.
It's no good that a real instrument can be wildly expressive during the course of a played note when played by a skilled player, but you only have a sample of a flat, static note available in your DAW.
However, what you can do in your example is set up a MIDI controller to crossfade between sample layers during the course of a note in the EXS24, using the modulation matrix. For example, you can set "Sample select" as a mod destination, with "Controller #1" as a source - now your mod wheel will select the sample layers.
You might need to go in an edit the instrument itself to get the setup you want, depending on how the source instrument is programmed.
From the EXS24 manual, for instance, it says this for sample select:
"Modulates the sample (zone) that is played.
By default, Sample Select is controlled by velocity—through the default Velocity to Sample Select modulation routing. This means that the received note velocity value determines which of the layered zones (in different velocity ranges) is heard as you play the keyboard softer or harder.
You aren’t limited to using velocity, however, to determine which sample is played. You could assign the modulation wheel source to the Sample Select destination, or use both velocity and the modulation wheel.
If you choose a continuous controller such as the modulation wheel, you can step through the velocity layers during playback. In this case, make use of the crossfade (Xfade) parameters to create smooth transitions between velocity split points.
When using multiple modulation sources, be aware that these can cause all velocity layers to run simultaneously—using up as many voices as there are layered zones. The CPU usage will increase accordingly." -
Photoshop CC -- OpenCL causes GPU temperature to rise instantly with no load
Hello all,
I've just installed Photoshop CC 64 and am finding that with "OpenCL" checked under performance, my graphics card (Nvidia GTX 670, 2 gb, with updated drivers) temperature immediately rises from 33 C to over 45 C, and "core" from 324 Mhz to 1,005 Mhz (yellow zone), with high memory usage, without even opening an image. This does not occur with CS 6 (13.1.2 x64) with OpenCL checked -- the temperature remains constantly in the 33-35 C range. Does anyone have any ideas why this is happening? Is it a bug in PS CC? (system: Windows 7, x64 pro, 24 gb ram, i5 3570k)
Thanks,
ArtWe think it's a driver bug, as it seems to be limited to certain cards and driver versions (well, that and the fact that Photoshop doesn't even touch OpenCL until you run a filter that uses it).
We're still investigating the problem. -
Website Grader - Form Not Found
I have built two forms for a client website. Yet every time I run Website Grader, I get this message:
B. Conversion Form: Not Found
The web page does not seem to contain any type of signup form where visitors can subscribe to a newsletter or future updates.
We would suggest adding some type of simple subscription form (at a minimum) that allows users to get email updates. This is a great way for visitors to express their interest in the company and it's offering and for the business to stay in touch. Though RSS usage continues to increase, email is still what a majority of site visitors know and understand.
Please contact me through forum for URL if needed.I have learned something about SEO from using some of these web tools. And quite often I'll overlook something that shows up when I run them.
It would be better to learn from the horse's mouth rather than to rely on these commercial tools -
http://www.seomoz.org/articles/search-ranking-factors.php
And for heavens sake, anytime you see something from Jon Spivey, stop and read it! 8)
Question: Are you still able to use your news reader for forum use, or do you now have to use the forum?
The NNTP forums have been closed. Let Adobe know how you feel about this.
My Profile says I have 0 Points. But there is one red bar. Is that a demerit?
Yes. Report to the Principal's office right away. Don't forget your Hall Pass. -
Should I care about "Damaged Database" files in ~/Library/Syndication/ ?
Hi, friends —
I haven't used RSS much.
I'm curious: What is the purpose of the database files contained in the ~/Library/Syndication folder — and is there any reason to save the several "Damaged" Database files that are retained there?
At the moment, this folder contains a "Database3" and four "Damaged" Databases (~old), one of which is labeled "Broken." Each file is ~8–20 MB.
From the name and contents (I opened and perused a "Damaged Database" file using TextEdit), I'm guessing these are associated with RSS...?
Thanks!
