Prstat output in Solaris 11

Hi all,
Looking at prstat output in Solaris 11, something has clearly changed from Solaris 10 prstat version, since the results are quite different.
Looking at an x86 server with S11+SRU7.5, the Oracle instance is eating(RSS) 57GB memory from a machine that only has 14GB memory in total. Is RSS calculated now with used memory+swap? if so, whats the SWAP column for then?
NPROC USERNAME SWAP RSS MEMORY TIME CPU
82 root 2755M 2003M 3.3% 4:19:21 0.8%
46 oracleoc 61G 57G 96% 0:22:08 0.0%
11 gdm 609M 143M 0.2% 0:02:53 0.0%
5 daemon 36M 13M 0.0% 0:01:31 0.0%
RSS
The resident set size of the process (RSS), in kilobytes
(K), megabytes (M), or gigabytes (G). The RSS value is
an estimate provided by proc(4) that might underestimate
the actual resident set size. Users who want to get more
accurate usage information for capacity planning should
use the -x option to pmap(1) instead.

I hope it's a bug.
Have you seen bug 7117588. It reads in part:
"prstat should account for shared pages when counting RSS and swap for zones, projects, tasks, and users. It does not due so after the fix for 6522129."
I am not sure how to read this statement but I guess it is/was broke.
GlenG

Similar Messages

Question of the output of "prstat -a" on solaris 10

Hi, all
Recently, I'm debugging a memory issue on the solaris 10 system. I executed the command like "prstat -U root -n 200 -s rss -a" to list all root user memory consumption and sorted by rss. I'm sure that the total number of root process is less than 100. My target is to find which root process takes the memory most. However I'm concerning the output of the above command. The summary on the bottom of the command showed that the total RSS of root was about 4G, the NPROC was about 150. When I did a "sum" of the RSS of all 150 root processes, the value was 1.5G which was far less than the 4G on the bottom summary.
Does anyone know if my understanding is correct or not? Should they be identical?
Thanks

Because of the way Solaris allocates and reports memory usage it is often (very) hard to account for the different values reported by the various commands. Most often the differences involve shared memory.
Why do you think you have a memory problem?
Do you think the problem is with virtual or physical memory?
Have you looked at:
How to Check and Analyze Solaris Memory Usage [ID 1009500.1]
GlenG

Prstat output

Dear all,
One of my server shows 99% used by Oracle continuously. I just want to know is there any issue related with lack of memory or Issue from Solaris 10 Operating System. Please see the output of prstat -a.
NPROC USERNAME SIZE RSS MEMORY TIME CPU
206 oracle 157G 154G 99% 0:05:39 0.3%
46 root 219M 92M 0.1% 1:21:56 0.1%
58 application 1020M 689M 0.4% 0:00:15 0.0%
1 noaccess 225M 110M 0.1% 0:35:31 0.0%
1 smmsp 7752K 936K 0.0% 0:00:05 0.0%
Best Regards
Bikash

Hi,
Extremely sorry if I had mentioned something wrong. The truncated output of �prstat �a� is the same what I mentioned above. I just want to know whether there is any issue relater to Operating System or memory if the output seems as below,
NPROC USERNAME SIZE RSS MEMORY TIME CPU
206 oracle 157G 154G 99% 0:05:39 0.3%
46 root 219M 92M 0.1% 1:21:56 0.1%
58 finadm 1020M 689M 0.4% 0:00:15 0.0%
1 noaccess 225M 110M 0.1% 0:35:31 0.0%
1 smmsp 7752K 936K 0.0% 0:00:05 0.0%
Total: 318 processes, 842 lwps, load averages: 0.16, 0.18, 0.21
Bikash

Explanation of prstat output

Dear All,
Pls explain to me in details
1. what you mean by the total virtual memory size of the process ?
2. resident set size of the process?
3. difference between cpu state and run state ?
these all the outputs of prstat command
Since iam new to solaris ,I require your help in detail
Thanks in advance
Regards
Yogesh

Virtual Size, or "Size" - the amount of memory allocated by the process.
Resident, or "RSS" - the amount of resident memory, aka memory that can't be swaped.
The CPU state is simply if the process is sleeping (not doing anything) waiting to get CPU (running) time or currently on the CPU.
.7/M.

