Solaris 10 v245 Busy/Slow System

UNIX Solaris
1

Hello All,

I have noticed that one of my servers, the busiest has become increasingly slow to respond and execute commands, the running applications appear to be fine though.

Here is some output from vmstat :-

 kthr      memory            page            disk          faults      cpu
 r b w   swap   free  	re  mf 	 pi po 	 fr   de sr m0 m1 m2 m3   in   sy   cs  us sy id
 0 0 0  7141808 755736 	52  2  	 0  1188 768  0  0  0  0  0 157 1000 20150 2076 14 12 74
 0 0 0  7137720 755616 	237 1325 0  1514 1067 0  0  0  0  3 197 1260 30616 2558 20 20 60
 3 0 0  7141808 755712 	59  0    0  1321 860  0  0  0  0  4 172 1095 21384 2162 12 13 75
 10 0 0 7139320 754224 	432 2535 0  1624 1093 0  0  0  0  0 211 1478 33216 2796 23 23 53
 0 0 0  7141808 755792 	59  152  0  1513 1069 0  0  0  0  1 191 1172 24833 2250 15 17 68
 0 0 0  7136904 753064 	402 2047 0  1385 1011 0  0  79 0  0 189 1481 33519 2846 24 22 54
 0 0 0  7139416 755088 	302 949  0  1592 1091 0  0  10 0  0 210 1249 29910 2500 19 21 60
 0 0 0  7141816 755864 	195 1070 0  1931 1293 0  0  0  0  0 267 1565 34584 2834 21 20 59

I am trying to determine if this can be tuned out with system parameters or if the server requires additional memory.

Any ideas?

Wes

2

Whats the uptime of the server? If its really too long, you might want to consider getting a downtime to reboot (as an interim solution).
What Solaris OS is it running? prstat -a.. checkout the process that is consuming high CPu/Mem utilisation first
In my example below, you can see that the flarcreate command is the most resourse utilising.....

root@unknown # prstat -a
   PID USERNAME  SIZE   RSS STATE  PRI NICE      TIME  CPU PROCESS/NLWP
 29535 root     2220K 1192K sleep    1    0   0:01:57  18% compress/1
 29489 root     1352K  744K sleep    1    0   0:01:06  10% wc/1
 29532 root       10M   10M sleep   60    0   0:00:12 1.6% cpio/1
 29534 root     1500K  736K sleep    1    0   0:00:09 1.5% computehash/1
 29533 root     1492K  720K sleep   24    0   0:00:04 0.7% computehash/1
 10698 root      136M   64M sleep   60    0   0:00:52 0.0% java/27
 11292 root      156M   84M sleep   60    0   0:02:13 0.0% java/30
 11589 noaccess  269M  115M sleep   31    0   0:00:54 0.0% java/26
 29569 siraj    7888K 3948K sleep   59    0   0:00:00 0.0% sshd/1
 12530 root      177M   68M sleep   18    0   0:00:20 0.0% java/37
  9991 root      208M  139M sleep   17    0   0:00:41 0.0% java/18
 29635 root     3628K 2828K cpu1    59    0   0:00:00 0.0% prstat/1
 29579 root     2952K 1728K sleep   59    0   0:00:00 0.0% bash/1
  9354 root       33M   25M sleep   59    0   0:00:06 0.0% Xorg/1
 29601 root     2948K 1732K sleep   59    0   0:00:00 0.0% bash/1
 NPROC USERNAME  SWAP   RSS MEMORY      TIME  CPU
    80 root      499M  547M   6.7%   0:08:19  32%
     1 noaccess  195M  172M   2.1%   0:00:54 0.0%
     5 siraj    2924K 7548K   0.1%   0:00:00 0.0%
     1 lp       1016K 1816K   0.0%   0:00:00 0.0%
     1 smmsp    1136K 4472K   0.1%   0:00:00 0.0%
Total: 97 processes, 419 lwps, load averages: 1.27, 0.57, 0.22
root@unknown # ptree 29535
8788  /usr/lib/ssh/sshd
  28547 /usr/lib/ssh/sshd
    28548 /usr/lib/ssh/sshd
      28554 -ksh
        28559 bash
          28561 -bash
            28583 bash
              29438 /bin/ksh /usr/sbin/flarcreate -n archive1 -c archive1.flar
                29487 /usr/bin/awk {print $1}
                  29488 /bin/ksh /usr/sbin/flarcreate -n archive1 -c archive1.f
                    29535 compress -fc
3

Thanks for the reply, the server has been up for 13 days so not very long.

Its currently running Solaris 10 with the recommended updates, here is the showrev output.

The server uses very little CPU as most of our application work is done in memory. The server has 8GB of physical memory and approx 8GB swap available.

---------- Post updated at 01:02 PM ---------- Previous update was at 12:18 PM ----------

Here is the output from "prstat -s size -n 50", it looks like each oracle process is eating memory :frowning:

4

try the ps command to sort out using memory. Based on this we can find which oracle process is consuming more memory and what tasks are going on using these processes

ps -eo pmem,pcpu,rss,vsz,args | sort -k 1 -r | more

Regards
Ramkumar

5

By the way, Generic_118833-36 is far too old. Has so many bug fixes with this release. Please update the patch anyway,

6

I think we may have found the issue, our Oracle 10 installations were completed without adding an oracle user project to define memory usage. We have since learned that without this it will take roots default values which are listed below.

Maybe 16.0EB is too much :eek:

---------- Post updated at 05:42 PM ---------- Previous update was at 05:41 PM ----------

I am looking on the SUN Software site for the latest patch release.