On Saturday 30 December 2017 15:02:05 Jon LaBadie wrote: > On Sat, Dec 30, 2017 at 08:33:42AM -0500, Ned Danieley wrote: > > and on the other side of the coin, is there any way to encourage the > > planner to do more level 0 dumps in a given day? > > > > I occasionally find, often after a kernel update/reboot, that amanda > > thinks that a large number of DLE are 'new', and fairly quickly > > moves them up to level 2. but then the planner leaves them at L2 for > > days or weeks, at the same time using less than 1 TB of my 2.5 TB > > tape. > > > > I know that I can force a level 0, but that has to be done on each > > DLE. it seems like there should be a way to tell just amanda to use > > a larger fraction of the tape, but I can't figure it out. > > As an ideal, amanda tries to get consistant size dumps > each run of amdump. As an approximation this size can be > estimated (using post-compression data) as the sum of > all level 0's divided by the number of runs per dumpcycle > plus the average daily level 1 sizes. > > So you can get larger runs by three means, get more data > (bigger level 0's), become more active (bigger level 1's) > or shorten the dumpcycle (or runs/dumpcycle). > > Sounds like you may want to shorten your dumpcycle. > > Note, dumpcycle is a "dumptype" parameter. I.e. thought > there is a global default value for dumpcycle, new dump- > types can be defined with a different dumpcycle. > > In my case I have my photos and my music backed up in > their own DLEs, 5 DLEs in the case of photos. These are > nearly static DLEs, so no need to back them up each week. > I defined a separate dumptype to have a 3 week dumpcycle > and these 6 DLEs use that dumptype. > > You may want to define a dumptype for your more precious > DLEs with a shorter dumpcycle which will cause them to > be dumped at level 0 more frequently. Or just shorten > your default dumpcycle. > Since a 10 day cycle was no help, I had several weeks back, gone back to a 7 day cycle. We'll see how that works with your PM'd settings. I assume I should watch the df report, currently: /dev/sdc2 960798056 668360948 243624668 74% /amandatapes
And that shows that commenting that stuff out helps, it was 88% a month ago. We'll see which way it tends in the next week or 4. I can, at that level of drive usage, put off swapping the 1TB for a 2TB I took out of a milling machine when I put a 60GB SSD in it. Those things are FAST! No mount, its just hanging on the cable. I need to round up some more of those, and some 3.5" mount adapters. Even 60Gigs is overkill for a milling machine unless using a code generator, as they generate megabytes of code that I can write by hand that does the same thing in 60 lines of gcode. Probably to a higher accuracy. Both are better than the machine, but... > Jon Have a better 2018, Jon, and thanks. 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>
