Recently I swapped in an SSD to be the new primary drive on my media center PC which was running Linux Mint 18.0 on the spinning SATA drive.
This post is basically a brief documentation of the basic steps involved in cloning or upgrading/cloning Linux Mint. Most likely this works fine for Ubuntu as well as Debian as they share a common ancestry. There are most likely limits to this scheme. I imagine things would break badly trying to do this across 17.3 and 18 for example. The base on those is a different version of Ubuntu, 14.04 vs 16.04. I might try to do a clone when the next whole number version of Mint comes along. Just pop in a drive that I don’t care about, or do it on a VM, such as Virtualbox for an experiment.
Plans
The plan is to relieve some of the storage duties of the spinning drive which was filling up. Plus a speed increase as the SSD can move data 4x faster than the spinning drive but more importantly with no moving parts the access time is minute in comparison the the spinning drive. Applications open much faster, boot time is cut by 75%, etc. If the machine needs to use swap it won’t grind down to a halt as well with a fast disk. This machine is a bit older, SATA II, but a Solid State Drive (SSD) still makes a big difference.
The idea is to clone over the home folder but exclude large data such as the ~/Music folder and leave that on the old drive and mount the drive as additional storage and use a symlink to it.
Goal
The goal of this post’s example is to do an upgrade to Linux Mint 18.3 from 18, clone over my user settings and reinstall all programs. Over the past year and a half that the machine has been in use there have been quite a few programs that have been installed on this machine. Many of them run from the command line or are libraries related to some of the machine learning code that gets run in the background on the machine. Needless to say it would be very hard to remember them and a lot of little things would be broken.
Step 1: Install Linux Mint from USB Stick or DVD
This step is pretty basic and is covered elsewhere on the web…
But needless to say you want to create a user that has the same name and User ID (UID) and Group ID (GID) as on the drive that you will be cloning from.
Step 2: Login on the new machine/drive setup kill your home directory and rsync the old one over
Mount the old drive, doing this from the GUI folder is a fine way to do it. Make note of where it mounts. You can always execute df from the command line to find where it mounted as well
It sounds crazy but, it does work. Login, open a terminal and execute…
rm -rf /home/yourusername
Everything is in memory that the OS needs right now to get through the next step, so nothing will crash and this will give you a blank slate to work with.
Next rsync over your home folder from the old drive ( /dev/sda in my case) making sure that you use the archive option. Using the v and h options as well is helpful as well to produce a lot of output in case you have to trace through a problem.
-v : verbose
-a : archive mode, archive mode allows copying files recursively and it also preserves symbolic links, file permissions, user & group ownerships and timestamps
# -h : human-readable, output numbers in a human-readable format
Example:
For me it went something like this…
rsync -avh /media/erick/B0B807B9-E4FC-499E-81AD-CDD246817F16/home/erick /home/
Next log out and then back in. Almost like magic everything should look familiar. The wallpaper on the desktop should look like whatever it was on the old setup, fonts and other desktop sizing customizations should be there. Open the browser and it should be like you left it in the old setup. It is almost like you hibernated the old setup and woke it up after teleporting it’s soul into the new drive.
But, wait the software needs attention
Step 3: Bring over the software too, sort of…
More like apt-get install it over is closer to the truth. I tried following a post on how to do this (https://askubuntu.com/questions/25633/how-to-migrate-user-settings-and-data-to-new-machine) but, what was described in it did not lead to success. The suggestion was the following…
oldmachine$ sudo dpkg --get-selections > installedsoftware
newmachine$ sudo dpkg --set-selections < installedsoftware
newmachine$ sudo apt-get --show-upgraded dselect-upgradeIt didn’t work but, it at least did the collection part. So I wound up using the first part…
oldmachine$ sudo dpkg --get-selections > installedsoftware…and then brute forced an install by doing some grep,rev,cut,rev again on the input file. Which basically flips every line in the file and removes the word “install” which is now at the beginning and backwards then flips it back over line by line.
The next line with the awk command prepends sudo apt-get install to the front of each line and saves the output to reinstall-software.sh
installedsoftware-to-apt-get-install.sh
#!/bin/bash
cat installedsoftware | grep "install" | rev | cut -c 12- | rev > cleaned-installed-software
awk '{printf "sudo apt-get install "$0"\n"}' cleaned-installed-software > reinstall-software.sh
Run the reinstall-software.sh script and it will do just what it says, install all of the software that was on the old setup. I believe there is an option for apt-get to preanswer Yes when it comes up and asks you the yes or no question about installing. I left it off so that I could review all the larger sized software being loaded. A few times I hit no by accident so had to re-run the script, no big deal.
Reboot is best now to avoid side-effects
Before going much further a reboot is probably in order as so much has changed on the machine.
For me, during the software install process, I was presented with a question about picking LightDM or another X- windows manager. I picked LightDM because I think that is what I had been using. After I was all done, I put the machine in suspend and it had a bit of trouble coming out of it, having a temporary error related to the X-windows manager. A blue screen came up and had a message about removing a temporary file. Just rebooting the machine cleared this up as the /tmp directory is flushed. Apparently this was something that was set before the upgrade, clone and software install process and did not get unset. Other than that I have seen no side effects from the process of upgrade, clone, software install.
Other Items
If you had files configured outside of the home directory such as /etc/hosts, you will obviously have to copy that over. Also, if you have any /etc/cron.hourly,weekly,monthlies that you put on the old machine. Also, it pays to do a dump of crontab’s using crontab -l > crontab-dump.txt on the old setup so they can be reconfigured to the same settings.
Cloning old to new box
This entire process can be used to clone one computer setup to another, old box to new one for example. Which brings us to…
Final Thoughts: Twin Machines
It is entirely possible to keep two machines in sync using the methods outlined above. I have not tried this but I am tempted to test it out at least. What I am thinking of is a laptop and desktop for instance. The desktop with it’s ability to hold multiple drives with ease works nice here It has one drive with the same OS as the “twin” laptop and is setup as multi OS boot. The steps above are executed, cloning the laptop setup and data to the desktop. It is entirely possible to keep cloning the home folder contents back and forth between the two to keep them sync’d. Even the software can be kept in sync using the method used above to re-install it.
It is possible to do this directly between them, both on at the same time. Or, through a server where they both get backed up to. The only caveat is overwriting and deletions. Such as care when using the –delete option with rsync. There is a potential for a race condition of sorts if settings and files get changed and then clobbered by a sync operation. If I were to try this I would start with a one direction sync. One device is the master and the other the slave. Deletions and settings changes get cloned from master to slave automatically only.
