On Monday 28 July 2014 22:29:19 Nathan Stratton Treadway did opine And Gene did reply: > On Mon, Jul 28, 2014 at 04:38:32 -0400, Gene Heskett wrote: > > So I'm still standing here on my stump and proclaiming that tars > > "--one- file-system" implementation is broken. And has been since > > somewhere post 1.22. > > On Mon, Jul 28, 2014 at 05:08:38 -0400, Gene Heskett wrote: > > Crazy. There is no evidence of that shown by an ls -l being executed > > anyplace in that tree. > > Gene, you asked about this on the bug-tar list back in March, and I > thought Paul succeeded in convincing you that all the directories that > cause these "directory is on a different filesystem; not dumped" > message are indeed mount points: > http://lists.gnu.org/archive/html/bug-tar/2014-03/msg00006.html > > Note that in a way the point of the Unix single-root file hierarchy is > that you don't notice in normal operation (i.e. "ls -l") where one > filesystem ends and the next begins (unlike, say, Windows, which has a > separate drive letter assigned to each filesystem). > > (Also, if in fact these directory entries _were_ symlinks, they > actually would show up in the "ls -l" output, with "-> <target>" at > the end of the output line.) > > As others have mentioned, you can easily tell what directories are the > roots of mounted filesystems by looking at the output of the "mount" > command (as you listed in > http://lists.gnu.org/archive/html/bug-tar/2014-07/msg00041.html > ). You'll see that all the directories getting these "different > filesystem" messages in your amreports are indeed mount points listed > there. > > All the mounts of type "proc", "sysfs", "tmpfs", etc. (or a device name > that doesn't start with "/dev/") are ones managed by the Ubuntu boot > process for various system purposes. They are created afresh each time > the system is started up, and so (as Joi mentioned) it doesn't do any > good to include them in your Amanda dumps. > > > In any case, keep in mind that (except for a few strange-case bugfixes) > the only one-filesystem-related thing that changed between earlier > versions and tar 1.27 is that tar now outputs a warning message > regarding the sudirectories that it runs across but doesn't back up due > to the filesystem check. In other words, these directories weren't > included in the dumps with tar 1.22 either -- it's just that they were > excluded silently. > > (So if indeed something was missing after you restored from a backup, > that issue has some other explanation....) > > Nathan Since it appears you are correct, maybe I should be asking for a --oh- shaddup switch too? :)
If one is printing the report as I had been for quite a while since a 32 dle backup would fit on one page, it would at least try to keep the nozzles clean on an ink squirter. But it also used about $400 worth of ink a year just to keep the nozzles semi-clean. Then tar started to get mouthy and it took 2 pages to print it. Processing all that, for 2 years worth of Epson ink, I could get a Brother HL-3170CDW color laser and one full set of spare toners, which has proven to be decent at color, runs 10,000% more dependable on paper feed, and despite my using it for book publishing occasionally, still is running on one of the original 4 toners in it 2 years later. So in a roundabout way, tar's mouthyness coaxed me into buying a better printer. And if the excludes work tonight, I'm happy, camping or... Thanks Nathan. Cheers, Gene Heskett -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author) Genes Web page <http://geneslinuxbox.net:6309/gene> US V Castleman, SCOTUS, Mar 2014 is grounds for Impeaching SCOTUS