Prstat output question

Hi all,
I have 2 problems I don't manage to explain when running prstat on Solaris 8 :
1- prstat does not see fsflush process !!!!
You can even try to force it : prstat -p 3 : it won't work...
2- prstat -a differs from prstat -aL when you look at TIME (and MEMORY) value for each user
Does anybody know the reason why ???
Thanks,
Jerome

Thanks for your answer about processes 0, 2 and 3
I found the answer to my second question : bug 4481976 : prstat without -L reports wrong CPU time for multithreaded processes
resolved by patch 109033-03....
Jerome

Question on top command output in solaris

Platform : Solaris 5.10
In the top output, if you look at the CPU colum, you can see that one process (2415) is consuming 95.84 % of CPU, another process (28533) is consuming 91.81%. How can this happen ? If one process is consuming 95.84 , then less 5% of CPU is available to all other processes . Right?
load averages: 4.24, 4.02, 3.89;                    up 57+10:17:11     14:36:35
252 processes: 245 sleeping, 1 zombie, 6 on cpu
CPU states: 40.7% idle, 46.7% user, 12.6% kernel, 0.0% iowait, 0.0% swap
Memory: 16G phys mem, 1002M free mem, 31G swap, 31G free swap
   PID USERNAME LWP PRI NICE SIZE   RES STATE    TIME    CPU COMMAND
2415 oracle     1   0    0 6267M 6182M cpu     45.7H 95.84% oracle   -------- ???
28533 oracle     2   0    0 5245M 5166M cpu     88:23 91.81% oracle   -------- ???
12062 oracle     2   0    0 5245M 5166M cpu    194:39 82.03% oracle
3459 oracle     1   0    0 1380K 996K cpu      9:55 73.00% gzip
2276 oracle     2 35    0 5244M 5165M sleep    0:16 12.84% oracle
13629 oracle     2 33    0 5244M 5165M sleep    4:51 12.44% oracle
7784 oracle     2 32    0 5349M 5270M sleep    1:51 11.85% oracle
5217 oracle     1 100 -20 5248M 5154M sleep   17.5H 3.05% oracle
5225 oracle     1 100 -20 5248M 5154M sleep   17.3H 2.87% oracle
14569 oracle     1 100 -20 6273M 6175M sleep   28.0H 1.61% oracle
14565 oracle     1 100 -20 6271M 6173M sleep   28.8H 1.56% oracle
1052 root      44 59    0   61M   36M sleep 319:29 1.41% crsd.bin
15734 oracle     3   0    0   33M   12M sleep 591:32 1.11% tnslsnr
5237 oracle     2 59    0 5275M 5181M sleep   79:47 0.51% oracle
5235 oracle     2 59    0 5271M 5177M sleep   81:17 0.44% oracle

I think you are right. This machine has 8 CPUs and there are 15 lines in the output . May be each line in the top result is for each core.
Result of psrinfo showing the number of CPUs
$ /usr/sbin/psrinfo
0       on-line   since 04/12/2011 09:19:34
1       on-line   since 04/12/2011 09:19:42
2       on-line   since 04/12/2011 09:19:44
3       on-line   since 04/12/2011 09:19:46
4       on-line   since 04/12/2011 09:19:48
5       on-line   since 04/12/2011 09:19:50
6       on-line   since 04/12/2011 09:19:52
7       on-line   since 04/12/2011 09:19:54Edited by: Tadeusz on Jun 8, 2011 8:13 AM

