retitle 737564 dump: change dumpdates to support incrementals by UUID=blah thanks
doing backups for uuid- and label-backed devices has worked since the switch to libblkid (roughly around 2004), ie. "dump -0f someplace UUID=whatever" and "dump -0f otherplace LABEL=tagyoureit" has and does work fine. there are two uuid-related issues: a) the documentation isn't uptodate, files-to-dump in dump.8 only says mountpoint or device, and doesn't mention what libblkid gained us, ie. it doesn't mention looking up devices by TOKEN=value. that is easy to fix and i'll take care of that on the next upload. b), which is the core of this bug report - dumpdates always only saves the device name, hence doesn't find reordered devices. i think the least messy way out is to rework the dumpdates stuff to stop canonicalising the memorised device name, and instead park whatever the user gave us for the device lookup. ie: if you dump -u /dev/sdy then "/dev/sdy" will be saved, just as before, but if you dump -u LABEL=foobardiddly then dumpdates will hold "LABEL=foobardiddly". pro: 100% backwards compatible, and minimal amount of code change. contra: you cannot mix incremental dumps of /dev/sdxyz with UUID=whatever and expect dump to find one by the other. i think that's quite acceptable a limitation, as long as it makes it into the documentation. i'll be working on this Real Soon Now(tm). regards az -- Alexander Zangerl + GPG Key 2FCCF66BB963BD5F + http://snafu.priv.at/ Never trust a computer you can't throw out the window. - S. Hunt
signature.asc
Description: Digital Signature
