Linux: Continuously synchronize files, one way

Question

Scenario: An IDE is set up on a Linux desktop box, editing PHP files locally. Every time I save a file, I want this change to appear on the linux server where Apache is running. The server has ssh (and samba and nfs for that matter).

As a reference, when I edited files on Windows, I finally came over WinSCP as the exact tool I needed - WinSCP have just this feature present, with initial synch and then continuous update, using the filesystem watch service: "Keep Remote Directory up to Date".

On Linux, one could argue that sshfs could be employed to sidestep the need for synchronization entirely. On windows, a samba-share would do the same. However, I want the IDE to work with local files (on a SSD disk!), not having to go over the network to do PHP indexing and whatnots, which takes ages.

But sshfs might be a part of the solution nevertheless - so that the continuous synchronization just needed to be done between two local directories.

Any ideas or pointers?

Answer 1

You can also use inotifywait from the inotify-tools package.

inotifywait -r -m -e close_write --format '%w%f' /tmp | while read MODFILE
do
    echo need to rsync $MODFILE ...
done

Answer 2

If you need to observe filesystem, then inotify is the way to do it. I would write a simple python script using pyinotify to perform sync when filesystem get changed. See documentation. You might also checkout the autosync.py for some inspiration. Have fun.

Answer 3

What I did once is have a bash script running ls -l in a loop (with some sleep) and comparing to the previous output. If it changed, do your synchronization.

#!/bin/bash

listcommand="ls -l $*"

newfilelist=$( $listcommand )
while true
do
   if [[ $oldfilelist != $newfilelist ]]
   then
      oldfilelist=$newfilelist
      # run your synchronization tool
   fi
   sleep 10 || exit 2 
   newfilelist=$( $listcommand )
done

Start this script in a new terminal with the file names as arguments (after putting in your synchronization tool).

(I used this to start a compilation, not to synchronize, but this would work a similar way.)

Answer 4

I suggest you to use an SCM System like Git or Bazaar that way you can solve two issues at the same time, including versioning control and a lot more features.

Answer 5

Much more simple approach:

Export your /var/www with samba (or nfs) and work on the files directly on the server.

Another solution: most IDE's allow various deployment configuration - check if there is one that suits you.

Or set a Source Code Management system like Git, Bazaar, etc.

Good luck !!