DeanHi, macjack and Dr. Smoke —
Muchas gracias to you both for your thoughtful replies!
I'd just finished restarting my eMac after removing the Damaged Databases to my desktop per macjack's suggestion when I saw your message, Dr. Smoke. Thanks for your analysis and for the citation.
I had also searched the KnowledgeBase (9 results) and ADC (3) for resources on syndication, but hadn't had a chance yet to read the Safari RSS Developer Release Note. As you've noted, it comes to the point of my question —
Quoting from the Note for anyone who stumbles across this thread and is interested:
"Purging the Database
Safari RSS stores all news articles in a database. There is not yet any mechanism to control how long articles live, or to delete articles on demand, so the database will grow monotonically.
If the database is taking up too much disk space, you can use the following workaround to delete it. You won't lose any of your subscription data (that's stored in Safari's bookmarks) and the SyndicationAgent will re-fetch all current articles the next time it runs, but older articles will be lost.
1. Quit Safari.
2. Quit the SyndicationAgent process by entering "killall SyndicationAgent" in a Terminal window.
3. Move the directory ~/Library/Syndication to the Trash, or delete it."
[Emphasis added.]
Thanks for pointing out the potential for growth of this database; it may be worth keeping an eye on for some folks — depending on RSS usage, I guess. Seems an odd design concept to me.
My current Database3 file covers ~ the last 9 weeks — but this may not be comparable to others', since I haven't too big a collection of RSS feeds.
I was ~surprised to find so little documentation about SyndicationAgent on Apple's website (no Man page, etc.). Does that color me naïve?
I scanned through my (Text-archived) log files without finding any close correlations between SyndicationAgent or Safari entries and these databases (in terms of time). Since the latest "Damaged Database" file is dated 4-½ months ago, this issue isn't a big concern — as I said, just curious...
Thanks again!
Warm regards,
Dean
I edited this message... But probably not enough! -
Any solutions to fast iPhone 4s battery drainage?
Just got iphone 4s today and it rapidly lost power as i was listening to Pandora, much much faster as the 3GS i've had for some time .
Apple has acknowledged that there is a battery leak in the 4s and with devices using iOS 5 and they have pledged to fix this in the coming weeks via a software update. I am having the same issue. Some have had success disabling location services as follows:
Settings-Location Services-[disable all non-essential apps that use Location Services] then at the bottom of that screen you will see System Services-disable Time Zone-iAds-Diagnostics and Usage.
I did this and it didn't help so I am one of those awaiting the update from Apple, however, this has helped many others.
Best of luck!
Ted -
Help/Info regarding the CL rating of RAM upgrades.
I'm upgrading my Early 2011 Macbook Pro 17 to 8GB Ram and I'm curious if anyone can offer advise regarding the CL rating. All the DDR3 1333GHz SODIMM kits have a CL9 rating. Is this a higher latency than the 4GB of ram that I have already? If so, will more latency actually do more harm than good??
Pardon me if I sound like a noob!
I'm running a stock MBP 17 8,1 2.2GHz i7 with Windows 7 Pro Bootcamp. (Windows & + Zone Alarm = 50% RAM usage when using Excel, UGH!)No, CL9 is a standard. I have the same model and put 8G from Omni Technologies in mine. It screams!
17" 2.2GHz i7 Quad-Core MacBook Pro 8G RAM 750G HD + OCZ Vertex 3 SSD Boot HD -
Soalris 10 Resource Management question
Hi,
Looking at the outoput of rcapstat, how much memory allocated would be suitable for avoiding paging. I guess you could set the cap at a large value, monitor then reduce the cap but I would like a calculated/derived approach. The server has enough physical memory, I belive the capping values need to be adjusted. Thanks.
xxxx # rcapstat -z
id zone nproc vm rss cap at avgat pg avgpg
1 yyyy 65 2231M 5382M 8192M 8537M 0K 1854M 0KHi,
My interprestation of rcapstat was incorrect. I should have read the man page correctly, which says
The paging statistics in the first report issued show the
activity since the daemon was started. Subsequent reports
reflect the activity since the last report was issued.
subsequent reports did show 0K. So I've incorrectly interpreted the output. -
How to dynamically set max-rss size for a local zone?
