Click to close the image preview
My Mozy for Linux Experience so far
I wanted to put together all of my notes with the Mozy for Linux beta testing so far, including some experiences along the way. Overall, it's been a positive experience and I'm looking forward to the possibility of rolling this backup solution out across a number of systems as a standard system backup implementation.
First though, a bit about my background and perspective to help convey where I'm starting from and how I'm approaching the use of the Mozy backup software for Linux systems. Professionally, I'm a Linux System Administrator and have been working in this field for more than 20 years. I've been using Linux, of one flavor or another, for more than 20 years as well, initially with a pre-1.0 release kernel.
Backups have been a part of this world for longer than I've been in it and a constant to be sure. I've used different backup solutions over the years but, before cost-effective disk-to-disk based systems, they traditionally consisted of juggling various sized tapes by hand and by robot. The software that I've used for backups has ranged from simple tar-based scripts to full-scale commercial products and I've written more than a few of them myself along the way, some of which are still in production use.
The thing about backups in general is that they're boring, or at least they become boring over time. Nobody ever received an award for successfully backing a system up - the tapes go in to a closet or a fire-safe and don't see the light of day until the next rotation. Disk-to-disk backups similarly aren't usually an exciting endeavour and sometimes seem that they
do nothing but cost a lot and take up resources. But, all in all, they are very important and when you recover that one file that saves the day, you're the hero and it's all worth it.
Being a 'Linux guy', Mozy is new to me as a backup solution provider so getting involved in the Linux client beta had me going in with no previous experience in their platform.
Initial Setup and Installation
The first steps of the early beta for me involved getting a Mozy account for the purposes of testing the client (separate from an existing enterprise account that we were working in concert with). There was some back and forth involved there, including getting a community account setup. Once we were past those issues, I patiently awaited the first release of the client while getting some system requirements ironed out with the beta team.
I downloaded and installed the initial release on a test system, running a RHEL-based distribution. The system itself had no special requirements and was several years old by this time, but still in great shape.
Installation of the package was exactly as I was hoping it would be, a smooth "rpm -ivh mozybackup_<version_info>.rpm" command which installed the package as I would expect any well-packaged RPM to do, placing files in standard locations. Including the README file in the standard /usr/share/doc/mozyutil/ location was a plus.
Activating the software took me a few tries, due to the account setup mentioned earlier, but it wasn't long before we had that ironed out and activation worked smoothly. I was able to verify that I had an active account using the "mozyutil account" command so I was ready to test.
The initial testing steps pretty much followed the provided documentation and worked quite well:
* Adding directories to the backup list.
* Modifying /etc/mozybackup.conf to include an additional entry in the msync.pathblacklist regular expression list. I wanted to exclude both /etc/mrtg/ and /etc/awstats/ on this system so I added the following to the list: /etc/(mrtg|awstats)(/.*)?
* Running manual backups using "mozyutil start", checking the status with "mozyutil state" and "tail -f /var/log/mozy.log".
* Checking what a completed backup looked like using "mozyutil history", "mozyutil filecount" and "mozyutil lastbackup".
* Testing restores using "mozyutil download " as well as the steps as documented using the Mozy web interface.
The results from these tests were all positive. I encountered no errors at any point and everything worked as documented. I was pretty happy with the client at this point. Being able to easily and successfully backup and restore data is nice because I don't have to go in and re-invent the wheel and so far the mozyutil command was proving to be a very good tool.
Continuous Mode Testing
One feature of the Mozy for Linux client that I was really interested in taking a look at was the "continuous" mode. The phrase certainly can conjure up a variety of ways that the backup software could achieve this so I was curious to see it in action.
The primary reason that I was looking forward to this new mode was that the systems for which this solution is destined are file-servers in typical office setups where PC and Mac end-users are updating files via CIFS shares all day long. It isn't uncommon to get a request for a file to be restored that wasn't backed up the previous night, or was modified multiple times in
the morning and one specific version in the middle is needed. I was hoping that this mode would fit the bill for this particular use model and allow for the ability for more granular backups and restores.
I enabled this mode using "mozyutil continuous on" and, with a "tail -f /var/log/mozy.log" running in another terminal, started poking at the file-system. I created files in directories in, and out of, the backup list, appended to files, replaced, renamed, deleted and so on. Each action would trigger mozy-daemon to upload a new copy of the file or remove it from the Mozy side as needed within only a few seconds of my actions. Files outside of the backed up directories did not trigger any actions as
Pretty cool! Especially for a boring old thing like backups! I'm still impressed that the Mozy for Linux team has included this feature in their client. I've looked at various ways to do this sort of file-system change detection in the past and, at least for the kernels on these particular systems, nothing has worked.
