Hi ,
I have 3 Dell Power edge R710 server.
In this server I will run Oracle EBS R12. Today Trying copy
One server to another two server . All three servers are taking huge memory.
1. Server1 : UAT using 32 GB RAM
2. Server2 : PRODDB Using 22 GB RAM
3. Server3 : PRODERP Using 17 GB RAM.

In my knowledge servers are not be able to free up the RAM.
Please suggest experts any immediate solution.

Regards
Sukanta Manna
Oracle DBA & Linux Admin.

What do you mean your servers can't free memory? It sounds like you're asking for advice on installing Linux. Then it also sounds like you're asking for advice on troubleshooting. Your question isn't clear. Please rephrase your question.

Hi Using this command I can free up RAM

free && sync && echo 3 > /proc/sys/vm/drop_caches && free

There was no problem - read this

Yes, as syg00 already pointed out unused RAM is wasted RAM, and file copy/move may quickly occupy all available RAM but a few hundred megabytes. Another fact might worth to add is that not all tools report this the same way. top, for example, may not show that the RAM was returned, or the GNOME System Monitor may not show that it was 'borrowed' in the first place.

In any case, even when acting on very large files or a large number of files (many times more than the available RAM+swap) if the kernel 'borrows' so much RAM there might be a related issue - such as incorrect boot options or block sizes making disk operations less than optimal, so when it happens repeatedly it might worth investigating a little.

Also, manually running free and sync is a pretty crazy move-especially since freeing up RAM like that means it writes it back out to disk. If you're manually copying data, *and* writing shitloads of RAM back to disk, you have a recipe for a mess.

What you didn't tell us is how much RAM was consumed as cache and as buffers...that's the difference between normal Linux memory utilization, and a problem.

GaWdly,

What you are asking the OP is the wrong way to go about this.. As I pointed out in my post trusting ANY tool when the basic configuration is worng IS crazy, not the other way around -asking a crazed tool to give you a sane answer. This is not UNIX only but the basic facts of life in general.

We could go about this all day long, but the fact is I signed up with linuxquestions after I have learned in a post here that some of my disks in my server I configured with the 'sync' options at boot, and started cp VirtualBox vdi files in the background from a shell, took 24 hours to forever to finish (48GB vdi files each on SATA3 internal drives, some SDD). I too trusted the tools, thus the gist of my post.

I used everything I could imagine while the server was running, I stopped all cp, mount -a, etc... until I get the damned sync out of my fstab and rebooted the server. After that 48GB files copied in seconds - not in hours or days.

The bottom line is - do not trust the tools, trust the facts and your intuition and experience.

I'm confused by your response. Nowhere did I ask the OP a question. I was merely backing others up, and providing some information about why the OP's request was being approached incorrectly.

I do disagree with you, wholeheartedly-in spite of your religious experience with ignoring tools. I'm pretty sure the facts of life don't play into memuse...

"Let's go with my gut on this tuning and performance issue, in spite of the data"-no man in a successful server environment ever said (at least when he had data available to him in order to make an informed decision).

I poke fun only to raise the tone of the thread. Tools, and performance data are very important, and to ignore them is perilous. It took you much time and many tries before you happened upon a fix for your problem...that wasn't intuition.

"What you didn't tell us is.."
say no more, it is a question, implied or not.

You read to be a very confused and upside-down person.

Assuming you are living in a flat when sub-zero temps starts to crack your bathroom pipes but the thremo dial stuck on torpical which one would you fix firts?

The OP posted high RAM usage, syg00 dismissed it, and I supported the fact by posting a personal experinece leading the issue back to STORAGE configuration. Then you come in asking for what ..? Ah, you have asked the men who's house was burning about the temperature of selected flames.

And here is the part which will surely bug you at least for a while:
Any system, being technology, science, culture, history, political or religious - deserves a full-scale investigation upon showing signs of trouble. If it was up to you I would still be growing peas in the Alps as a peasant or worse - having my throat cut and sacrificed to a planet called Jupiter.

You are not only wrong, but also ignorant.

I have re-re-re-read this thread trying to understand what you think you read here... and remain baffled. I think that you have completely mis-read GaWdly's post and suggest that you reconsider it. Perhaps it is a language barrier both ways.

In any event, the personal insults are not welcome in any context, especially here on LQ.

1 members found this post helpful.

It is not a personal attack of course.

It was the reply to my post to GawdLy from GawdLy on 06:24 which is now obviously escalating into a personal issue.

I would love to discuss this issue with you or any other member of Linuxqestions.org. The problem at hand is, what I believe, a post by me is taken as a personal attack - which was not meant to be.
Then, it became one, which is my fault.

GaWdLy

I apologize for all the things I wrote on this forum which you take as a personal attack or demeaning to your person.
Sincerely
Sandor (my real name)

I am truly sorry, but I wish the OP would show up and tell us it was STORAGE, not RAM.

Sandor, I still have no idea what you're getting on about. As long as it stops here, we're good. Apology accepted.