I thought this might be useful data. We had a good run with the older release last night. This left the "changer" pointer pointing at slot 5, we put in todays tape, per normal practice, and when I run amcheck here are the results:
ama...@amanda:/vtapes/DailyDump/vtape5$ amcheck DailyDump Amanda Tape Server Host Check ----------------------------- Holding disk /dumpdisk: 544744 MB disk space available, using 544644 MB slot 5: not an amanda tape (Numerical argument out of domain) slot 6: read label `DailyDump05', date `20090109' NOTE: skipping tape-writable test Tape DailyDump05 label ok Server check took 7.007 seconds As you can see, it determined that what was in "slot" 5 was not valid, & then incremented to the next "slot", found a matched RAIT pair, and this also matched the next tape label in the tape list, so it is happy. But what _seems_ to happening in the new release is something different. I have seen evidence that it decrements the changer "slot" rather than incrementing it, and that it decrements by 2 instead of 1, but I can't prove these at the moment. -- One of the main causes of the fall of the roman empire was that, lacking zero, they had no way to indicate successful termination of their C programs.
