>On one hand, I have been thinking about running amanda once a week for >totals directly to tape, and the rest of weekdays without tapes so >incrementals are sent to holding disk, and then before next total on >friday with tapes, I do amflush...
FYI, 2.4.3 (now in beta testing) has an "autoflush" parameter that will do the flush for you as part of the normal run. Normally you would have to use one tape for the amflush and one for the amdump. If you don't have all that much data, you might get by with just one tape for both (but only if you use 2.4.3). >... Also, with amflush, will indexes be maintained without >any trouble? So index server finds stuff as if they were on separated tapes. It will work fine. A datestamp is associated with the amdump (not the amflush) and maintained as part of the index and used during a restore. Even though you'll have several backup images of each file system on the flush tape, they will have unique datestamps. And if you do a restore before the amflush, amrecover will find the images in your holding disk and use them directly from there, without any tape needed. Also, Jon LaBadie is correct that Amanda tries to balance the amount of tape used for each run, not maximize it. John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