Console output on Solaris 7 and Solaris 8

Hi All...
It seems to me strange and impossible, but...
I load 32-bit driver : add_drv -m '* 0600 root root' <driver_name> on Solaris 7 and 8.
Both systems are 32-bit. The same source, the same define flags ( I use -D_KERNEL -DSUNDDI ).
On 8 I can see my driver's console output, on 7 - no way!
I produce console output using function cmn_err(CE_NOTE, format ...);
One interesting thing else: after my driver has come in kernel refuses to attach it, so the driver remains unattached. After rem_drv I checked dmesg output, there I saw this :
NOTICE: <driver name>: 64-bit driver module not found
What could it be?? Any suggestions are welcome
Thank you in advance
Andrew

Sorry, computer was booted with 64-bits kernel.
Andrew

No sound output on Solaris 8 and an Ultra 5

I have recently installed Solaris 8 on a Sun Ultra 5 after a few minor problems, everything seems to be working great.
However, I am unable to use any of the system sound notifications under KDE or Gnome. KDE starts the arts sound server just fine without any errors. Audio CD works great but no other audio.
Any suggestions?
Thank you.

Bummer. So you still can't boot OS 9, and get no sound device listed in the OS X Sound Preference Pane? Just checking, did you also press the CUDA reset button after reseating the card? (It's on the side of the motherboard between the PCI card slots and the side, at the rear of the DT case.)
For me, the sound started working again magically after I moved. So, my computer was unplugged for a couple of days and possibly bouncing around in the car a little. Not sure if a loose connection was restored during the move, or being unplugged from the wall power for a few days helped the computer recognize the card again.
Of course there's still the possibility your card is no longer working.

Lpstat -t -o commands hangs with no output on SOlaris 10

Hi All,
We have ( SunOS 5.10 Generic_127111-06 sun4u sparc SUNW,SPARC-Enterprise),on a box .
The issue is when ever lpstat command is used to see weather there are any queues in the printer it hangs...
But print queue is getting cleared want to know weather its a Bug in Solaris 10...........
We have configured the printer on port 9100...
cosuaor19:> cat /etc/lp/printers/ALBH_SHIP_LN1/configuration
Banner: off
Content types: simple,postscript
Device: /dev/null
Interface: /usr/lib/lp/model/netstandard
Printer type: hplaser
Modules:
Options: dest=192.100.9.106:9100,protocol=tcp
cosuaor19:> telnet 192.100.9.106 9100
Trying 192.100.9.106...
Connected to 192.100.9.106.
Escape character is '^]'.
^]
telnet> quit
Connection to 192.100.9.106 closed.
cosuaor19:>
So any help on this would be appreciated ...

gdb shows that asynchronous i/o for sockets is handled only in user space via creating threads.
Additional info can be found on http://mail.opensolaris.org/pipermail/perf-discuss/2007-February/001621.html too.

Nslookup output changes between Solaris 8 and 10

I have a customer that has asked why and if the output can be changed to match?
We have noticed a difference on the nslookup reverse lookup with
Solaris 10.
On Solaris 8 we get:
# nslookup x.x.x.x
Server: dnsserver.csd.disa.mil
Address: x.x.x.x
Name: hostname.my.af.mil
Address: x.x.x.x
On Solaris 10 we get:
# nslookup x.x.x.x
Server: x.x.x.x
Address: x.x.x.x#53
x.x.x.x(server ip in reverse).in-addr.arpa name =
hostname.my.af.mil.
Why is the output on Solaris 10 different when doing an nslookup on
the IP
address? The guys who work on the application noticed it and
opened a
ticket. Is there something I can do to get the output to come out
like it
does on Solaris 8.Any and all help would be appreciated....

