Am 04-Sep-2021 21:52:11 +0200 schrieb ph...@caerllewys.net: > On 9/4/21 1:50 PM, neumei...@mail.de wrote: > > Hello bacula-community, > > i have a question. I've read that chapter of the bacula-documentation: > > https://www.bacula.org/11.0.x-manuals/en/main/Automated_Disk_Backup.html > > It states:"Now since each of these different kinds of saves needs to > > remain valid for differing periods, the simplest way to do this (and > > possibly the only) is to have a separate Pool for each backup type.". > > The Tutorial uses three pools, a incremental one, a differential one, > > and a full one to accomplish a basic GFS-scheme-backup. I got my > > bacula-Installation running with multiple clients/file-deamons and do my > > backups on a harddrive. > > > > *My questions:* > > I am wondering if there is a possibility to use four pools with a added > > second incremental-pool? > > > > *The problem I see:* > > is that an incremental backup always references the last incremental-, > > differential- or full-backup and saves only the files that have changed > > since then. In my scheme i have two different incremental-backups, the > > daily and the weekly one. The weekly one always references the daily one > > that came the day before(saturday), right? When i keep the weekly one > > around for more than 4weeks(i have chosen 5weeks) and the daily one only > > for more than 10 days(i have chosen 14days) the daily files are going to > > be deleted after 14days but the monthly one still references them. So > > there will be a "hole" in the backupset? > > > > I figured out, if it would be possible to have multiple "levels" of > > incremental-backups, like i could say that the weekly incremental-backup > > references only the weekly backups(level0) and the daily > > incremental-backups only the daily ones(level1) the problem would be > > gone. The same problem would occur when i would keep the full-backup for > > to short around. For example if the full backup overwrites the only > > existing one, there will be no full backup left, that the > > differential-backup could reference and I lose all backups i made to > > that point. > > A Bacula incremental Job references the most recent backup *of any > level* of the same Job. If you had another incremental that used a > different pool, it would have to be a different job, and then it would > reference the last completed backup of *that* job. > > If you are doing a "daily" and a "weekly" incremental of the same job, > then the "weekly" either isn't really "weekly" or isn't really > incremental. Unless what you're trying to do is somehow have one > incremental job receive special treatment and have a longer retention > time than other incremental jobs. It's not quite clear what you mean, > or what you're trying to accomplish. > > What I do is the following: > > I run an incremental backup six nights a week (which is based upon the > last incremental or higher run of that Job, so it records all changes > since the previous night), with one month retention time. > > The seventh night, once per *week*, I run a Differential backup (based > upon the last Differential or higher, so it records all changes since > the last Differential or Full), with two months retention time. > > Once per *month*, I run a Full backup (which rebases everything), which > has six months retention. > > And then finally after the Full backup, I run a Copy job which copies > the Full backup to removable disks in a separate Pool with one year > retention, which are then stored offline (and air-gapped). > > Now what you COULD do, I suppose, is run a Copy job once a week which > copies the most recent Incremental volume(s) to a separate pool with > longer expiration. But those copied incrementals would then reference > incremental jobs that quite possibly wouldn't exist any more by the time > you came to need to refer to them, and you would have succeeded in > setting up exactly the "hole in the backup set" situation you appear to > be trying to avoid. So don't do that. > > What it sounds like to me is that your "weekly incremental" job should > be a Differential job, and your monthly job should be either a Full, or > a Differential followed by a Virtual Full. Your Full backup should > never "overwrite the only existing one" unless you're doing something > wrong or you only have enough backup space to store one Full backup at a > time, in which case you have a deeper problem than trying to fudge > retention times. > > It is *possible* that the situation here is that you are asking the > wrong questions because you don't properly understand the terms. You > may be misunderstanding the purpose and behaviors of the different Job > levels. If in doubt, ask. We're always happy to answer. >
Hello Phil, First of all thank you for your answer. I have thought some time about it. Actually I don't know that there is a deeper purpose of the different job levels and I would really appreciate it if someone please could clear that up for me. But I'm quiet sure that I do understand the behaviors of them. However In case I did get something wrong I will write out the behaviors and would acknowledge if someone could point out if I should got something wrong. Thank you. For a given job: -an incremental backup references the most recent backup of any kind(incremental,differential,full) and backups all the data that changed since then -a differential backup references the most recent differential or full backup and backups all the data that changed since then -a full backup references nothing and backups all the data You have got me right. You wrote: "If you are doing a "daily" and a "weekly" incremental of the same job, then the "weekly" either isn't really "weekly" or isn't really incremental." That's exactly what I was attempting to do. My goal was to shrink the space all my backups need by introducing a second incremental-pool with a longer retention period, but this leads straight to the problem you have described under the "Now what you COULD do, ..."-paragraph that I am trying to avoid. I'm Sorry that I didn't made it clearer. Thank you for pointing out the Virtual Full option. I didn't know about that. I know that my full-backup should never overwrite the only existing one. That was only an example that should clear up what I meant with "hole in the backup set". I think that in that case someone would need more backup space. If I'm wrong please correct me. With what I have learned from you so far I am going to implement the following scheme(one job, one client): Three pools: one incremental(dailyPool), one differential(monthlyPool) and a full pool(halfannualPool). - incremental backup every night with a volume-retention of 40days - differential backup every 1st february-june and august-december (at night, same time) with a volume-retention of 7months - full backup on january 1st and july 1st (at night, same time) with a volume-retention of 12months file- and job-retention for that single client are set to the maximum volume-retention. -> File Retention = 12 months , Job Retention = 12 months Are there any mistakes? What can I do better? There just popped up another question in my head: - should I preferably use the Virtual Full option to make full backups or the normal full backup-option? Are there any downsides? Thank you for your help. Sebastian ------------------------------------------------------------------------------------------------- FreeMail powered by mail.de - MEHR SICHERHEIT, SERIOSITÄT UND KOMFORT _______________________________________________ Bacula-users mailing list Bacula-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
