mariotti / bmu Goto Github PK

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

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

by double checking source/dest, and the like.

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.

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

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.

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

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.

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.

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.

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

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

This at first should index the *.filelsit

But it might index the content of the backup too.