removing files and changing the ownership of the file is taking long time

[prasanth.ps]

Hi,

This is regarding some perfomance issue am facing in the recently loaded HPUX servers.
We have loaded the hpux 11.31 in BL 860 servers.we presented a mount poin from the SAN (EVA 4100).
This mout point is 100 gb in space and file occupied is 50 gb. we are trying to delete the contents of the mount point it is taking long time.

Same issue am facing while changing the owner ship of the mount point.The preswent server is having dual itanium 2 cpu and 8 gb of memory and swap space i assigned is 16 gb.

If am running the top in the server in that time its giving the 90 % idle while doing the above process.

Can u please tell some remody for this problem? I s it becuase i assigned 16 gb as swap space?

Thaxn $ rgds,

Prasanth.

[Dennis Handly]

>we are trying to delete the contents of the mount point it is taking long time.

There are other threads that say if you want to delete the whole filesystem, it is faster to newfs than "rm -rf".
http://forums.itrc.hp.com/service/forums/questionanswer.do?threadId=1287338

>Is it because I assigned 16 gb as swap space?

I don't see how this will slow things down, especially if you aren't under memory pressure.

[Steven E. Protter]

Shalom Prasanth,

I've not seen any definitive data that can allow me to help you.

There are many possible causes of the problem you write about:

* Too many files in one directory, which causes great stress when any operation is run against any file.

* EVA configuration issues. Is the LUN raid 1, raid 5 or some other configuration.

* I don't see how swap space is relevant without seeing data on memory use. swapinfo -tam at a most basic level. swap space allocation has little impact on what appears to be an input/output issue.

Need more information to help you.

SEP

[prasanth.ps]

Hi All,

This particular mount point is having the oracle database instance and its in vraid 5 in SAN stoarage.

Its HP UX 11i v3 so we were using the native multipathing for adding the alternate path in the volume group too.

If any more input u want please ask me.

aThanx and rgds,

Prasanth.

[Bill Hassell]

> This particular mount point is having the oracle database instance and its in vraid 5 in SAN stoarage.

The question still remains: how many files are you changing or removing? Are you using -R option for chmod? Are you using rm -r? If you don't know how many files and directories are in this mountpoint, use these commands:

find /old_oracle_moiuntpoint -type f | wc
find /old_oracle_moiuntpoint -type d | wc

If there are more than a few hundred, don't bother removing the files. Just umount the mountpoint, run newfs on the raw volume and mount it back. Billions of files and directories can be removed in a 1-2 seconds with newfs.

[prasanth.ps]

Hi All,

Thanx for the immediate response.
We cant do the newfs because in this mount point 2 instance is running we wnat remove only one instance.

So i feel only the option is delete the files and directories.

i didnt use the -R option with the chmod for changing the permission.

I will check the total no of files while today.

Thanx and rgds,

Prasanth.

[Dennis Handly]

>I will check the total # of files while today.

I'm not sure how helpful this will be, other than you can schedule your nap when you do the rm or chmod. :-)

[prasanth.ps]

Hi,

I had my nap as u wish.

:)

Atlast we decided to go for change the two instance in s same mount point.So we can play with the filesystem command as i wish.

Still am not unserstanding that what is the difference between the deleting technology in both windows and unix. Deleting "windows" files(same size and number) is taking less than the Unix One.


Thanx to all of ur support.

Thanx and rgds,

Prasanth

[prasanth.ps]

Thanx to all