The fact that their client is working on a kernel that's more than 10 years old at this point was exciting news. Continuous mode, pending a full release of the Mozy for Linux, may likely be a centerpiece of this particular file-server solution and a great selling point.
Testing on an Unsupported System
That's right, at this point we stepped up the challenge for the Mozy for Linux beta crew. Not only did we have a system that wasn't a fit for the listed system requirements, it wasn't even running a normal distribution.
At the same time that I've been testing Mozy for Linux, we've also been evaluating some NAS devices to replace these existing Linux file-servers. The specific device that I've tested to date has a completely custom busy-box based Linux distribution using a newer 3.2 kernel and unique package management system.
The first step in testing the Mozy for Linux client on this NAS device was to get the files there. This was a simple extraction of the RPM-packaged files and copying them over to the NAS system:
* rpm2cpio mozybackup_<version_info>.rpm | cpio -idmv
The above command is used to simply extract the files from an RPM package. After copying these files over to the NAS and trying to execute the /usr/sbin/mozy-daemon binary, I was able to see which libraries it might be missing. I was in luck, the only library that mozy-daemon couldn't find was 'libcom_err.so.2' but this system came with 'libcom_err.so.3.' I created a symlink to make the library visible to mozy-daemon and it started right up.
During this phase of setup, I did run in to a small problem where the mozy-daemon gave an error:
Unable to connect to mozybackup service at /var/run/mozy.sock
Verify that you have sufficient privileges and that the service is started.
The solution, from the Mozy for Linux beta team, was to remove that /var/run/mozy.sock file and start the service again. This did the trick. I had been reworking the /etc/init.d/mozybackup script to get mozy-daemon starting and stopping successfully and had to do a bit more to make sure that it was handling this file correctly as the paths in the NAS file-system layout are different than those assumed by the init script.
Once that was ironed out, testing consisted of the same steps as above including continuous mode, all of which worked as expected. All in all a great test and, pending any failures, this NAS device evaluation is quite enhanced by the ability for us to get the Mozy for Linux client running successfully on it.
During the course of the beta to date, there have been two new releases. Both have been applied to both of my test systems - including the unsupported NAS device - with no issues. No interruption in data or features, continuous mode continues to run - as is now expected - and new features have been added including a 'mozyadmin' system group to allow non-root users to use the mozyutil interface. We don't have a need for that particular feature but having worked with different role-based access/permission schemes in the past, this is a simple and effective one.
Followup Testing and Beta Team Discussions
The additional testing, past the stages described above, has consisted of letting mozy-daemon run on continuous mode and periodically checking the logs, performing restores and simply checking that it's continued to run without errors or dying for any reason.
Over the course of the beta, I've been participating in conference calls with the Mozy for Linux beta team and the team continues to impress. The developers are knowledgeable and involved in the entire process, and the ongoing updates and discussions regarding upcoming features has been promising.
Based upon the beta experience so far, I'm looking forward to continued excellence in the upcoming releases, additions of features including possible items like organizational backup sets, a data shuttle service, API access to web features and improved reporting. Some of these we've only touched on in the beta conference calls but they keep coming up so I have
a reasonable expectation that the dev team is seriously looking at all of them.
As a Linux guy - I don't do Windows - the Mozy for Linux beta team has appealed to me, the Linux System Administrator, with how they've built, packaged and delivered this solution. I didn't have to translate documentation in my head written by and for a non-technical user as it is all clear and accurate, unlike my own tendency to ramble on.
The backups *work* and continue to work and I haven't been able to break them so far using a reasonable system to test from and even an unreasonable system. The dev team didn't try to re-invent the wheel in how the software starts, runs and stops. I don't have to use the web interface for *everything*, in fact I don't need to use it that much at all - mostly just for restoring large sets of data. I like all of that.
Solved! Go to Solution.
Thanks for the in depth write up! I'll spend some time reading over this today and pass it along to the technicians so they can see all of your feedback.
Thank you Mike (@TummySupport2 ) for providing an excellent summary of your experience!
You have definitely been a great help in the Linux Beta!
As you have mentioned about Continuous Mode, this has also surpassed our expectations, namely how well it has been used and the speed. I too use it on my personal machine.
We look forward to hearing more of your experience!