I am going to implement a script which will use NFS to replicate data between two SCO unix servers. It will take files with mtime -1 and copy the data periodically .In this regard my questions are :
Is this approach good and reliable enough with respect to loading servers, accuracy and redundancy.
Also is there any way to detect the changed file as it changes without using a long search with find's mtime, as this searches the whole bulk of data each time it seraches for modified files less than 1 day old. If I have 100gb of data (increasing with hours) , mtime command have more data each time to search and will become slow with time.
I have more questions than answers.
Are the source files in all file systems?
Is the environment such that there is a small (relatively) number of large files involved,
or are a large number of small files that are created by various means and rarely modified.?
How many processes are there that can change a file.
Do you care if the permissions and ownership change?
Yes I have 3 file systems/partitions \(on the source server\) which are used by three different departments when they run their concerned application software/processing software.
Environment is such that daily new files are generated \(these files are large files either in small no. or large nos.\)
Typically there shouln't be more than 2 processes simantaneously changing the file data.
Permisions and ownership don't matter as such but files with changed attributes should be copied as it is on the destination server.
Basically I thought of having the eaxct replica of my source server.