HP-UX

Hello,

I have a very big directory with lots of small files. Just doing a single "ls" takes eternity, so I even can't know how many files the directory has.

I need to get the newest files in the last 15 minutes, but I can't get an eficient way of doing it. I've tried find, perl, shell script, etc. Is there any more eficient way of doing this? Maybe avoiding reading all the files, or going to the end of the directory...

PS: Using HP-UX B.11.31 U ia64

Any help?
Thanks.

To count the number of files in the directory. A "find" does not sort the output and will be quicker than "ls".

Code:

find . -type f -print|wc -l

To find out the size of the directory file itself:

Code:

ls -lad .

A directory file above one megabyte is pretty large. I would expect say 25,000 files in a megabyte size directory.
The performance of a mature directory can be improved by periodically re-creating the directory file.

Well, my directory file as 180MB ...

Code:

$ ls -lad .
drwxrwxr-x  10 interf     sapsys     189300736 Jun  4 17:57 .
$

There is no really easy to do it other than finding a more efficient data storage structure.

However it you are using vxfs and are deleting files are an amount of time has elapsed you could try a directory defrag to see if the directory is bigger than it needs to be and reduce it if possible. The is not the current number of files but the highest number ever that matters.

Any luck with finding out how many files you have?
What processing you you need to do in the directory?
Is archiving an option?

The inode count is the total number of files and directories on a filesystem. It is possible to run out of inodes long before you fill a filesystem and well worth being aware of how many inodes you have left:

Code:

bdf -i