machine to machine synchronisation

[rodgoslin]

I've recently had a lot of trouble with Mageia failing. I've not found the reason, but a complete rebuild has got me going again. But it has taken almost three weeks to get the machine back to where it was. So, not having found out why, It may be software, it may be hardware, I don't entirely trust the machine, yet. I thought of setting up another PC, as a complete copy of this one. Which I've done. So for software the two machines are identical (I hope). So now I've come to the data. A synchronisation programme seems to be the answer. Unison seems to fit the bill, for syncing on remote machines. However, when I run it, it crashes , wit a segmentation fault, every time at the same point. That of defining the local directory to sync from. So I'm looking for another. Most seem to be local sychronisations (Why would you need to sync local directories?). I was going to try Bittorrent Sync, but thought to ask here first. Aside from this, it's not in the repositories, but I did install it on an earlier build, just to see hat it did.

[rodgoslin]

Running the application from the command line results in the following command line return:-

[rod@down ~]$ unison

(unison:9491): Gtk-CRITICAL **: gtk_tree_model_filter_get_value: assertion 'GTK_TREE_MODEL_FILTER (model)->priv->stamp == iter->stamp' failed

(unison:9491): GLib-GObject-WARNING **: gtype.c:4221: type id '0' is invalid

(unison:9491): GLib-GObject-WARNING **: can't peek value table for type '<invalid>' which is not currently referenced
Segmentation fault
[rod@down ~]$

I've also tried it on the second machine, with exactly the same result. Has anyone run Unison without problems?

[doktor5000]

That's a known issue, see https://bugs.mageia.org/show_bug.cgi?id=15138 and viewtopic.php?f=7&t=9900 for the fix.

[jiml8]

For protection against failures, why would you want to sync instead of just backing up? If something happens that results in software corruption on your main machine, sync will just propagate that corruption.

Sync is great when you are collaborating with others, or when you want a "hot spare" that you can failover to if necessary, but what you described as your reason for doing this sounds to me like you just need a backup system in place. That would be less expensive to set up, easier to maintain, and probably more robust in the case of most failures.

Over the years I have worked with Linux, I personally have experienced multiple and sometimes severe failures. I have even experienced multiple simultaneous hard drive failures. But, by virtue of my backup system, I have always recovered very quickly (though sometimes I could not recover until I had purchased new hardware).

[rodgoslin]

jim18, yes, I have done backups (backintime), but as I wrote in my originating post, it took me three weeks to get back to what I had before. Or nearly what I had before. On my original build, I'd got rid of all (most) those irritating bells and whistles, like active corners, and I've to go through it all again. My thought was for a second PC, in sync with the first, that I could fall back on, instantly, if I had another failure. A backup system would not solve this, I'd have to update the clone, at every backup to keep the two in step. I note that the original bug report on Unison was submitted in Jan 2015, so am not anticipating an earl solution. In the meantime I intend to try Bittorrent Sync, if I can fathom out how it works and how to use it on a server/vlient basis

[doktor5000]

I note that the original bug report on Unison was submitted in Jan 2015, so am not anticipating an earl solution.

The solution is in the bugreport and in the linked forum thread, just change the GTK theme.

[jiml8]

Rod, my point is that with a proper backup system, you can recover everything you had pretty much with one command. The time it takes is the time required to copy everything back into place from disk to disk, or from another machine to your machine. I once recovered - completely - from a double hard drive failure that carried away both my system and my hot backup of my system. It took 4 hours, including the time required to physically replace the failed drives, and boot into a rescue system, connect to the NAS that had the second backup of my system, and copy the whole thing back into place on the new hard drives. The actual copy consisted of one rsync command.

If recovering from backup took you weeks, you need a better backup system - which, of course, is your point.

Here is how I backup. viewtopic.php?f=41&t=5957

I have used this method for over a decade, and have never found a better way to do it. I have recovered - completely - from total catastrophes in minutes to hours, depending totally on how much stuff had to be copied back.

By the way, I use btsync routinely. An associate and I are working on the same project, and we find it to be a convenient way to sync some of our stuff. My NAS serves him and he syncs here, and his NAS serves me, and I sync there. Btsync is fast, and uses encrypted connections across the internet - which is important because our stuff is sensitive. However, I would not consider it a good solution as a backup; it syncs, which means a change on one system is immediately reflected on the other one. Given that many catastrophes come about because of user error (I have actually deleted /usr in the past... ) syncing for backup isn't a good plan. If you delete /usr on your system, /usr on the synced system will also vanish unless you are quick about pulling the plug. Btsync is fast; it will push the change. So when you screw up and delete /usr, it will tell the peers it is connected to that they also should delete /usr.

[rodgoslin]

It didn't take three weeks to complete the rebuild. It took most of the time getting a rebuild that worked. Originally, there were four HDD's. One with OS and user partitions, One as an overspill drive, another was a small drive, which only had a few things on it, and the backup drive. . In all, I reinstalled the OS, and data fourteen times, using three of the original disks, two others I was holding as spares and a 4TB new drive that was strictly a spare for my Drobo-fs. All went the same way. They'd lock up, or run slow. On reboot they'd only boot up into single user mode. At no point could I ascertain why they were going down. Now, I've restricted the internal drives to one, and put the backup drive in a USB drive housing, and all seems well, but my trust in the things has taken a bit of a knock. Which is why I want an exact clone to swap in if it goes down again. I've been running Linux now for about 16 years, through Mandrake, Mandriva through to Mageia, and have never had the problems as of late.