On Wednesday 30 January 2002 01:27 pm, John R. Jackson wrote: >>I wonder how hard it would be to train amanda to rotate holding >>areas (partitions) on a raid instead of tapes? ... > >That's already part of 2.4.3 (or the amanda-242-tapeio CVS branch).
Humm, might be a good thing to know. How-to Docs location? Back on the trail of this miss-behaving drive, I was right, in an afternoons worth of messing with this thing, both by hand where it works nominally with dd, mt and mtx, and mt reports no softerrors ever, but let this same amanda-2.4.3b2-20020129 or one of the earlier ones too have a whack at it and it spends half the day running to the cleaning tape in slot 4! Useage count just for today is about 52 since it will normally increment it by 4 for each full run of amcheck or similar scanning the slots util. Immediately after amanda has errored out, an mt -f /dev/nst0 status will return a BOT ONLINE IM_REP_EN with a 41010000 status, and if BOT isn't valid, its 81010000. Two things enter into this that might be contributors to the problem, first being the revision number of this particular Python 28849-XXX drive, which is 4.CM according to dmesg. The second is that originally the dc dipswitch was on when I labeled the tapes, and I've been fighting with that ever since. I did succeed in getting a title header written to one out of 6 tapes that doesn't turn the compression led back on when a "mt -f /dev/nst0 rewind && dd if=dev/nst0" is done. BTW, I can do that without error 50 times in a row, and even wrote one nearly 20 meg file with tar and recovered it, again using tar. dd also works if the file is some even multiple of 512 bytes long. If amanda wasn't so concerned with the cleaning tape, I think I'd have a working system, but when amanda sends it to slot 4 and then seems to forget it has to reload the correct 'next' slot when doing a show or such, so it says theres an io error for every slot, usually accompanied by a 'while rewinding' when there isn't even a loaded tape! I haven't tried taking the cleaning tape out of the configuration yet. This same code, compiled on my machine at home, using the same model of drive but possibly a newer rev version, runs quite nicely now. Cheers John & Thomas, Gene
