oracle login slow

[Akif_1]

Hello,

OS:HP-UX 11.11
Oracle Application server
RAM:8GB
swap:12GB
Kernel parameters:
nkthread 7184
nproc 4096

total processes : 1179
Total forms connections : 744
Total HTTP server processes: 242


Iam facing a problem of Oracle Application 9.0.1.4.0 process damm slow access to the oracle login users via java applet.

Although the RAM usage is only approx 15% , But swapping is higher, Why it is not consuming more RAM for fast processing.

Does this require to change "any kernel parameters"

Kindly find below more details about the memory utilization:


total processes : 1179
Total forms connections : 744
Total HTTP server processes: 242

# swapinfo -tam
Mb Mb Mb PCT START/ Mb
TYPE AVAIL USED FREE USED LIMIT RESERVE PRI NAME
dev 4096 734 3362 18% 0 - 1 /dev/vg00/lvol2
dev 4092 738 3354 18% 0 - 1 /dev/vgswapapp1_1/lvswap1_1
dev 4092 738 3354 18% 0 - 1 /dev/vgswapapp1_2/lvswap1_2
localfs 800 0 800 0% 800 0 1 /logs/paging
reserve - 9625 -9625
memory 6301 762 5539 12%
total 19381 12597 6784 65% - 0 -

#Regard's

[Akif_1]

Hi All,

Any valuable input's will be appreciated .....

[Kenan Erdey]

Hi,

you might have been started jvm with low memory, as connection increases jvm makes more garbage collection with more durations. look at the oc4j logs in opmn/logs directory anc check GC durations.

[Akif_1]

Hello,

I appreciate your quick response,

The 172.1.100.4 belongs to infrastructure DB node.

Kindly find an attached for your review and feedback to over-come with this issue .

Thanks& Regards

[Akif_1]

Attached file :

[Dennis Handly]

>But swapping is higher, Why it is not consuming more RAM for fast processing?

You can't tell if it is swapping unless you provide the "vmstat 1 8" and look at the "po" column.

[Akif_1]

vmstat 1 8 Output:
=================

procs memory page faults cpu
r b w avm free re at pi po fr de sr in sy cs us sy id
54 1 0 2294623 99604 55 0 6 8 0 0 50 753 6057 1586 8 5 88
54 1 0 2294623 98919 411 1 35 0 0 0 0 2289 12734 7550 35 14 51
54 1 0 2294623 98141 348 0 28 0 0 0 0 2280 12726 7288 33 17 50
54 1 0 2294623 97525 329 0 23 0 0 0 0 2197 12648 7367 29 15 56
54 1 0 2294623 97437 307 0 24 0 0 0 0 2117 12356 7164 20 12 68
54 1 0 2294623 98032 253 0 19 0 0 0 0 2000 11877 6758 19 9 72
54 1 0 2294623 97307 192 0 13 0 0 0 0 1896 11460 6408 28 13 59
54 1 0 2294623 100893 158 0 19 0 0 0 0 1845 11081 6369 22 14 65

[Akif_1]

memory page
avm free re at pi po
2294623 99604 55 0 6 8
2294623 98919 411 1 35 0
2294623 98141 348 0 28 0
2294623 97525 329 0 23 0
2294623 97437 307 0 24 0
2294623 98032 253 0 19 0
2294623 97307 192 0 13 0
2294623 100893 158 0 19 0

[Dennis Handly]

>avm free re at pi po
>2294623 99604 55 0 6 8
>2294623 98919 411 1 35 0

Basically you aren't paging out (swapping).

[Akif_1]

Hi Dennis,

But below output shows system swapping 65% where as real time memory RAM only 12%,


total processes : 1179
Total forms connections : 744
Total HTTP server processes: 242

# swapinfo -tam
Mb Mb Mb PCT START/ Mb
TYPE AVAIL USED FREE USED LIMIT RESERVE PRI NAME
dev 4096 734 3362 18% 0 - 1 /dev/vg00/lvol2
dev 4092 738 3354 18% 0 - 1 /dev/vgswapapp1_1/lvswap1_1
dev 4092 738 3354 18% 0 - 1 /dev/vgswapapp1_2/lvswap1_2
localfs 800 0 800 0% 800 0 1 /logs/paging
reserve - 9625 -9625
memory 6301 762 5539 12%
total 19381 12597 6784 65% - 0 -

Regard's

[Akif_1]

Any valuable input will be appreciated .....

[Dennis Handly]

>But below output shows system swapping 65% where as real time memory RAM only 12%,

This isn't what that means.

TYPE AVAIL USED FREE USED
dev 4096 734 3362 18%
dev 4092 738 3354 18%
dev 4092 738 3354 18%
localfs 800 0 800 0%
reserve - 9625 -9625
memory 6301 762 5539 12%
total 19381 12597 6784 65%

What this says is:
65% of your total swap has been reserved.
Only 12% of pseudo swap is being used.

So this means that you may have asked for 12.6 Gb of VM. You may have only touched a very small part of it.

So you have reserved 12.6 Gb of VM but you haven't started actively swapping the whole 12.6 Gb. From vmstat(1m) po field.

You have reserved 9.6 Gb and only used 2.1 Gb of your device swap.

[Akif_1]

Thanks Dennis for your feedbackup ,

How can i use this memory completely for smooth production.

I mean which area need to be workout to over come with this issue.

Thanks & Regard's

[Akif_1]

Hello Dennis,


Is this bcz i change dbc_max_pct=20 from dbc_max_pct=50 where as on the other node like database dbc_max_pct=35 from dbc_max_pct50.

And the DB memory consumption is 77% swapping 44%.

Do i need to make the application dbc_max_pct=35 , so it use the memory as DB.

Regard's

[Dennis Handly]

>Is this because I change dbc_max_pct=20 from 50 where as on the other node like database dbc_max_pct=35 from 50.

>Do I need to make the application dbc_max_pct=35, so it use the memory as DB.

I assume you don't have Oracle on this machine?
Yes, you could try increasing it.

[Arockiasamy K]

Dear Akif,

Please check the resolv.conf, wheather the DNS server is reachable or not.

That maybe a problem.

Regards,
Arockiasamy K