memory issue on HPUX 11.23 ia BOX--

[Trng]

Hi ,

my OVO is reporting high physical memory utilisation my 11.23 box.i checked with swapinfo -atm.output pasted below

dev 8192 2177 6015 27% 0 - 1 /dev/vg00/lvol2
reserve - 3638 -3638
memory 4066 1614 2452 40%
total 12258 7429 4829 61%



but kmeminfoIA reported 100% memory utilisation.whay such difference ..which should i believe.

Physical memory usage summary (in page/byte/percent):

Physical memory = 1040926 4.0g 100%
Free memory = 14670 57.3m 1%
User processes = 452317 1.7g 43% details with -user
System = 563231 2.1g 54%
Kernel = 324562 1.2g 31% kernel text and data
Dynamic Arenas = 119839 468.1m 12% details with -arena
vx_global_pool = 23718 92.6m 2%
spinlock = 14394 56.2m 1%
vx_inode_cache = 13892 54.3m 1%
VFD_BT_NODE = 7652 29.9m 1%
vx_buffer_cache = 6688 26.1m 1%
Other arenas = 53495 209.0m 5% details with -arena
Super page pool = 47801 186.7m 5% details with -kas
Static Tables = 129715 506.7m 12% details with -static
nbuf = 75904 296.5m 7% bufcache headers
pfdat = 24396 95.3m 2%
vhpt = 8192 32.0m 1%
text = 7349 28.7m 1% vmunix text section
inode = 4896 19.1m 0%
Other tables = 8977 35.1m 1% details with -static
Buffer cache = 238669 932.3m 23% details with -bufcache
UFC file mrg = 0 0.0 0%
UFC meta mrg = 0 0.0 0%






is there any inbuilt utility to check physical memory utilisation(pls excuse glance from this thread)


regards
skr

[Venkatesh BL]

About 933MB (> 20% of RAM) is used by the buffer cache. What are the dbx_min_pct and dbc_max_pct values? May be you can tune those a little.

[Trng]

it is

dbc_max_pct 50 Default Immed
dbc_min_pct 5 Default Immed

[Jeeshan]

set your dbc_max_pct to 20.

[Emil Velez]

You are using 1 GB of your 4 GB of memory as buffer cache.. This is not really necessary since you can have that memory used by processes.

Change DBC_max to 15 or so so the DBC only uses 600 MB maximum

[Trng]

but anyone can say why kmeminofia output and swapinfo output showing different physical memory utilisation ??

[Gokul Chandola]

Hi,
Both are different command and for different time the uses of Swap is/ may change.

Regards,
Gokul Chandola

[Tim Nelson · Accepted solution]

This line.
dev 8192 2177 6015 27% 0 - 1 /dev/vg00/lvol2

shows that at one point in time, since the last reboot, that 2177KB worth of processes were swapped out to disk and have space reserved. It does not mean / show what is currently happening. Use either glance or vmstat to review the Page Out rate.

And as a general practice always use stats that have been collected over a period of time, not just snapshot at a moment in time.

e.g. vmstat 5 5

[Trng]

my customer not allowing to install glance ..is thery any inbuilt utility to check physical memory utilisation on hpux 11.23,dont want any script .

regds
skr

[Bill Hassell]

> my customer not allowing to install glance ..
>is thery any inbuilt utility to check physical memory utilisation on hpux 11.23,
> dont want any script .

Well, the standard tools are very cryptic and require a lot of knowledge to interpret. The commands are:

ipcs -bmop
vmstat 2 10
top
sar -w 1 10
UNIX95=1 ps -ef -ovsz,pid,ppid,stime,args

None of these will give you a complete picture of what is happening. You'll need to learn about shared memory, process spaces, paging. The HP-UX Performance book is a good start.

[Dennis Handly]

>my customer not allowing to install glance

Because they don't want to spend the money? There is a trial version.
Also, they might as well take that money and buy more memory if their performance is bad.