Depending on what I'm testing and its importance I may just # cp tapelist.yesterday tapelist or manually alter it.
-----Original Message----- From: Debra S Baddorf <[email protected]> Sent: Monday, November 26, 2018 3:25 PM To: Cuttler, Brian R (HEALTH) <[email protected]> Cc: Debra S Baddorf <[email protected]>; Debra S Baddorf <[email protected]>; amanda-users <[email protected]> Subject: Re: Configuration Rollback [Was: Reusable] ATTENTION: This email came from an external source. Do not open attachments or click on links from unknown senders or unexpected emails. > On Nov 26, 2018, at 2:19 PM, Cuttler, Brian R (HEALTH) > <[email protected]> wrote: > > Deb, > > I'm with you, periodically I have to test something, or will start an > additional backup to help move data after a failure of some sort but in > general I allow cron to start # amdump once/day. My understanding matches > yours, balance will get thrown off, amanda may advance dumps but you chew up > a lot of tape and do a lot of I/O for little overall gain if you are running > multiple dumps as a matter of course. When I have something that failed/was missed OR a new node or DLE to add, I do amdump config —no-taper node DLE so test the dump, but leave the results on my holding disk. It’ll be flushed when tonight’s backup starts. So I don’t HAVE to chew up tape, just to test things. :) Deb Baddorf > > I have even gone so far as run run multiple dumps across a very small > (perhaps "1") DLE to test include/exclude or get a new dumptype (snapshot) or > compress(pig/parallel zip) tested, but do not run multiple dumps as a matter > of routine. > > In my mind running many small DLEs can be self-defeating, as can running very > few very large once, each hitting a different set of constraints. > > My samba shares are on separate ZFS mount points and I snapshot them. My home > directories are also on separate ZFS mount points but individual backups were > untenable so I glob them by letter, but that means I can't do snapshots. > > Based on the latest emails I think Chris may have moved on, but he has these > additional answers for when he cycles back. > > Thanks, > Brian > > > -----Original Message----- > From: Debra S Baddorf <[email protected]> > Sent: Monday, November 26, 2018 3:00 PM > To: amanda-users <[email protected]> > Cc: Debra S Baddorf <[email protected]>; Debra S Baddorf > <[email protected]>; Cuttler, Brian R (HEALTH) > <[email protected]> > Subject: Re: Configuration Rollback [Was: Reusable] > > ATTENTION: This email came from an external source. Do not open attachments > or click on links from unknown senders or unexpected emails. > > >> >> -----Original Message----- >> From: [email protected] <[email protected]> >> On Behalf Of Debra S Baddorf >> Sent: Monday, November 26, 2018 2:04 PM >> To: amanda-users <[email protected]> >> Cc: Debra S Baddorf <[email protected]> >> Subject: Re: Configuration Rollback [Was: Reusable] >> >> ATTENTION: This email came from an external source. Do not open attachments >> or click on links from unknown senders or unexpected emails. >> >> >>> On Nov 24, 2018, at 9:47 AM, Chris Nighswonger >>> <[email protected]> wrote: >>> >>> On Fri, Nov 23, 2018 at 6:47 PM Jon LaBadie <[email protected]> wrote: >>> On Wed, Nov 21, 2018 at 11:55:21AM -0800, Chris Miller wrote: >>>> Hi Folks, >>>> >>>> I have written some very small DLEs so I can rip through weeks of >>>> backups in minutes. I've learned some things. >>> >> >> Am I wrong in thinking that you cannot do extra backups, to get to the end >> of your “dumpcycle” faster? >> Dumpcycle is explicitly in days, not number of times run. Isn’t it? >> >> Deb Baddorf >> Fermilab >> >> >> >> On Nov 26, 2018, at 1:21 PM, Cuttler, Brian R (HEALTH) >> <[email protected]> wrote: >> >> Deb, >> >> Not sure if I'm understanding your question. If so I believe Amanda was >> built with the concept of a once/day run schedule taking into account >> runs/cycle as well as days/dumpcycle (for instance 5 run days in a one week >> dump cycle). >> >> Brian > > Yes, I agree. At one point, Chris was trying to speed this up by doing 21 > runs in 10 minutes or so. > Perhaps he has stopped that, and people are just continuing to quote that > line (above). It’s that line that’s bothering me. > > He did ask "Can I specify "dumpcycle" as an elapsed count rather than an > elapsed time?” > Per the wiki help files, dumpcycle seems to be explicitly in “days” and > cannot be changed or sped up. > ( > https://urldefense.proofpoint.com/v2/url?u=https-3A__protect2.fireeye. > com_url-3Fk-3D42d018d223e76922.42d2e1e7-2D4a9f6eb7b4f1e3bb-26u-3Dhttps > -3A__wiki.zmanda.com_index.php_Dumpcycle&d=DwIGaQ&c=gRgGjJ3BkIsb5y6s49 > QqsA&r=HMrKaRiCv4jddln9fLPIOw&m=l_nUa8ln6_AcIwhv7dPsFLOJjoUvcO0FZC_Go- > Kk0_Q&s=jgJt912esNKK5Qjd43o1uniBJyWyfB4otVSX4o4CDL8&e= ) > > Chris also asked "Are "dumpcycle" and "runspercycle" conflicting with each > other?” > Even if you occasionally do extra amdump runs (I do), that doesn’t bother > the “runspercycle” count. It still results in “at least one level 0 within > dumpcycle days”. Runspercycle just lets amanda gauge numbers for balance > adjusting, and maybe promoting a DLE to do an early level 0. > Doing 21 runs in NN minutes will use up all your vtapes, but will not force > more than 1 level 0. > > Maybe I’m being pedantic, and he’s looking at other issues now. If so, > nevermind me! > :) > > Deb Baddorf > > >
