Incremental archive running as full
Posted by Andre Kuehnemund on 08 February 2016 19:41
The P5 log showed the following:
... Notice: ARCHIVE: plan '10008', host 'localhost', filter '', pool 'AB_Archive', copypool '', index 'Default-Archive'
The default index options of an archive plan are:
Index path options: 'Original path'
Path prefix: 'No extension'.
These default options keep P5 in charge of managing the archive index. Any file that was archived can be found under its original source path. In case it has been archived multiple times, P5 can easily detect this by looking at the archive snapshots it manages.
"Reverting to full due to alternating root" means that the customer used a custom index root path under which the data to be archived was going to be inserted into the archive index database. What this does, though, is that it makes it impossible for P5 to know whether or not a file has been backed up previously - as the same file could have been archived one time under index path '/a/b/c/d', the next time under '/b/c/d', the next time under '/b' etc. An incremental archive cannot not work under such conditions. The only way for P5 to determine whether or not a file had been archived, would be to compare every file to be archived with every file entry in the archive index database - something that would simply not be feasible.
In case your archive plan is configured with the standard settings (i.e. 'Original path' & 'No extension'), please be aware that specifying an index location for a manual archive job does the same as specifying a custom index root path in the archive plan you're using. The custom path specified is where P5 will insert the data to be archived into the index database. Therefore, when running manual archive jobs that you wish to run as incremental archive jobs - make sure the archive plan index settings are set to the above defaults AND you're not specifying a custom index path in the manual archive settings.