vmstat

[Roshan]

Solaris Sparc 11

Hello Team,

I have a few questions as per link below

Identifying High CPU Usage with vmstat - High CPU Usage
http://www.dba-oracle.com/t_tuning_cpu_usage_vmstat.htm
Identifying High CPU Usage with vmstat - High CPU Usage, identifying ram memory bottlenecks, ..


"To aid in this, the vmstat utility gives the sr column to designate the memory page scan rate. If we see the scan rate rising steadily, we will have hit the page-stealing daemon's first threshold, indicating that entire program's RAM memory regions are being paged-out to the swap disk. As the process continues, we will begin to see high page-in numbers as the entire process in paged-back into RAM memory."

The sr increases from 3-21 and pi from 1 to 10 does this mean memory is a concern?

From my system the output vmstat shows:

ora12c@dware1:/ogg/ora12c$ vmstat 6 10

 kthr   memory      page      disk     faults   cpu

 r b w  swap free re mf pi po fr de sr am s0 s1 s2  in  sy  cs us sy id

 0 0 0 152543200 21845880 1296 2001 124 158 137 0 0 14 1249 55 246 60751 40872 196621 6 9 85

 0 0 0 150007936 18537680 1303 4071 0 0 0 0 0 2 750 82 11 57776 56985 130320 7 8 84

 0 0 0 150183040 18669584 521 1398 0 0 0 0 0 4 607 48 11 52178 22613 130569 5 8 88

 0 0 0 150180248 18664264 574 919 0 0 0 0 0 5 569 40 12 56273 18087 202522 6 9 85

 0 0 0 150175584 18664872 318 601 0 0 0 0 0 4 736 31 10 59895 17471 284399 8 12 79

 0 0 0 150175824 18676200 316 545 0 0 0 0 0 5 362 19 12 51924 17160 72556 4 6 90

 0 0 0 150190064 18679152 250 297 0 0 0 0 0 6 349 23 11 53857 15773 114230 4 6 90

 0 0 0 150168864 18648200 1375 2894 171 0 0 0 0 6 628 70 19 57535 27296 192475 5 8 86

 0 0 0 150072624 18556888 402 1329 0 0 0 0 0 4 933 37 12 59075 25724 162592 6 7 88

 0 0 0 150013376 18510120 327 287 0 699 699 0 0 49 1008 29 11 64712 24819 142569 5 7 88



I see the SR increases. Does this mean memory real RAM memory on the Oracle server is exceeded?

ora12c@dware1:/ogg/ora12c$ prtconf|grep -i mem

Memory size: 81920 Megabytes

  memory (driver not attached)

  virtual-memory (driver not attached)

  reboot-memory (driver not attached)

"In sum, page-out (po) operations are a normal part of virtual memory operation, but page-in (pi) operations indicate that the server has excessive RAM demands."

I also see the pi almost near to the page out on my server. Does this mean memory is a concern?

Thanks,

Roshan

[ErmanArslansOracleBlog]

In Solaris, it is pi,po and sr columns which may be considered as direct indicators for addressing memory pressure. So you are checking the right columns.
But you also need to check the swapping activity.. (by running vmstat with -S argument) .. The si and so operations...
I see a decrease in you free pages.. Those values are in KBs..  But, you should still have around 17GB free memory there. So I don't think you are running out of memory, but you got a memory demand..
Having a memory demand is not a bad thing in all the cases.. These Operating Systems are designed to page out and page in.. We don't just copy all the code into the memory in one go and we always try to be ready for a future shortage..

3-21? 1-10? How do you conclude that?
Especially the SR column.. Please send me a easy-to-read output for the values of these SR column.(+ Si,SO,Pİ and PO)
Normally, The sr column under the page column indicates the scan rate. High scan rates are caused by a shortage of available memory. YES.. However; we got around 17Gb free memory there.. Normally those scans are started when the server reaches the "lotsfree".. It is set to 1/64  of the physical memory.. (unless you modified it somehow)