SPARC doesn't need a split boot loader like BIOS/x86. The OBP can directly load from any slice. So 'boot disk:a' would boot from slice 0 and 'boot disk:g' would boot from slice 6.
Darren

Xsun (high CPU utilization) on Solaris 10 Sparc

hi
i have sun blade 1500 and am running solaris 10 on it. the machine is a 2 CPU (750Mhz) 4GB Sparc with the lastest cluster patch.
The Xsun process is alway at 50% util and the windowmaker (wmaker) is at 27%.
The Xsun is alway using all available CPU and the machine is really slow.. any help on what patch will fix the Xsun process ? any operation on the machine increases the Xsun's CPU util.
thx
Sriram

Hi sridhar,
can i know which platform you are using....
if it is solaris,
can you paste the details of the prstat -L -p wlpid 1 1 ---> which gives the light wieght pid threads
and also (pstact wlspid) for lwpid to process id mapping
or you can follow these steps for finding which thread is causing the hight cpu utilization
1.find the highest usage lwpid in prstat output
2.find the lwpid in pstack output and get the matching thread number
3.convert the thread number to hexadecimal
4.find the hexadecimal thread number in the server thread dump (nid= xxx)
5.determine what thread was doing to cause the high cpu usage
you can find similar way if it is linux..
ps u -Lp wlspid and thread dump
Thank you,
Bob
Edited by: Bob on Sep 21, 2010 10:18 AM
Edited by: Bob on Sep 21, 2010 10:24 AM

Please assist - strange prstat / unable to kill process (-9)

Hi Guys,
I have some questions and I hope someone here will be able to help. Kindly refer to below for my prstat output.
1. I have this ora10g process taking 8.3 CPU (sftp-server2/1). Just now, our dba attempted to perform a sftp from a remote macine to this machine, however, the process has already been terminated. We tried to perform a kill -9 <pid>, however, the process is still there? Any other way to kill it since rebooting the mission critical server is not an option.
2. From prstat -a, we can see the total RSS by ora10g is 362G. It won't be possible right as our server has only 32 GB of memory. Anyone can advise? Is it normal?
prtdiag:
System Configuration: Sun Microsystems sun4u Sun Fire V890
System clock frequency: 150 MHz
Memory size: 32768 Megabytes
prstat -a:
PID USERNAME SIZE RSS STATE PRI NICE TIME CPU PROCESS/NLWP
15443 ora10g 3152K 1608K cpu1 0 0 2:12:45 8.3% sftp-server2/1
20170 nsm 1730M 1445M cpu4 0 0 16:10:22 7.1% oracle/1
20139 nsm 642M 449M sleep 59 0 2:45:36 3.4% oracle/1
27064 ora10g 641M 447M sleep 59 0 0:00:17 0.7% oracle/1
20147 nsm 642M 549M sleep 59 0 1:54:47 0.7% oracle/1
29529 nsm 2135M 1030M sleep 59 0 0:01:12 0.6% oracle/1
20175 nsm 642M 449M sleep 59 0 1:34:41 0.4% oracle/1
20161 nsm 642M 449M sleep 59 0 1:20:21 0.3% oracle/1
20145 nsm 642M 449M sleep 59 0 1:21:10 0.3% oracle/1
20150 nsm 642M 449M sleep 59 0 0:50:01 0.2% oracle/1
10230 oracle9i 2143M 1015M sleep 59 0 3:04:46 0.2% oracle/35
28159 nsm 642M 453M sleep 59 0 0:01:06 0.2% oracle/1
20171 nsm 643M 449M sleep 59 0 0:44:10 0.2% oracle/1
20172 nsm 1730M 981M sleep 59 0 0:46:05 0.2% oracle/1
20144 nsm 450M 349M sleep 59 0 0:52:51 0.2% oracle/1
NPROC USERNAME SIZE RSS MEMORY TIME CPU
56 nsm 26G 18G 3.1% 33:25:26 16%
513 ora10g 503G 362G 65% 22:26:19 9.6%
183 oracle9i 373G 180G 32% 4:24:06 2.0%
87 root 489M 192M 0.0% 2:58:27 0.4%
7 jgxu 20M 13M 0.0% 0:00:22 0.1%
Total: 850 processes, 8444 lwps, load averages: 3.35, 2.83, 4.06
Please advise.
thanks

