Search Results

Providing iozone reads and writes more data than you have RAM available then your tests should be correct.

The problem occurs when you read or write that which could potentially exist in disk...

If you're using the Net::FTP method (above) then you can provide the user/pass details within the perl script and it'll handle the login for you.

Sure, we don't have compilers on all the production machines we look after, but we have a build box which is of the same arch, so once compiled there it goes into my home dir which gets copied to all...

You shouldn't need administration, just a login really. Coreutils can compile happily in your home dir :)

When I find that date can't do things like this I often just build GNU Coreutils on the destination. The time it takes to compile is often far less than tweaking to get the date handling right myself...

The trouble with attempting to test disks is that you can never really be certain that you're reading data that hasn't been cached somewhere.

A good way to try and dump what's in cache is to be...

You can use date itself to perform this,

$ for i in \
"17-jun-2013" \
"29-mar-2013" \
"26-Jul-2012 04:00:00" \
"Mar 26 23:59:59 2012" \
"27-Jul-2012 07:56:43" ; \
do \
date -d...

If you're already executing a perl script, why not make the perl script use the FTP module and that way control the flow of execution through the FTP upload? Calling a bash script doesn't make it too...