Rsync - What's YOUR setup like?

[StuartN]

Yes, run it as root, and make sure the permissions get preserved, but do a check to make sure the drive is mounted. As for how you do that, I'm not totally sure: google it.

This cludge will check if something is mounted:

cat /proc/mounts | grep /media/BigBadBackup
if [ $? -eq 1 ] ; then
  # BigBadBackup is mounted, so we can rsync to it
fi

But if you don't need to rsync as root, then don't run it as root. Using the minimum permission to get the job done is safer. You should use -a which is equivalent to all of -r recurse; -l copy symlinks as symlinks; -ptgo permissions, times, group and owner; and -D preserve device / specials if superuser. Using -a means your backup is pretty much identical to your original, and will function the same if restored.

[Roasted]

Using -a means your backup is pretty much identical to your original, and will function the same if restored.

That's actually kind of what I want, to be honest.

Is there any way we can break down that grep command so I understand it easier? I have 4 drives in my system, 2 of which are backups. 500/500gb and 250/250gb... the 500s are mine + backup and the 250s are samba network file storage + backup.

The mount points are:

/media/localbackup
/media/storage
/media/storagebackup

Each command is:

rsync -a --progress --delete /source /destination

How would I incorporate that grep command into each individual rsync command, seeing as though I'm running 3 commands in a bin/bash script.

[StuartN]

Is there any way we can break down that grep command so I understand it easier?

The first part, cat /proc/mounts, is a list of currently mounted filesystems - you could equally use the mount command. The output is piped into grep to search for a particular mountpoint.

Note that your second mount at /mount/storage would also match your third at /mount/storagebackup, so add the -w (whole word) option to grep.

The variable $? is (I think) a success flag. Whatever it is, echo $? shows you 0 if a command returned a result, or 1 if it did not - most commands return 0. Grep returns 0 if the string was matched and 1 if it was not.

So we just check each mountpoint in turn and run rsync if it is present.

cat /proc/mounts | grep -w /media/localbackup
if [ $? -eq 0 ] ; then
  rsync -a --progress --delete source1 /media/localbackup
fi

cat /proc/mounts | grep -w /media/storage
if [ $? -eq 0 ] ; then
  rsync -a --progress --delete source2 /media/storage
fi

cat /proc/mounts | grep -w /media/storagebackup
if [ $? -eq 0 ] ; then
  rsync -a --progress --delete source3 /media/storagebackup
fi

I have my backup script on the desktop to click whenever I want to feel like I have done something important. I could add it to crontab, but that would be too organized.

[Roasted]

Well, geez. I'm not exactly a programmer but that still is confusing to me.

I just thought there would be an easier command to execute which first checks the mount point then pipes it to run. That was just my assumption though... something like:

sudo if /media/localbackup exists | rsync -a --progress --delete /home/jason /media/localbackup

[StuartN]

Well, geez. I'm not exactly a programmer but that still is confusing to me.

My apologies, I made a mistake in that example. The tests should all be [ $? -eq 0 ], that is run the rsync if the grep was successful and not [ $? -eq 1 ] which would run rsync if the grep failed (did not return any matches).

NB: $? is set by every command, so echo $? is zero if the last command "succeeded", e.g. it is not zero after ls this_file_does_not_exist, which can be very useful.

[Roasted]

My apologies, I made a mistake in that example. The tests should all be [ $? -eq 0 ], that is run the rsync if the grep was successful and not [ $? -eq 1 ] which would run rsync if the grep failed (did not return any matches).

NB: $? is set by every command, so echo $? is zero if the last command "succeeded", e.g. it is not zero after ls this_file_does_not_exist, which can be very useful.

LOL. I wouldn't have known the difference any other way.

So, say I didn't come here and get this help from you guys. What should I do then? Should I just run the script with root and just bank on hoping the drives stay active?

I guess worst case scenario, if a drive DOES go down and the script kicks in anyway, it would just kick 300gb of data to the empty mount point, which would just dump it on my root partition and max it out. At that point all I should need to do is delete the data, or at worst - boot to a LiveCD and remove it from the unmounted mount point.

Is my train of thought correct?

[Roasted]

New question - Is it possible to rsync to the drive itself rather than its mount point?

/dev/sdb1 gets mounted to /media/localbackup

I can rsync to /media/localbackup, I got that.

But can I run my rsync to /de/sdb1?

I was just thinking, if /dev/sdb1 goes down, it's gone, so if the rsync script kicks in, it won't see it, whereas it'll see /media/localbackup regardless of whether or not it's mounted cause it's still a folder where data can be put into.

[StuartN]

New question - Is it possible to rsync to the drive itself rather than its mount point?

I am sure someone will correct me if I am wrong, but I would not think that rsync can handle a raw device. If the device has not been mounted then there is no protocol to write to it - i.e. rsync would have no knowledge of the filesystem, let alone any paths or access permissions.

[Roasted]

I am sure someone will correct me if I am wrong, but I would not think that rsync can handle a raw device. If the device has not been mounted then there is no protocol to write to it - i.e. rsync would have no knowledge of the filesystem, let alone any paths or access permissions.

/dev/sdb1 = /media/localbackup

I tried my rsync script, but plugged in /dev/sdb1 for /media/localbackup and it barfed out a bunch of errors. So I guess the answer to that is a "no."

Question - if I would utilize a program such as GRsync, would that program be able to detect if a HDD is active/mounted prior to executing the command? I'm just trying to find a more automated way to handle this, mostly because I want to make sure I do it right and these commands are kind of "whoa, wtf??" to me.

[CharlesA]

I don't know if it will know if a device is mounted or not. What I did was mount my external to something like /sync and then run something like this:

rsync -ar /storage /sync

-a = archive, copies all permissions and whatnot
-r = recursive, copying everything recursively