mariotti / bmu

BackMeUp - A tool to make backups for personal use and maybe system wide too

WARNING this code is at GAMMA stage

Wrong settings can ovewrite your data.

bmu

BackMeUp - A tool to make backups for personal use. Also system wide and embedded devices.

How it currently works: we want to improve this

• install: run backmeup.install.sh
• backup: run backmeup.sh as many times as you need or set it on a cron job
• index: run backmeup.updatedb.sh any now and then or setup a nightly cron job
• search: run backmeup.locate.sh

Target Requirements

• A backup which is readable by "almost any" system. As a matter of facts the current backup system depends only on the backup device and not on the backup software. The basic software currently is "updatedb" and "locate" from the findutils package, which are widely available as default system software. There are plans to introduce additional (read to add on top of the current system) indexing software to improve the search.

• A backup which I explore by simply listing files with "ls" or with any modern file manager.

Why another backup tool?

I was, and I am still, not happy with all the tools around. Also lately I started to use a light mac machine for quick everyday work. My current linux box is even lighter, and I use it for high load programming. This made things even worst: sharing a personal backup system within OSs.

The Mac OS Time Machine was a nice discovery and finding that I can do the same with the Unix command "rsync" was even better.

What I propose is something between a backup, a time machine and close to version control if you "control" it.

Future directions

• The first is indeed to make it stable and generally usable.
• A GUI
• An archive facility to compress very old data which will still include an indexing/search facility

News

Tested on linux

Finally I got back to my linux box and tested it!

I realised the problem is not linux or mac but more gnu version or mlocate.

This will need to create an other issue. gnu or m?

bmu's Issues

Add a machine ID in the "update" backup

This is useful in the future. It has to be at first only in the timed backups from rsync. It has to default to a fixed value if it is not used. I suggest BMUMACID.

enhance locate indexing and test it

Design a common workflow. No code.

Design a common workflow and requirements for different scenarios. The least example is: 2 machines backing up the same project. Consider that a back rsync is available.

Create a lock facility

The first local approach will be simply to create a .lock file which might have to be removed by hands. The second step is to create these lock files:
.lock
.lock-corrent -> .lock
.lock-
.lock-master

.lock-master might still requires by hand removal but will be present on special maintenance operation and might wait for all other locks to be closed.

.lock- are used for concurrency, which are indeed using the .lock-master as a semaphore.

.lock and .lock-local are used mainly for local installations.

Create a command for cron

Create a command which does also the checking for the backup dir and eventually some disk space, etc.. which indeed has to run under cron.

Implement travis-ci

define indexing behaviour: when to index

implement zindex

From:
https://github.com/mattgodbolt/zindex

As for codesearch we first target the listfiles.
Then eventually the content.

Make it working for path projects with /

We might want to run:
backmeup.sh projects/bmu
This will simply not work now..

Prove it is running.

• It makes backups
• It makes dated backups
• It creates a searchable index

Create mac/linux scripts

Implement codesearch indexing

From:
https://github.com/google/codesearch

This at first should index the *.filelsit

But it might index the content of the backup too.

verify code documentation

test install script in diferent conditions

rename all variables to BMU_

create a first install script

A very first bare version has to copy files, (in HOME/bin ?) and ansk and creates backups dirs

Define installation path and run mode

We can use:
</usr, /usr/local>/bin/
</usr, /usr/local>/lib/bmu
/etc/bmu

What goes where?

Also, I want to keep a single place (directory) install.

Create a manual

Remove the GAMMA stage: what do we need to get to BETA?

by double checking source/dest, and the like.