Hi there,

I'm not entirely sure whether this is a bug or a feature-request.

To reproduce this, setup a folder-structure like this:
/tmp/bkuptest/master/SeriesA
/tmp/bkuptest/isolated/SeriesA
/tmp/bkuptest/isolated/configs
/tmp/bkuptest/source/SeriesA

Put some files into the source/SeriesA folder (doesn't matter that much, I used a Copy of /bin/), and place a SeriesA.cfg into isolated/configs/ that contains the following Options:
sourceDir=/tmp/bkuptest/source/SeriesA/
backupDir=/tmp/bkuptest/isolated
mergeBackupDir=/tmp/bkuptest/master
series=SeriesA
otherBackupSeries= 0:SeriesA
cpIsGnu=yes
lateLinks=yes
lateCompress=yes
doNotDelete=yes
keepAll=30m # Just for testing purposes, would be much higher in real world scenario

You may add other config options if you want to.
Now run:
storeBackup.pl -f /tmp/bkuptest/isolated/configs/SeriesA.cfg
Afterwards run:
storeBackupMergeIsolatedBackup.pl -f /tmp/bkuptest/isolated/configs/SeriesA.cfg --force
And you will get an Error:
ERROR   2016.08.13 15:26:12  2762 didn't find a backup in /tmp/bkuptest/isolated/SeriesA, exiting

This is somewhat understandable, because the isolated-mode has not been set up properly. But since this is the first backup of this series ever (non in master, non in isolated before), all the merge-tool would have to do is move the backup-folder from the isolated-pool to the master-pool and then run storeBackupUpdateBackup.pl. (Doing so by hand works!)

Is there a specific reason the merge-tool doesn't do this? Or have you just not thought about such scenario yet? (Of course other reasons like "No time for this" are valid as well.)

Why would that even be useful?
Well, imagine you have an environment where several machines do their backups on a centralized storage, and you don't want any machine to be able to destroy all backups at once. Then the isolated-mode is a very good option, because all machines can do their backups into an isolated folder, and the server merges them to the master-pool at night, wich is presented as a read-only storage to all other machines. So everyone can access their backups, but can't destroy them.
Now imagine you get a new machine to this setup, which of course, doesn't have any backups yet. It gets to exactly this scenario.

Of course one could solve this problem in other ways (additional shell scripts), but I was curious whether there is hard reasons behind this behavior, or if it would be hard to implement such a feature.


Greetings

Lookbehind