Hi,
It may b possible coz of some temproray issue or may be dispatcher connection is closed.. first try starting java stack , if it doen't work out try starting dispatcher and the j2ee stack.
Regards
Aashish Sinha

JWSDP 1.6 install issue on Solaris 2.8

I am attempting to install JWSDL 1.6 on Solaris 8 and am getting the following message:
This machine is missing the following Solaris(TM) patches:
108940-24 or higher
This patch is installed.
I believe that the error is due to the install script running the comand 'showrev -p' which does not list the installed patches.
Does anyone have work around or fix for this?
Thanks,
Jeff

I found a fix. More acurately, I fixed the install script.
It turns out that the "showrev -p" command produces different output on Solaris 8 than it does on solaris 9 or 10.
I changed the installation script to use "patchadd -p" instead and all is well.
Jeff

URLDecoder behaviour in Solaris

I run following piece of code in solaris and windows env. and get different results. Pls can someone give idea what is wrong in my solaris setup.
//---------CODE START--------------
public class TestEncBasic
public static void main(String arg[])
String strURL = "%DC" ;
String strTest = java.net.URLDecoder.decode(strURL);
for (int i = 0 ; i < strTest.length() ; i++)
System.out.println("char =" + strTest.charAt(i) + " int =" + (int)strTest.charAt(i) ) ;
System.out.println("System encoding =" + System.getProperty("file.encoding"));
// ---------CODE END----------
Output in Solaris :
char =? int =65533
System encoding =646
java full version "1.3.1_06-b01"
OS : SunOS amazon 5.8 Generic_108528-03 sun4u sparc SUNW,Ultra-4
Output in Windows2000 :
char =▄ int =220
System encoding =Cp1252
java full version "1.3.1_04-b02"
Output in Linux :
char =� int =220
System encoding =ISO-8859-15
java full version "Blackdown-1.3.1-FCS"
OS: Linux conjurer 2.4.18-3 #1 Thu Apr 18 07:37:53 EDT 2002 i686 unknown

The URLDecoder method you used uses the default charset encoding for the system you are on. You should use the method where you can specify the charset. a good charset to use is UTF-8 since it is a charset that java supports.
//---------CODE START--------------
public class TestEncBasic
public static void main(String arg[])
String strURL = "%DC" ;
String strTest = java.net.URLDecoder.decode(strURL, "UTF-8");
for (int i = 0 ; i < strTest.length() ; i++)
System.out.println("char =" + strTest.charAt(i) + " int =" + (int)strTest.charAt(i) ) ;
System.out.println("System encoding =" + System.getProperty("file.encoding"));
// ---------CODE END----------
This code should return the same answer for all systems

Solar Powering a 17" MacBook Pro

Hello,
I'm attempting to solar power my 17" MacBook Pro while away tent camping. I have a Brunton 52 Watt folding solar array, Brunton 12V DC-to-AC inverter, Solar controller, cigar plug cable for output from solar array. The output from the solar aray in full sun is 24V DC. I can lower the voltage to 14V if I cover up 2/3rds of the cells.
I've connected the DC side of the 12V inverter to the solar array and the AC side to the MacBook Pro power brick. The inverter is between the solar array and the MBP. With solar array in full sun and connecting the three pieces, there is no AC power output from the inverter going to the MacBook Pro brick. Can anyone answer why I have no 120V output from the AC side of the inverter to the MBP?
Thanks,
Peter

If you are going to use your MBP power adapter you are forced to do the DC to AC conversion and deal with the inherent power loss (10%-15%).
You need to calculate the real world output of the solar panels (not the advertised figures) and take into account that the amperage will fluctuate with weather conditions and panel orientation.
My guess is that you will need a minimum of 5 amp output (using 24 v output panels) plus a fudge factor to take into account less than optimum conditions. The good news is that more panels can be added if the need so requires.
Good luck.
Ciao.

Prstat output in Solaris 11

Similar Messages

Maybe you are looking for