Running solaris 10 05/08.
Following situation: Three local zones in a Veritas cluster. The zones
can fail over but normally not all three are on the same physical box.
I want to be able to restrict physical memory usage for the zones in
situations where certain combinations of them end up on the same
physical box. So, I need to be able to dynamically change memory
resources. I know how to do this with, for example locked memory:
prctl -n zone.max-locked-memory -r -v <mem> `pgrep -z <zone name>
init`
but i need to do this for physical memory and I can't see that there
is a zone resource for this? I thought this was added in 05/08?Hi
You can do this by using rcapd from the global zone
- start rcapd from the global zone:
svcadm enable svc:/system/rcap:default
- set a physical memory caping value for each zone
rcapadm -z your_zone -m max_rss_value
- check the memory use for all the zone capped
rcapstat -gz
You can set the rss capping value in the zone configuration:
# zonecfg -z your_zone
zonecfg:busi-app-prod> add capped-memory
zonecfg:busi-app-prod:capped-memory> set physical=900m
zonecfg:busi-app-prod:capped-memory> end
zonecfg:busi-app-prod> verify
zonecfg:busi-app-prod> commit
zonecfg:busi-app-prod> exit
This will be take in account at the next reboot of this zone.
Have a also a look to rcapadm to tune rcapd
Bye
Fred -
Wrong time zone with rss feed in iWeb
I'm using the RSS widget in iWeb, and all the RSS items are nine hours off - meaning they're on California time (I'm in Norway). The time zone is wrong even when I look at the feed in iWeb, so I don't think it's a server issue (and I'm not publishing to MobileMe).
I've found a couple of other questions about this on the board, but no answers, so here we go again... anyone? I've tried several different feeds.Same question! There must be some command-line hack to fix this. Any one figured it out?
It's posted times in USA Pacific time by default (GMT -8). This is iWeb 3.0.1. -
I've noticed that when I check my 'Blogs' RSS page (consisting of Engadget, Gizmodo, Lifehacker, Boing-Boing, 43 Folders, Dvorak, and TUAW) my processor usage spikes to 100% and the feeds take a really long time to load. My guess is this is because so many of these feeds include pictures and can get rather long. Has anyone else noticed this, and is there anything I can do about it? Or should I just grin and bear it?
Nevermind, no big deal. Probably because the entries are often an entire post, graphics and all.
Maybe you are looking for
-
Thinking about switching to Intel Mac Mini... questions for the cognoscenti
Greetings: I've been checking out the new Intel Mac Minis and I was thinking of purchasing one to replace my AV computer (PC) in the living room, which the GF currently uses. She's a typical user: watches video files/DVDs, uses iTunes excessively , s
-
Handling standard toolbar event in CL_SALV_TABLE
Hi Experts, I have created a ALV report using class CL_SAL_TABLE. I have created some buttons in the application toolbar of ALV by which I can jump to other ALV views. Now my requirement is if I click on the Exit (Shift+F3) button on the standard too
-
i m new one of sap i need to do the mass upload to infotype 2011 please help me..
-
Hi, Incase of upgrade from R3 4.6c to ECC(new box), we need to point the existing BW landscape to the new ECC landscape. 1. In this scenario, can we follow the document http://www.sdn.sap.com/irj/scn/go/portal/prtroot/docs/library/uuid/70a7b233-8be5-
-
Uncommitted data warning popup not working
Hi ADF Experts, JDEV 11.1.1.7.0 I have a jspx page with af:document uncommettedDataWarning="on". Inside jspx I have a region. In the jsff fragment I have inputText with autoSubmit="true", The popup does not appear if I try to close the browser or ref